From 8c42ee903bd620e89465780f8c61cf77d2e5148f Mon Sep 17 00:00:00 2001 From: njindal Date: Fri, 17 Mar 2023 15:57:38 +0530 Subject: [PATCH 1/4] [Img2Img]: Copyover img2img pipeline --- .../clip_guided_stable_diffusion_img2img.py | 735 ++++++++++++++++++ 1 file changed, 735 insertions(+) create mode 100644 examples/community/clip_guided_stable_diffusion_img2img.py diff --git a/examples/community/clip_guided_stable_diffusion_img2img.py b/examples/community/clip_guided_stable_diffusion_img2img.py new file mode 100644 index 000000000000..7642c73ace1d --- /dev/null +++ b/examples/community/clip_guided_stable_diffusion_img2img.py @@ -0,0 +1,735 @@ +# Copyright 2023 The HuggingFace Team. All rights reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import inspect +from typing import Callable, List, Optional, Union + +import numpy as np +import PIL +import torch +from packaging import version +from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer + +from diffusers import DiffusionPipeline +from diffusers.configuration_utils import FrozenDict +from diffusers.models import AutoencoderKL, UNet2DConditionModel +from diffusers.schedulers import KarrasDiffusionSchedulers + +from diffusers.utils import ( + PIL_INTERPOLATION, + deprecate, + is_accelerate_available, + is_accelerate_version, + logging, + randn_tensor, + replace_example_docstring, +) +from diffusers.pipelines.stable_diffusion import StableDiffusionPipelineOutput +from diffusers.pipelines.stable_diffusion.safety_checker import StableDiffusionSafetyChecker + +logger = logging.get_logger(__name__) # pylint: disable=invalid-name + +# TODO: Change this +EXAMPLE_DOC_STRING = """ + Examples: + ```py + >>> import requests + >>> import torch + >>> from PIL import Image + >>> from io import BytesIO + + >>> from diffusers import StableDiffusionImg2ImgPipeline + + >>> device = "cuda" + >>> model_id_or_path = "runwayml/stable-diffusion-v1-5" + >>> pipe = StableDiffusionImg2ImgPipeline.from_pretrained(model_id_or_path, torch_dtype=torch.float16) + >>> pipe = pipe.to(device) + + >>> url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg" + + >>> response = requests.get(url) + >>> init_image = Image.open(BytesIO(response.content)).convert("RGB") + >>> init_image = init_image.resize((768, 512)) + + >>> prompt = "A fantasy landscape, trending on artstation" + + >>> images = pipe(prompt=prompt, image=init_image, strength=0.75, guidance_scale=7.5).images + >>> images[0].save("fantasy_landscape.png") + ``` +""" + + +def preprocess(image): + if isinstance(image, torch.Tensor): + return image + elif isinstance(image, PIL.Image.Image): + image = [image] + + if isinstance(image[0], PIL.Image.Image): + w, h = image[0].size + w, h = map(lambda x: x - x % 8, (w, h)) # resize to integer multiple of 8 + + image = [np.array(i.resize((w, h), resample=PIL_INTERPOLATION["lanczos"]))[None, :] for i in image] + image = np.concatenate(image, axis=0) + image = np.array(image).astype(np.float32) / 255.0 + image = image.transpose(0, 3, 1, 2) + image = 2.0 * image - 1.0 + image = torch.from_numpy(image) + elif isinstance(image[0], torch.Tensor): + image = torch.cat(image, dim=0) + return image + + +class CLIPGuidedStableDiffusionImg2ImgPipeline(DiffusionPipeline): + r""" + Pipeline for text-guided image to image generation using Stable Diffusion. + + This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods the + library implements for all the pipelines (such as downloading or saving, running on a particular device, etc.) + + Args: + vae ([`AutoencoderKL`]): + Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations. + text_encoder ([`CLIPTextModel`]): + Frozen text-encoder. Stable Diffusion uses the text portion of + [CLIP](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel), specifically + the [clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14) variant. + tokenizer (`CLIPTokenizer`): + Tokenizer of class + [CLIPTokenizer](https://huggingface.co/docs/transformers/v4.21.0/en/model_doc/clip#transformers.CLIPTokenizer). + unet ([`UNet2DConditionModel`]): Conditional U-Net architecture to denoise the encoded image latents. + scheduler ([`SchedulerMixin`]): + A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of + [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`]. + safety_checker ([`StableDiffusionSafetyChecker`]): + Classification module that estimates whether generated images could be considered offensive or harmful. + Please, refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for details. + feature_extractor ([`CLIPFeatureExtractor`]): + Model that extracts features from generated images to be used as inputs for the `safety_checker`. + """ + _optional_components = ["safety_checker", "feature_extractor"] + + # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.__init__ + def __init__( + self, + vae: AutoencoderKL, + text_encoder: CLIPTextModel, + tokenizer: CLIPTokenizer, + unet: UNet2DConditionModel, + scheduler: KarrasDiffusionSchedulers, + safety_checker: StableDiffusionSafetyChecker, + feature_extractor: CLIPFeatureExtractor, + requires_safety_checker: bool = True, + ): + super().__init__() + + if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 1: + deprecation_message = ( + f"The configuration file of this scheduler: {scheduler} is outdated. `steps_offset`" + f" should be set to 1 instead of {scheduler.config.steps_offset}. Please make sure " + "to update the config accordingly as leaving `steps_offset` might led to incorrect results" + " in future versions. If you have downloaded this checkpoint from the Hugging Face Hub," + " it would be very nice if you could open a Pull request for the `scheduler/scheduler_config.json`" + " file" + ) + deprecate("steps_offset!=1", "1.0.0", deprecation_message, standard_warn=False) + new_config = dict(scheduler.config) + new_config["steps_offset"] = 1 + scheduler._internal_dict = FrozenDict(new_config) + + if hasattr(scheduler.config, "clip_sample") and scheduler.config.clip_sample is True: + deprecation_message = ( + f"The configuration file of this scheduler: {scheduler} has not set the configuration `clip_sample`." + " `clip_sample` should be set to False in the configuration file. Please make sure to update the" + " config accordingly as not setting `clip_sample` in the config might lead to incorrect results in" + " future versions. If you have downloaded this checkpoint from the Hugging Face Hub, it would be very" + " nice if you could open a Pull request for the `scheduler/scheduler_config.json` file" + ) + deprecate("clip_sample not set", "1.0.0", deprecation_message, standard_warn=False) + new_config = dict(scheduler.config) + new_config["clip_sample"] = False + scheduler._internal_dict = FrozenDict(new_config) + + if safety_checker is None and requires_safety_checker: + logger.warning( + f"You have disabled the safety checker for {self.__class__} by passing `safety_checker=None`. Ensure" + " that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered" + " results in services or applications open to the public. Both the diffusers team and Hugging Face" + " strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling" + " it only for use-cases that involve analyzing network behavior or auditing its results. For more" + " information, please have a look at https://github.com/huggingface/diffusers/pull/254 ." + ) + + if safety_checker is not None and feature_extractor is None: + raise ValueError( + "Make sure to define a feature extractor when loading {self.__class__} if you want to use the safety" + " checker. If you do not want to use the safety checker, you can pass `'safety_checker=None'` instead." + ) + + is_unet_version_less_0_9_0 = hasattr(unet.config, "_diffusers_version") and version.parse( + version.parse(unet.config._diffusers_version).base_version + ) < version.parse("0.9.0.dev0") + is_unet_sample_size_less_64 = hasattr(unet.config, "sample_size") and unet.config.sample_size < 64 + if is_unet_version_less_0_9_0 and is_unet_sample_size_less_64: + deprecation_message = ( + "The configuration file of the unet has set the default `sample_size` to smaller than" + " 64 which seems highly unlikely. If your checkpoint is a fine-tuned version of any of the" + " following: \n- CompVis/stable-diffusion-v1-4 \n- CompVis/stable-diffusion-v1-3 \n-" + " CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- runwayml/stable-diffusion-v1-5" + " \n- runwayml/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the" + " configuration file. Please make sure to update the config accordingly as leaving `sample_size=32`" + " in the config might lead to incorrect results in future versions. If you have downloaded this" + " checkpoint from the Hugging Face Hub, it would be very nice if you could open a Pull request for" + " the `unet/config.json` file" + ) + deprecate("sample_size<64", "1.0.0", deprecation_message, standard_warn=False) + new_config = dict(unet.config) + new_config["sample_size"] = 64 + unet._internal_dict = FrozenDict(new_config) + + self.register_modules( + vae=vae, + text_encoder=text_encoder, + tokenizer=tokenizer, + unet=unet, + scheduler=scheduler, + safety_checker=safety_checker, + feature_extractor=feature_extractor, + ) + self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) + self.register_to_config(requires_safety_checker=requires_safety_checker) + + # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_sequential_cpu_offload + def enable_sequential_cpu_offload(self, gpu_id=0): + r""" + Offloads all models to CPU using accelerate, significantly reducing memory usage. When called, unet, + text_encoder, vae and safety checker have their state dicts saved to CPU and then are moved to a + `torch.device('meta') and loaded to GPU only when their specific submodule has its `forward` method called. + Note that offloading happens on a submodule basis. Memory savings are higher than with + `enable_model_cpu_offload`, but performance is lower. + """ + if is_accelerate_available() and is_accelerate_version(">=", "0.14.0"): + from accelerate import cpu_offload + else: + raise ImportError("`enable_sequential_cpu_offload` requires `accelerate v0.14.0` or higher") + + device = torch.device(f"cuda:{gpu_id}") + + if self.device.type != "cpu": + self.to("cpu", silence_dtype_warnings=True) + torch.cuda.empty_cache() # otherwise we don't see the memory savings (but they probably exist) + + for cpu_offloaded_model in [self.unet, self.text_encoder, self.vae]: + cpu_offload(cpu_offloaded_model, device) + + if self.safety_checker is not None: + cpu_offload(self.safety_checker, execution_device=device, offload_buffers=True) + + # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_model_cpu_offload + def enable_model_cpu_offload(self, gpu_id=0): + r""" + Offloads all models to CPU using accelerate, reducing memory usage with a low impact on performance. Compared + to `enable_sequential_cpu_offload`, this method moves one whole model at a time to the GPU when its `forward` + method is called, and the model remains in GPU until the next model runs. Memory savings are lower than with + `enable_sequential_cpu_offload`, but performance is much better due to the iterative execution of the `unet`. + """ + if is_accelerate_available() and is_accelerate_version(">=", "0.17.0.dev0"): + from accelerate import cpu_offload_with_hook + else: + raise ImportError("`enable_model_offload` requires `accelerate v0.17.0` or higher.") + + device = torch.device(f"cuda:{gpu_id}") + + if self.device.type != "cpu": + self.to("cpu", silence_dtype_warnings=True) + torch.cuda.empty_cache() # otherwise we don't see the memory savings (but they probably exist) + + hook = None + for cpu_offloaded_model in [self.text_encoder, self.unet, self.vae]: + _, hook = cpu_offload_with_hook(cpu_offloaded_model, device, prev_module_hook=hook) + + if self.safety_checker is not None: + _, hook = cpu_offload_with_hook(self.safety_checker, device, prev_module_hook=hook) + + # We'll offload the last model manually. + self.final_offload_hook = hook + + @property + # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline._execution_device + def _execution_device(self): + r""" + Returns the device on which the pipeline's models will be executed. After calling + `pipeline.enable_sequential_cpu_offload()` the execution device can only be inferred from Accelerate's module + hooks. + """ + if not hasattr(self.unet, "_hf_hook"): + return self.device + for module in self.unet.modules(): + if ( + hasattr(module, "_hf_hook") + and hasattr(module._hf_hook, "execution_device") + and module._hf_hook.execution_device is not None + ): + return torch.device(module._hf_hook.execution_device) + return self.device + + # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline._encode_prompt + def _encode_prompt( + self, + prompt, + device, + num_images_per_prompt, + do_classifier_free_guidance, + negative_prompt=None, + prompt_embeds: Optional[torch.FloatTensor] = None, + negative_prompt_embeds: Optional[torch.FloatTensor] = None, + ): + r""" + Encodes the prompt into text encoder hidden states. + + Args: + prompt (`str` or `List[str]`, *optional*): + prompt to be encoded + device: (`torch.device`): + torch device + num_images_per_prompt (`int`): + number of images that should be generated per prompt + do_classifier_free_guidance (`bool`): + whether to use classifier free guidance or not + negative_prompt (`str` or `List[str]`, *optional*): + The prompt or prompts not to guide the image generation. If not defined, one has to pass + `negative_prompt_embeds`. instead. If not defined, one has to pass `negative_prompt_embeds`. instead. + Ignored when not using guidance (i.e., ignored if `guidance_scale` is less than `1`). + prompt_embeds (`torch.FloatTensor`, *optional*): + Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not + provided, text embeddings will be generated from `prompt` input argument. + negative_prompt_embeds (`torch.FloatTensor`, *optional*): + Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt + weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input + argument. + """ + if prompt is not None and isinstance(prompt, str): + batch_size = 1 + elif prompt is not None and isinstance(prompt, list): + batch_size = len(prompt) + else: + batch_size = prompt_embeds.shape[0] + + if prompt_embeds is None: + text_inputs = self.tokenizer( + prompt, + padding="max_length", + max_length=self.tokenizer.model_max_length, + truncation=True, + return_tensors="pt", + ) + text_input_ids = text_inputs.input_ids + untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="pt").input_ids + + if untruncated_ids.shape[-1] >= text_input_ids.shape[-1] and not torch.equal( + text_input_ids, untruncated_ids + ): + removed_text = self.tokenizer.batch_decode( + untruncated_ids[:, self.tokenizer.model_max_length - 1 : -1] + ) + logger.warning( + "The following part of your input was truncated because CLIP can only handle sequences up to" + f" {self.tokenizer.model_max_length} tokens: {removed_text}" + ) + + if hasattr(self.text_encoder.config, "use_attention_mask") and self.text_encoder.config.use_attention_mask: + attention_mask = text_inputs.attention_mask.to(device) + else: + attention_mask = None + + prompt_embeds = self.text_encoder( + text_input_ids.to(device), + attention_mask=attention_mask, + ) + prompt_embeds = prompt_embeds[0] + + prompt_embeds = prompt_embeds.to(dtype=self.text_encoder.dtype, device=device) + + bs_embed, seq_len, _ = prompt_embeds.shape + # duplicate text embeddings for each generation per prompt, using mps friendly method + prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1) + prompt_embeds = prompt_embeds.view(bs_embed * num_images_per_prompt, seq_len, -1) + + # get unconditional embeddings for classifier free guidance + if do_classifier_free_guidance and negative_prompt_embeds is None: + uncond_tokens: List[str] + if negative_prompt is None: + uncond_tokens = [""] * batch_size + elif type(prompt) is not type(negative_prompt): + raise TypeError( + f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !=" + f" {type(prompt)}." + ) + elif isinstance(negative_prompt, str): + uncond_tokens = [negative_prompt] + elif batch_size != len(negative_prompt): + raise ValueError( + f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:" + f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches" + " the batch size of `prompt`." + ) + else: + uncond_tokens = negative_prompt + + max_length = prompt_embeds.shape[1] + uncond_input = self.tokenizer( + uncond_tokens, + padding="max_length", + max_length=max_length, + truncation=True, + return_tensors="pt", + ) + + if hasattr(self.text_encoder.config, "use_attention_mask") and self.text_encoder.config.use_attention_mask: + attention_mask = uncond_input.attention_mask.to(device) + else: + attention_mask = None + + negative_prompt_embeds = self.text_encoder( + uncond_input.input_ids.to(device), + attention_mask=attention_mask, + ) + negative_prompt_embeds = negative_prompt_embeds[0] + + if do_classifier_free_guidance: + # duplicate unconditional embeddings for each generation per prompt, using mps friendly method + seq_len = negative_prompt_embeds.shape[1] + + negative_prompt_embeds = negative_prompt_embeds.to(dtype=self.text_encoder.dtype, device=device) + + negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1) + negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1) + + # For classifier free guidance, we need to do two forward passes. + # Here we concatenate the unconditional and text embeddings into a single batch + # to avoid doing two forward passes + prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds]) + + return prompt_embeds + + # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.run_safety_checker + def run_safety_checker(self, image, device, dtype): + if self.safety_checker is not None: + safety_checker_input = self.feature_extractor(self.numpy_to_pil(image), return_tensors="pt").to(device) + image, has_nsfw_concept = self.safety_checker( + images=image, clip_input=safety_checker_input.pixel_values.to(dtype) + ) + else: + has_nsfw_concept = None + return image, has_nsfw_concept + + # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.decode_latents + def decode_latents(self, latents): + latents = 1 / self.vae.config.scaling_factor * latents + image = self.vae.decode(latents).sample + image = (image / 2 + 0.5).clamp(0, 1) + # we always cast to float32 as this does not cause significant overhead and is compatible with bfloat16 + image = image.cpu().permute(0, 2, 3, 1).float().numpy() + return image + + # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_extra_step_kwargs + def prepare_extra_step_kwargs(self, generator, eta): + # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature + # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers. + # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502 + # and should be between [0, 1] + + accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys()) + extra_step_kwargs = {} + if accepts_eta: + extra_step_kwargs["eta"] = eta + + # check if the scheduler accepts generator + accepts_generator = "generator" in set(inspect.signature(self.scheduler.step).parameters.keys()) + if accepts_generator: + extra_step_kwargs["generator"] = generator + return extra_step_kwargs + + def check_inputs( + self, prompt, strength, callback_steps, negative_prompt=None, prompt_embeds=None, negative_prompt_embeds=None + ): + if strength < 0 or strength > 1: + raise ValueError(f"The value of strength should in [0.0, 1.0] but is {strength}") + + if (callback_steps is None) or ( + callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0) + ): + raise ValueError( + f"`callback_steps` has to be a positive integer but is {callback_steps} of type" + f" {type(callback_steps)}." + ) + + if prompt is not None and prompt_embeds is not None: + raise ValueError( + f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to" + " only forward one of the two." + ) + elif prompt is None and prompt_embeds is None: + raise ValueError( + "Provide either `prompt` or `prompt_embeds`. Cannot leave both `prompt` and `prompt_embeds` undefined." + ) + elif prompt is not None and (not isinstance(prompt, str) and not isinstance(prompt, list)): + raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}") + + if negative_prompt is not None and negative_prompt_embeds is not None: + raise ValueError( + f"Cannot forward both `negative_prompt`: {negative_prompt} and `negative_prompt_embeds`:" + f" {negative_prompt_embeds}. Please make sure to only forward one of the two." + ) + + if prompt_embeds is not None and negative_prompt_embeds is not None: + if prompt_embeds.shape != negative_prompt_embeds.shape: + raise ValueError( + "`prompt_embeds` and `negative_prompt_embeds` must have the same shape when passed directly, but" + f" got: `prompt_embeds` {prompt_embeds.shape} != `negative_prompt_embeds`" + f" {negative_prompt_embeds.shape}." + ) + + def get_timesteps(self, num_inference_steps, strength, device): + # get the original timestep using init_timestep + init_timestep = min(int(num_inference_steps * strength), num_inference_steps) + + t_start = max(num_inference_steps - init_timestep, 0) + timesteps = self.scheduler.timesteps[t_start:] + + return timesteps, num_inference_steps - t_start + + def prepare_latents(self, image, timestep, batch_size, num_images_per_prompt, dtype, device, generator=None): + if not isinstance(image, (torch.Tensor, PIL.Image.Image, list)): + raise ValueError( + f"`image` has to be of type `torch.Tensor`, `PIL.Image.Image` or list but is {type(image)}" + ) + + image = image.to(device=device, dtype=dtype) + + batch_size = batch_size * num_images_per_prompt + if isinstance(generator, list) and len(generator) != batch_size: + raise ValueError( + f"You have passed a list of generators of length {len(generator)}, but requested an effective batch" + f" size of {batch_size}. Make sure the batch size matches the length of the generators." + ) + + if isinstance(generator, list): + init_latents = [ + self.vae.encode(image[i : i + 1]).latent_dist.sample(generator[i]) for i in range(batch_size) + ] + init_latents = torch.cat(init_latents, dim=0) + else: + init_latents = self.vae.encode(image).latent_dist.sample(generator) + + init_latents = self.vae.config.scaling_factor * init_latents + + if batch_size > init_latents.shape[0] and batch_size % init_latents.shape[0] == 0: + # expand init_latents for batch_size + deprecation_message = ( + f"You have passed {batch_size} text prompts (`prompt`), but only {init_latents.shape[0]} initial" + " images (`image`). Initial images are now duplicating to match the number of text prompts. Note" + " that this behavior is deprecated and will be removed in a version 1.0.0. Please make sure to update" + " your script to pass as many initial images as text prompts to suppress this warning." + ) + deprecate("len(prompt) != len(image)", "1.0.0", deprecation_message, standard_warn=False) + additional_image_per_prompt = batch_size // init_latents.shape[0] + init_latents = torch.cat([init_latents] * additional_image_per_prompt, dim=0) + elif batch_size > init_latents.shape[0] and batch_size % init_latents.shape[0] != 0: + raise ValueError( + f"Cannot duplicate `image` of batch size {init_latents.shape[0]} to {batch_size} text prompts." + ) + else: + init_latents = torch.cat([init_latents], dim=0) + + shape = init_latents.shape + noise = randn_tensor(shape, generator=generator, device=device, dtype=dtype) + + # get latents + init_latents = self.scheduler.add_noise(init_latents, noise, timestep) + latents = init_latents + + return latents + + @torch.no_grad() + @replace_example_docstring(EXAMPLE_DOC_STRING) + def __call__( + self, + prompt: Union[str, List[str]] = None, + image: Union[torch.FloatTensor, PIL.Image.Image] = None, + strength: float = 0.8, + num_inference_steps: Optional[int] = 50, + guidance_scale: Optional[float] = 7.5, + negative_prompt: Optional[Union[str, List[str]]] = None, + num_images_per_prompt: Optional[int] = 1, + eta: Optional[float] = 0.0, + generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None, + prompt_embeds: Optional[torch.FloatTensor] = None, + negative_prompt_embeds: Optional[torch.FloatTensor] = None, + output_type: Optional[str] = "pil", + return_dict: bool = True, + callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None, + callback_steps: int = 1, + ): + r""" + Function invoked when calling the pipeline for generation. + + Args: + prompt (`str` or `List[str]`, *optional*): + The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`. + instead. + image (`torch.FloatTensor` or `PIL.Image.Image`): + `Image`, or tensor representing an image batch, that will be used as the starting point for the + process. + strength (`float`, *optional*, defaults to 0.8): + Conceptually, indicates how much to transform the reference `image`. Must be between 0 and 1. `image` + will be used as a starting point, adding more noise to it the larger the `strength`. The number of + denoising steps depends on the amount of noise initially added. When `strength` is 1, added noise will + be maximum and the denoising process will run for the full number of iterations specified in + `num_inference_steps`. A value of 1, therefore, essentially ignores `image`. + num_inference_steps (`int`, *optional*, defaults to 50): + The number of denoising steps. More denoising steps usually lead to a higher quality image at the + expense of slower inference. This parameter will be modulated by `strength`. + guidance_scale (`float`, *optional*, defaults to 7.5): + Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598). + `guidance_scale` is defined as `w` of equation 2. of [Imagen + Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale > + 1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`, + usually at the expense of lower image quality. + negative_prompt (`str` or `List[str]`, *optional*): + The prompt or prompts not to guide the image generation. If not defined, one has to pass + `negative_prompt_embeds`. instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` + is less than `1`). + num_images_per_prompt (`int`, *optional*, defaults to 1): + The number of images to generate per prompt. + eta (`float`, *optional*, defaults to 0.0): + Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to + [`schedulers.DDIMScheduler`], will be ignored for others. + generator (`torch.Generator`, *optional*): + One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html) + to make generation deterministic. + prompt_embeds (`torch.FloatTensor`, *optional*): + Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not + provided, text embeddings will be generated from `prompt` input argument. + negative_prompt_embeds (`torch.FloatTensor`, *optional*): + Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt + weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input + argument. + output_type (`str`, *optional*, defaults to `"pil"`): + The output format of the generate image. Choose between + [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`. + return_dict (`bool`, *optional*, defaults to `True`): + Whether or not to return a [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] instead of a + plain tuple. + callback (`Callable`, *optional*): + A function that will be called every `callback_steps` steps during inference. The function will be + called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`. + callback_steps (`int`, *optional*, defaults to 1): + The frequency at which the `callback` function will be called. If not specified, the callback will be + called at every step. + Examples: + + Returns: + [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] or `tuple`: + [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] if `return_dict` is True, otherwise a `tuple. + When returning a tuple, the first element is a list with the generated images, and the second element is a + list of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" + (nsfw) content, according to the `safety_checker`. + """ + # 1. Check inputs. Raise error if not correct + self.check_inputs(prompt, strength, callback_steps, negative_prompt, prompt_embeds, negative_prompt_embeds) + + # 2. Define call parameters + if prompt is not None and isinstance(prompt, str): + batch_size = 1 + elif prompt is not None and isinstance(prompt, list): + batch_size = len(prompt) + else: + batch_size = prompt_embeds.shape[0] + device = self._execution_device + # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2) + # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1` + # corresponds to doing no classifier free guidance. + do_classifier_free_guidance = guidance_scale > 1.0 + + # 3. Encode input prompt + prompt_embeds = self._encode_prompt( + prompt, + device, + num_images_per_prompt, + do_classifier_free_guidance, + negative_prompt, + prompt_embeds=prompt_embeds, + negative_prompt_embeds=negative_prompt_embeds, + ) + + # 4. Preprocess image + image = preprocess(image) + + # 5. set timesteps + self.scheduler.set_timesteps(num_inference_steps, device=device) + timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength, device) + latent_timestep = timesteps[:1].repeat(batch_size * num_images_per_prompt) + + # 6. Prepare latent variables + latents = self.prepare_latents( + image, latent_timestep, batch_size, num_images_per_prompt, prompt_embeds.dtype, device, generator + ) + + # 7. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline + extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta) + + # 8. Denoising loop + num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order + with self.progress_bar(total=num_inference_steps) as progress_bar: + for i, t in enumerate(timesteps): + # expand the latents if we are doing classifier free guidance + latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents + latent_model_input = self.scheduler.scale_model_input(latent_model_input, t) + + # predict the noise residual + noise_pred = self.unet(latent_model_input, t, encoder_hidden_states=prompt_embeds).sample + + # perform guidance + if do_classifier_free_guidance: + noise_pred_uncond, noise_pred_text = noise_pred.chunk(2) + noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond) + + # compute the previous noisy sample x_t -> x_t-1 + latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs).prev_sample + + # call the callback, if provided + if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0): + progress_bar.update() + if callback is not None and i % callback_steps == 0: + callback(i, t, latents) + + # 9. Post-processing + image = self.decode_latents(latents) + + # 10. Run safety checker + image, has_nsfw_concept = self.run_safety_checker(image, device, prompt_embeds.dtype) + + # 11. Convert to PIL + if output_type == "pil": + image = self.numpy_to_pil(image) + + # Offload last model to CPU + if hasattr(self, "final_offload_hook") and self.final_offload_hook is not None: + self.final_offload_hook.offload() + + if not return_dict: + return (image, has_nsfw_concept) + + return StableDiffusionPipelineOutput(images=image, nsfw_content_detected=has_nsfw_concept) From 246449b63b78d038cc3a80bfe927bd1b72013298 Mon Sep 17 00:00:00 2001 From: njindal Date: Fri, 17 Mar 2023 20:04:44 +0530 Subject: [PATCH 2/4] [Img2Img]: img2img pipeline --- examples/community/README.md | 110 ++- .../clip_guided_stable_diffusion_img2img.py | 919 +++++++----------- 2 files changed, 428 insertions(+), 601 deletions(-) diff --git a/examples/community/README.md b/examples/community/README.md index 1521f16bffaf..1600b835b1f1 100644 --- a/examples/community/README.md +++ b/examples/community/README.md @@ -6,29 +6,30 @@ Please have a look at the following table to get an overview of all community examples. Click on the **Code Example** to get a copy-and-paste ready code example that you can try out. If a community doesn't work as expected, please open an issue and ping the author on it. -| Example | Description | Code Example | Colab | Author | -|:---------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-----------------------------------------------------------:| -| CLIP Guided Stable Diffusion | Doing CLIP guidance for text to image generation with Stable Diffusion | [CLIP Guided Stable Diffusion](#clip-guided-stable-diffusion) | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/CLIP_Guided_Stable_diffusion_with_diffusers.ipynb) | [Suraj Patil](https://github.com/patil-suraj/) | -| One Step U-Net (Dummy) | Example showcasing of how to use Community Pipelines (see https://github.com/huggingface/diffusers/issues/841) | [One Step U-Net](#one-step-unet) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) | -| Stable Diffusion Interpolation | Interpolate the latent space of Stable Diffusion between different prompts/seeds | [Stable Diffusion Interpolation](#stable-diffusion-interpolation) | - | [Nate Raw](https://github.com/nateraw/) | -| Stable Diffusion Mega | **One** Stable Diffusion Pipeline with all functionalities of [Text2Image](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py), [Image2Image](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py) and [Inpainting](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py) | [Stable Diffusion Mega](#stable-diffusion-mega) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) | -| Long Prompt Weighting Stable Diffusion | **One** Stable Diffusion Pipeline without tokens length limit, and support parsing weighting in prompt. | [Long Prompt Weighting Stable Diffusion](#long-prompt-weighting-stable-diffusion) | - | [SkyTNT](https://github.com/SkyTNT) | -| Speech to Image | Using automatic-speech-recognition to transcribe text and Stable Diffusion to generate images | [Speech to Image](#speech-to-image) | - | [Mikail Duzenli](https://github.com/MikailINTech) -| Wild Card Stable Diffusion | Stable Diffusion Pipeline that supports prompts that contain wildcard terms (indicated by surrounding double underscores), with values instantiated randomly from a corresponding txt file or a dictionary of possible values | [Wildcard Stable Diffusion](#wildcard-stable-diffusion) | - | [Shyam Sudhakaran](https://github.com/shyamsn97) | -| [Composable Stable Diffusion](https://energy-based-model.github.io/Compositional-Visual-Generation-with-Composable-Diffusion-Models/) | Stable Diffusion Pipeline that supports prompts that contain "|" in prompts (as an AND condition) and weights (separated by "|" as well) to positively / negatively weight prompts. | [Composable Stable Diffusion](#composable-stable-diffusion) | - | [Mark Rich](https://github.com/MarkRich) | -| Seed Resizing Stable Diffusion| Stable Diffusion Pipeline that supports resizing an image and retaining the concepts of the 512 by 512 generation. | [Seed Resizing](#seed-resizing) | - | [Mark Rich](https://github.com/MarkRich) | -| Imagic Stable Diffusion | Stable Diffusion Pipeline that enables writing a text prompt to edit an existing image| [Imagic Stable Diffusion](#imagic-stable-diffusion) | - | [Mark Rich](https://github.com/MarkRich) | -| Multilingual Stable Diffusion| Stable Diffusion Pipeline that supports prompts in 50 different languages. | [Multilingual Stable Diffusion](#multilingual-stable-diffusion-pipeline) | - | [Juan Carlos Piñeros](https://github.com/juancopi81) | -| Image to Image Inpainting Stable Diffusion | Stable Diffusion Pipeline that enables the overlaying of two images and subsequent inpainting| [Image to Image Inpainting Stable Diffusion](#image-to-image-inpainting-stable-diffusion) | - | [Alex McKinney](https://github.com/vvvm23) | -| Text Based Inpainting Stable Diffusion | Stable Diffusion Inpainting Pipeline that enables passing a text prompt to generate the mask for inpainting| [Text Based Inpainting Stable Diffusion](#image-to-image-inpainting-stable-diffusion) | - | [Dhruv Karan](https://github.com/unography) | -| Bit Diffusion | Diffusion on discrete data | [Bit Diffusion](#bit-diffusion) | - |[Stuti R.](https://github.com/kingstut) | -| K-Diffusion Stable Diffusion | Run Stable Diffusion with any of [K-Diffusion's samplers](https://github.com/crowsonkb/k-diffusion/blob/master/k_diffusion/sampling.py) | [Stable Diffusion with K Diffusion](#stable-diffusion-with-k-diffusion) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) | -| Checkpoint Merger Pipeline | Diffusion Pipeline that enables merging of saved model checkpoints | [Checkpoint Merger Pipeline](#checkpoint-merger-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | -Stable Diffusion v1.1-1.4 Comparison | Run all 4 model checkpoints for Stable Diffusion and compare their results together | [Stable Diffusion Comparison](#stable-diffusion-comparisons) | - | [Suvaditya Mukherjee](https://github.com/suvadityamuk) | -MagicMix | Diffusion Pipeline for semantic mixing of an image and a text prompt | [MagicMix](#magic-mix) | - | [Partho Das](https://github.com/daspartho) | -| Stable UnCLIP | Diffusion Pipeline for combining prior model (generate clip image embedding from text, UnCLIPPipeline `"kakaobrain/karlo-v1-alpha"`) and decoder pipeline (decode clip image embedding to image, StableDiffusionImageVariationPipeline `"lambdalabs/sd-image-variations-diffusers"` ). | [Stable UnCLIP](#stable-unclip) | - |[Ray Wang](https://wrong.wang) | -| UnCLIP Text Interpolation Pipeline | Diffusion Pipeline that allows passing two prompts and produces images while interpolating between the text-embeddings of the two prompts | [UnCLIP Text Interpolation Pipeline](#unclip-text-interpolation-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | -| UnCLIP Image Interpolation Pipeline | Diffusion Pipeline that allows passing two images/image_embeddings and produces images while interpolating between their image-embeddings | [UnCLIP Image Interpolation Pipeline](#unclip-image-interpolation-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | +| Example | Description | Code Example | Colab | Author | +|:--------------------------------------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|--------------------------------------------------------------:| +| CLIP Guided Stable Diffusion | Doing CLIP guidance for text to image generation with Stable Diffusion | [CLIP Guided Stable Diffusion](#clip-guided-stable-diffusion) | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/CLIP_Guided_Stable_diffusion_with_diffusers.ipynb) | [Suraj Patil](https://github.com/patil-suraj/) | +| One Step U-Net (Dummy) | Example showcasing of how to use Community Pipelines (see https://github.com/huggingface/diffusers/issues/841) | [One Step U-Net](#one-step-unet) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) | +| Stable Diffusion Interpolation | Interpolate the latent space of Stable Diffusion between different prompts/seeds | [Stable Diffusion Interpolation](#stable-diffusion-interpolation) | - | [Nate Raw](https://github.com/nateraw/) | +| Stable Diffusion Mega | **One** Stable Diffusion Pipeline with all functionalities of [Text2Image](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py), [Image2Image](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py) and [Inpainting](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py) | [Stable Diffusion Mega](#stable-diffusion-mega) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) | +| Long Prompt Weighting Stable Diffusion | **One** Stable Diffusion Pipeline without tokens length limit, and support parsing weighting in prompt. | [Long Prompt Weighting Stable Diffusion](#long-prompt-weighting-stable-diffusion) | - | [SkyTNT](https://github.com/SkyTNT) | +| Speech to Image | Using automatic-speech-recognition to transcribe text and Stable Diffusion to generate images | [Speech to Image](#speech-to-image) | - | [Mikail Duzenli](https://github.com/MikailINTech) +| Wild Card Stable Diffusion | Stable Diffusion Pipeline that supports prompts that contain wildcard terms (indicated by surrounding double underscores), with values instantiated randomly from a corresponding txt file or a dictionary of possible values | [Wildcard Stable Diffusion](#wildcard-stable-diffusion) | - | [Shyam Sudhakaran](https://github.com/shyamsn97) | +| [Composable Stable Diffusion](https://energy-based-model.github.io/Compositional-Visual-Generation-with-Composable-Diffusion-Models/) | Stable Diffusion Pipeline that supports prompts that contain "|" in prompts (as an AND condition) and weights (separated by "|" as well) to positively / negatively weight prompts. | [Composable Stable Diffusion](#composable-stable-diffusion) | - | [Mark Rich](https://github.com/MarkRich) | +| Seed Resizing Stable Diffusion | Stable Diffusion Pipeline that supports resizing an image and retaining the concepts of the 512 by 512 generation. | [Seed Resizing](#seed-resizing) | - | [Mark Rich](https://github.com/MarkRich) | +| Imagic Stable Diffusion | Stable Diffusion Pipeline that enables writing a text prompt to edit an existing image | [Imagic Stable Diffusion](#imagic-stable-diffusion) | - | [Mark Rich](https://github.com/MarkRich) | +| Multilingual Stable Diffusion | Stable Diffusion Pipeline that supports prompts in 50 different languages. | [Multilingual Stable Diffusion](#multilingual-stable-diffusion-pipeline) | - | [Juan Carlos Piñeros](https://github.com/juancopi81) | +| Image to Image Inpainting Stable Diffusion | Stable Diffusion Pipeline that enables the overlaying of two images and subsequent inpainting | [Image to Image Inpainting Stable Diffusion](#image-to-image-inpainting-stable-diffusion) | - | [Alex McKinney](https://github.com/vvvm23) | +| Text Based Inpainting Stable Diffusion | Stable Diffusion Inpainting Pipeline that enables passing a text prompt to generate the mask for inpainting | [Text Based Inpainting Stable Diffusion](#image-to-image-inpainting-stable-diffusion) | - | [Dhruv Karan](https://github.com/unography) | +| Bit Diffusion | Diffusion on discrete data | [Bit Diffusion](#bit-diffusion) | - | [Stuti R.](https://github.com/kingstut) | +| K-Diffusion Stable Diffusion | Run Stable Diffusion with any of [K-Diffusion's samplers](https://github.com/crowsonkb/k-diffusion/blob/master/k_diffusion/sampling.py) | [Stable Diffusion with K Diffusion](#stable-diffusion-with-k-diffusion) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) | +| Checkpoint Merger Pipeline | Diffusion Pipeline that enables merging of saved model checkpoints | [Checkpoint Merger Pipeline](#checkpoint-merger-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | + Stable Diffusion v1.1-1.4 Comparison | Run all 4 model checkpoints for Stable Diffusion and compare their results together | [Stable Diffusion Comparison](#stable-diffusion-comparisons) | - | [Suvaditya Mukherjee](https://github.com/suvadityamuk) | + MagicMix | Diffusion Pipeline for semantic mixing of an image and a text prompt | [MagicMix](#magic-mix) | - | [Partho Das](https://github.com/daspartho) | +| Stable UnCLIP | Diffusion Pipeline for combining prior model (generate clip image embedding from text, UnCLIPPipeline `"kakaobrain/karlo-v1-alpha"`) and decoder pipeline (decode clip image embedding to image, StableDiffusionImageVariationPipeline `"lambdalabs/sd-image-variations-diffusers"` ). | [Stable UnCLIP](#stable-unclip) | - | [Ray Wang](https://wrong.wang) | +| UnCLIP Text Interpolation Pipeline | Diffusion Pipeline that allows passing two prompts and produces images while interpolating between the text-embeddings of the two prompts | [UnCLIP Text Interpolation Pipeline](#unclip-text-interpolation-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | +| UnCLIP Image Interpolation Pipeline | Diffusion Pipeline that allows passing two images/image_embeddings and produces images while interpolating between their image-embeddings | [UnCLIP Image Interpolation Pipeline](#unclip-image-interpolation-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | +| CLIP Guided Img2Img Stable Diffusion Pipeline | Doing CLIP guidance for image to image generation with Stable Diffusion | [CLIP Guided Stable Diffusion](#clip-guided-img2img-stable-diffusion) | - | [Nipun Jindal](https://github.com/nipunjindal/) | @@ -1033,4 +1034,67 @@ The resulting images in order:- ![result4](https://huggingface.co/datasets/NagaSaiAbhinay/UnCLIPImageInterpolationSamples/resolve/main/starry_to_flowers_4.png) ![result5](https://huggingface.co/datasets/NagaSaiAbhinay/UnCLIPImageInterpolationSamples/resolve/main/starry_to_flowers_5.png) +### CLIP Guided Img2Img Stable Diffusion + +CLIP guided Img2Img stable diffusion can help to generate more realistic images with an initial image +by guiding stable diffusion at every denoising step with an additional CLIP model. + +The following code requires roughly 12GB of GPU RAM. + +```python +from io import BytesIO + +import requests +import torch +from diffusers import DiffusionPipeline +from PIL import Image +from transformers import CLIPFeatureExtractor, CLIPModel + +feature_extractor = CLIPFeatureExtractor.from_pretrained( + "laion/CLIP-ViT-B-32-laion2B-s34B-b79K" +) +clip_model = CLIPModel.from_pretrained( + "laion/CLIP-ViT-B-32-laion2B-s34B-b79K", torch_dtype=torch.float16 +) + + +guided_pipeline = DiffusionPipeline.from_pretrained( + "CompVis/stable-diffusion-v1-4", + # custom_pipeline="clip_guided_stable_diffusion", + custom_pipeline="/home/njindal/diffusers/examples/community/clip_guided_stable_diffusion.py", + clip_model=clip_model, + feature_extractor=feature_extractor, + torch_dtype=torch.float16, +) +guided_pipeline.enable_attention_slicing() +guided_pipeline = guided_pipeline.to("cuda") + +prompt = "fantasy book cover, full moon, fantasy forest landscape, golden vector elements, fantasy magic, dark light night, intricate, elegant, sharp focus, illustration, highly detailed, digital painting, concept art, matte, art by WLOP and Artgerm and Albert Bierstadt, masterpiece" + +url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg" + +response = requests.get(url) +init_image = Image.open(BytesIO(response.content)).convert("RGB") + +image = guided_pipeline( + prompt=prompt, + num_inference_steps=30, + image=init_image, + strength=0.75, + guidance_scale=7.5, + clip_guidance_scale=100, + num_cutouts=4, + use_cutouts=False, +).images[0] +display(image) +``` + +Init Image + +![img2img_init_clip_guidance](https://huggingface.co/datasets/njindal/images/resolve/main/clip_guided_img2img_init.jpg) + +Output Image + +![img2img_clip_guidance](https://huggingface.co/datasets/njindal/images/resolve/main/clip_guided_img2img.jpg) + diff --git a/examples/community/clip_guided_stable_diffusion_img2img.py b/examples/community/clip_guided_stable_diffusion_img2img.py index 7642c73ace1d..43a2a229954f 100644 --- a/examples/community/clip_guided_stable_diffusion_img2img.py +++ b/examples/community/clip_guided_stable_diffusion_img2img.py @@ -1,85 +1,87 @@ -# Copyright 2023 The HuggingFace Team. All rights reserved. -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. - import inspect -from typing import Callable, List, Optional, Union +from typing import List, Optional, Union import numpy as np import PIL import torch -from packaging import version -from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer - -from diffusers import DiffusionPipeline -from diffusers.configuration_utils import FrozenDict -from diffusers.models import AutoencoderKL, UNet2DConditionModel -from diffusers.schedulers import KarrasDiffusionSchedulers - +from torch import nn +from torch.nn import functional as F +from torchvision import transforms +from transformers import CLIPFeatureExtractor, CLIPModel, CLIPTextModel, CLIPTokenizer + +from diffusers import ( + AutoencoderKL, + DDIMScheduler, + DiffusionPipeline, + LMSDiscreteScheduler, + PNDMScheduler, + UNet2DConditionModel, +) from diffusers.utils import ( PIL_INTERPOLATION, - deprecate, - is_accelerate_available, - is_accelerate_version, - logging, randn_tensor, - replace_example_docstring, ) -from diffusers.pipelines.stable_diffusion import StableDiffusionPipelineOutput -from diffusers.pipelines.stable_diffusion.safety_checker import StableDiffusionSafetyChecker - -logger = logging.get_logger(__name__) # pylint: disable=invalid-name +from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipelineOutput -# TODO: Change this EXAMPLE_DOC_STRING = """ Examples: - ```py - >>> import requests - >>> import torch - >>> from PIL import Image - >>> from io import BytesIO - - >>> from diffusers import StableDiffusionImg2ImgPipeline - - >>> device = "cuda" - >>> model_id_or_path = "runwayml/stable-diffusion-v1-5" - >>> pipe = StableDiffusionImg2ImgPipeline.from_pretrained(model_id_or_path, torch_dtype=torch.float16) - >>> pipe = pipe.to(device) + ``` + from io import BytesIO - >>> url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg" + import requests + import torch + from diffusers import DiffusionPipeline + from PIL import Image + from transformers import CLIPFeatureExtractor, CLIPModel - >>> response = requests.get(url) - >>> init_image = Image.open(BytesIO(response.content)).convert("RGB") - >>> init_image = init_image.resize((768, 512)) + feature_extractor = CLIPFeatureExtractor.from_pretrained( + "laion/CLIP-ViT-B-32-laion2B-s34B-b79K" + ) + clip_model = CLIPModel.from_pretrained( + "laion/CLIP-ViT-B-32-laion2B-s34B-b79K", torch_dtype=torch.float16 + ) - >>> prompt = "A fantasy landscape, trending on artstation" - >>> images = pipe(prompt=prompt, image=init_image, strength=0.75, guidance_scale=7.5).images - >>> images[0].save("fantasy_landscape.png") + guided_pipeline = DiffusionPipeline.from_pretrained( + "CompVis/stable-diffusion-v1-4", + # custom_pipeline="clip_guided_stable_diffusion", + custom_pipeline="/home/njindal/diffusers/examples/community/clip_guided_stable_diffusion.py", + clip_model=clip_model, + feature_extractor=feature_extractor, + torch_dtype=torch.float16, + ) + guided_pipeline.enable_attention_slicing() + guided_pipeline = guided_pipeline.to("cuda") + + prompt = "fantasy book cover, full moon, fantasy forest landscape, golden vector elements, fantasy magic, dark light night, intricate, elegant, sharp focus, illustration, highly detailed, digital painting, concept art, matte, art by WLOP and Artgerm and Albert Bierstadt, masterpiece" + + url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg" + + response = requests.get(url) + init_image = Image.open(BytesIO(response.content)).convert("RGB") + + image = guided_pipeline( + prompt=prompt, + num_inference_steps=30, + image=init_image, + strength=0.75, + guidance_scale=7.5, + clip_guidance_scale=100, + num_cutouts=4, + use_cutouts=False, + ).images[0] + display(image) ``` """ -def preprocess(image): +def preprocess(image, w, h): if isinstance(image, torch.Tensor): return image elif isinstance(image, PIL.Image.Image): image = [image] if isinstance(image[0], PIL.Image.Image): - w, h = image[0].size - w, h = map(lambda x: x - x % 8, (w, h)) # resize to integer multiple of 8 - image = [np.array(i.resize((w, h), resample=PIL_INTERPOLATION["lanczos"]))[None, :] for i in image] image = np.concatenate(image, axis=0) image = np.array(image).astype(np.float32) / 255.0 @@ -91,415 +93,97 @@ def preprocess(image): return image -class CLIPGuidedStableDiffusionImg2ImgPipeline(DiffusionPipeline): - r""" - Pipeline for text-guided image to image generation using Stable Diffusion. - - This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods the - library implements for all the pipelines (such as downloading or saving, running on a particular device, etc.) - - Args: - vae ([`AutoencoderKL`]): - Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations. - text_encoder ([`CLIPTextModel`]): - Frozen text-encoder. Stable Diffusion uses the text portion of - [CLIP](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel), specifically - the [clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14) variant. - tokenizer (`CLIPTokenizer`): - Tokenizer of class - [CLIPTokenizer](https://huggingface.co/docs/transformers/v4.21.0/en/model_doc/clip#transformers.CLIPTokenizer). - unet ([`UNet2DConditionModel`]): Conditional U-Net architecture to denoise the encoded image latents. - scheduler ([`SchedulerMixin`]): - A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of - [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`]. - safety_checker ([`StableDiffusionSafetyChecker`]): - Classification module that estimates whether generated images could be considered offensive or harmful. - Please, refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for details. - feature_extractor ([`CLIPFeatureExtractor`]): - Model that extracts features from generated images to be used as inputs for the `safety_checker`. - """ - _optional_components = ["safety_checker", "feature_extractor"] - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.__init__ - def __init__( - self, - vae: AutoencoderKL, - text_encoder: CLIPTextModel, - tokenizer: CLIPTokenizer, - unet: UNet2DConditionModel, - scheduler: KarrasDiffusionSchedulers, - safety_checker: StableDiffusionSafetyChecker, - feature_extractor: CLIPFeatureExtractor, - requires_safety_checker: bool = True, - ): +class MakeCutouts(nn.Module): + def __init__(self, cut_size, cut_power=1.0): super().__init__() - if hasattr(scheduler.config, "steps_offset") and scheduler.config.steps_offset != 1: - deprecation_message = ( - f"The configuration file of this scheduler: {scheduler} is outdated. `steps_offset`" - f" should be set to 1 instead of {scheduler.config.steps_offset}. Please make sure " - "to update the config accordingly as leaving `steps_offset` might led to incorrect results" - " in future versions. If you have downloaded this checkpoint from the Hugging Face Hub," - " it would be very nice if you could open a Pull request for the `scheduler/scheduler_config.json`" - " file" - ) - deprecate("steps_offset!=1", "1.0.0", deprecation_message, standard_warn=False) - new_config = dict(scheduler.config) - new_config["steps_offset"] = 1 - scheduler._internal_dict = FrozenDict(new_config) + self.cut_size = cut_size + self.cut_power = cut_power - if hasattr(scheduler.config, "clip_sample") and scheduler.config.clip_sample is True: - deprecation_message = ( - f"The configuration file of this scheduler: {scheduler} has not set the configuration `clip_sample`." - " `clip_sample` should be set to False in the configuration file. Please make sure to update the" - " config accordingly as not setting `clip_sample` in the config might lead to incorrect results in" - " future versions. If you have downloaded this checkpoint from the Hugging Face Hub, it would be very" - " nice if you could open a Pull request for the `scheduler/scheduler_config.json` file" - ) - deprecate("clip_sample not set", "1.0.0", deprecation_message, standard_warn=False) - new_config = dict(scheduler.config) - new_config["clip_sample"] = False - scheduler._internal_dict = FrozenDict(new_config) - - if safety_checker is None and requires_safety_checker: - logger.warning( - f"You have disabled the safety checker for {self.__class__} by passing `safety_checker=None`. Ensure" - " that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered" - " results in services or applications open to the public. Both the diffusers team and Hugging Face" - " strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling" - " it only for use-cases that involve analyzing network behavior or auditing its results. For more" - " information, please have a look at https://github.com/huggingface/diffusers/pull/254 ." - ) + def forward(self, pixel_values, num_cutouts): + sideY, sideX = pixel_values.shape[2:4] + max_size = min(sideX, sideY) + min_size = min(sideX, sideY, self.cut_size) + cutouts = [] + for _ in range(num_cutouts): + size = int(torch.rand([]) ** self.cut_power * (max_size - min_size) + min_size) + offsetx = torch.randint(0, sideX - size + 1, ()) + offsety = torch.randint(0, sideY - size + 1, ()) + cutout = pixel_values[:, :, offsety: offsety + size, offsetx: offsetx + size] + cutouts.append(F.adaptive_avg_pool2d(cutout, self.cut_size)) + return torch.cat(cutouts) - if safety_checker is not None and feature_extractor is None: - raise ValueError( - "Make sure to define a feature extractor when loading {self.__class__} if you want to use the safety" - " checker. If you do not want to use the safety checker, you can pass `'safety_checker=None'` instead." - ) - is_unet_version_less_0_9_0 = hasattr(unet.config, "_diffusers_version") and version.parse( - version.parse(unet.config._diffusers_version).base_version - ) < version.parse("0.9.0.dev0") - is_unet_sample_size_less_64 = hasattr(unet.config, "sample_size") and unet.config.sample_size < 64 - if is_unet_version_less_0_9_0 and is_unet_sample_size_less_64: - deprecation_message = ( - "The configuration file of the unet has set the default `sample_size` to smaller than" - " 64 which seems highly unlikely. If your checkpoint is a fine-tuned version of any of the" - " following: \n- CompVis/stable-diffusion-v1-4 \n- CompVis/stable-diffusion-v1-3 \n-" - " CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- runwayml/stable-diffusion-v1-5" - " \n- runwayml/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the" - " configuration file. Please make sure to update the config accordingly as leaving `sample_size=32`" - " in the config might lead to incorrect results in future versions. If you have downloaded this" - " checkpoint from the Hugging Face Hub, it would be very nice if you could open a Pull request for" - " the `unet/config.json` file" - ) - deprecate("sample_size<64", "1.0.0", deprecation_message, standard_warn=False) - new_config = dict(unet.config) - new_config["sample_size"] = 64 - unet._internal_dict = FrozenDict(new_config) +def spherical_dist_loss(x, y): + x = F.normalize(x, dim=-1) + y = F.normalize(y, dim=-1) + return (x - y).norm(dim=-1).div(2).arcsin().pow(2).mul(2) + + +def set_requires_grad(model, value): + for param in model.parameters(): + param.requires_grad = value + + +class CLIPGuidedStableDiffusion(DiffusionPipeline): + """CLIP guided stable diffusion based on the amazing repo by @crowsonkb and @Jack000 + - https://github.com/Jack000/glid-3-xl + - https://github.dev/crowsonkb/k-diffusion + """ + def __init__( + self, + vae: AutoencoderKL, + text_encoder: CLIPTextModel, + clip_model: CLIPModel, + tokenizer: CLIPTokenizer, + unet: UNet2DConditionModel, + scheduler: Union[PNDMScheduler, LMSDiscreteScheduler, DDIMScheduler], + feature_extractor: CLIPFeatureExtractor, + ): + super().__init__() self.register_modules( vae=vae, text_encoder=text_encoder, + clip_model=clip_model, tokenizer=tokenizer, unet=unet, scheduler=scheduler, - safety_checker=safety_checker, feature_extractor=feature_extractor, ) - self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1) - self.register_to_config(requires_safety_checker=requires_safety_checker) - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_sequential_cpu_offload - def enable_sequential_cpu_offload(self, gpu_id=0): - r""" - Offloads all models to CPU using accelerate, significantly reducing memory usage. When called, unet, - text_encoder, vae and safety checker have their state dicts saved to CPU and then are moved to a - `torch.device('meta') and loaded to GPU only when their specific submodule has its `forward` method called. - Note that offloading happens on a submodule basis. Memory savings are higher than with - `enable_model_cpu_offload`, but performance is lower. - """ - if is_accelerate_available() and is_accelerate_version(">=", "0.14.0"): - from accelerate import cpu_offload - else: - raise ImportError("`enable_sequential_cpu_offload` requires `accelerate v0.14.0` or higher") - - device = torch.device(f"cuda:{gpu_id}") - - if self.device.type != "cpu": - self.to("cpu", silence_dtype_warnings=True) - torch.cuda.empty_cache() # otherwise we don't see the memory savings (but they probably exist) - - for cpu_offloaded_model in [self.unet, self.text_encoder, self.vae]: - cpu_offload(cpu_offloaded_model, device) - - if self.safety_checker is not None: - cpu_offload(self.safety_checker, execution_device=device, offload_buffers=True) - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_model_cpu_offload - def enable_model_cpu_offload(self, gpu_id=0): - r""" - Offloads all models to CPU using accelerate, reducing memory usage with a low impact on performance. Compared - to `enable_sequential_cpu_offload`, this method moves one whole model at a time to the GPU when its `forward` - method is called, and the model remains in GPU until the next model runs. Memory savings are lower than with - `enable_sequential_cpu_offload`, but performance is much better due to the iterative execution of the `unet`. - """ - if is_accelerate_available() and is_accelerate_version(">=", "0.17.0.dev0"): - from accelerate import cpu_offload_with_hook - else: - raise ImportError("`enable_model_offload` requires `accelerate v0.17.0` or higher.") - - device = torch.device(f"cuda:{gpu_id}") - - if self.device.type != "cpu": - self.to("cpu", silence_dtype_warnings=True) - torch.cuda.empty_cache() # otherwise we don't see the memory savings (but they probably exist) - - hook = None - for cpu_offloaded_model in [self.text_encoder, self.unet, self.vae]: - _, hook = cpu_offload_with_hook(cpu_offloaded_model, device, prev_module_hook=hook) - - if self.safety_checker is not None: - _, hook = cpu_offload_with_hook(self.safety_checker, device, prev_module_hook=hook) - - # We'll offload the last model manually. - self.final_offload_hook = hook - - @property - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline._execution_device - def _execution_device(self): - r""" - Returns the device on which the pipeline's models will be executed. After calling - `pipeline.enable_sequential_cpu_offload()` the execution device can only be inferred from Accelerate's module - hooks. - """ - if not hasattr(self.unet, "_hf_hook"): - return self.device - for module in self.unet.modules(): - if ( - hasattr(module, "_hf_hook") - and hasattr(module._hf_hook, "execution_device") - and module._hf_hook.execution_device is not None - ): - return torch.device(module._hf_hook.execution_device) - return self.device - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline._encode_prompt - def _encode_prompt( - self, - prompt, - device, - num_images_per_prompt, - do_classifier_free_guidance, - negative_prompt=None, - prompt_embeds: Optional[torch.FloatTensor] = None, - negative_prompt_embeds: Optional[torch.FloatTensor] = None, - ): - r""" - Encodes the prompt into text encoder hidden states. - - Args: - prompt (`str` or `List[str]`, *optional*): - prompt to be encoded - device: (`torch.device`): - torch device - num_images_per_prompt (`int`): - number of images that should be generated per prompt - do_classifier_free_guidance (`bool`): - whether to use classifier free guidance or not - negative_prompt (`str` or `List[str]`, *optional*): - The prompt or prompts not to guide the image generation. If not defined, one has to pass - `negative_prompt_embeds`. instead. If not defined, one has to pass `negative_prompt_embeds`. instead. - Ignored when not using guidance (i.e., ignored if `guidance_scale` is less than `1`). - prompt_embeds (`torch.FloatTensor`, *optional*): - Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not - provided, text embeddings will be generated from `prompt` input argument. - negative_prompt_embeds (`torch.FloatTensor`, *optional*): - Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt - weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input - argument. - """ - if prompt is not None and isinstance(prompt, str): - batch_size = 1 - elif prompt is not None and isinstance(prompt, list): - batch_size = len(prompt) - else: - batch_size = prompt_embeds.shape[0] - - if prompt_embeds is None: - text_inputs = self.tokenizer( - prompt, - padding="max_length", - max_length=self.tokenizer.model_max_length, - truncation=True, - return_tensors="pt", - ) - text_input_ids = text_inputs.input_ids - untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="pt").input_ids - - if untruncated_ids.shape[-1] >= text_input_ids.shape[-1] and not torch.equal( - text_input_ids, untruncated_ids - ): - removed_text = self.tokenizer.batch_decode( - untruncated_ids[:, self.tokenizer.model_max_length - 1 : -1] - ) - logger.warning( - "The following part of your input was truncated because CLIP can only handle sequences up to" - f" {self.tokenizer.model_max_length} tokens: {removed_text}" - ) - - if hasattr(self.text_encoder.config, "use_attention_mask") and self.text_encoder.config.use_attention_mask: - attention_mask = text_inputs.attention_mask.to(device) - else: - attention_mask = None - - prompt_embeds = self.text_encoder( - text_input_ids.to(device), - attention_mask=attention_mask, - ) - prompt_embeds = prompt_embeds[0] - - prompt_embeds = prompt_embeds.to(dtype=self.text_encoder.dtype, device=device) - - bs_embed, seq_len, _ = prompt_embeds.shape - # duplicate text embeddings for each generation per prompt, using mps friendly method - prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1) - prompt_embeds = prompt_embeds.view(bs_embed * num_images_per_prompt, seq_len, -1) - - # get unconditional embeddings for classifier free guidance - if do_classifier_free_guidance and negative_prompt_embeds is None: - uncond_tokens: List[str] - if negative_prompt is None: - uncond_tokens = [""] * batch_size - elif type(prompt) is not type(negative_prompt): - raise TypeError( - f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !=" - f" {type(prompt)}." - ) - elif isinstance(negative_prompt, str): - uncond_tokens = [negative_prompt] - elif batch_size != len(negative_prompt): - raise ValueError( - f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:" - f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches" - " the batch size of `prompt`." - ) - else: - uncond_tokens = negative_prompt - - max_length = prompt_embeds.shape[1] - uncond_input = self.tokenizer( - uncond_tokens, - padding="max_length", - max_length=max_length, - truncation=True, - return_tensors="pt", - ) - - if hasattr(self.text_encoder.config, "use_attention_mask") and self.text_encoder.config.use_attention_mask: - attention_mask = uncond_input.attention_mask.to(device) - else: - attention_mask = None - - negative_prompt_embeds = self.text_encoder( - uncond_input.input_ids.to(device), - attention_mask=attention_mask, - ) - negative_prompt_embeds = negative_prompt_embeds[0] - - if do_classifier_free_guidance: - # duplicate unconditional embeddings for each generation per prompt, using mps friendly method - seq_len = negative_prompt_embeds.shape[1] - - negative_prompt_embeds = negative_prompt_embeds.to(dtype=self.text_encoder.dtype, device=device) - - negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1) - negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1) - - # For classifier free guidance, we need to do two forward passes. - # Here we concatenate the unconditional and text embeddings into a single batch - # to avoid doing two forward passes - prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds]) - - return prompt_embeds - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.run_safety_checker - def run_safety_checker(self, image, device, dtype): - if self.safety_checker is not None: - safety_checker_input = self.feature_extractor(self.numpy_to_pil(image), return_tensors="pt").to(device) - image, has_nsfw_concept = self.safety_checker( - images=image, clip_input=safety_checker_input.pixel_values.to(dtype) - ) - else: - has_nsfw_concept = None - return image, has_nsfw_concept - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.decode_latents - def decode_latents(self, latents): - latents = 1 / self.vae.config.scaling_factor * latents - image = self.vae.decode(latents).sample - image = (image / 2 + 0.5).clamp(0, 1) - # we always cast to float32 as this does not cause significant overhead and is compatible with bfloat16 - image = image.cpu().permute(0, 2, 3, 1).float().numpy() - return image - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_extra_step_kwargs - def prepare_extra_step_kwargs(self, generator, eta): - # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature - # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers. - # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502 - # and should be between [0, 1] + self.normalize = transforms.Normalize(mean=feature_extractor.image_mean, std=feature_extractor.image_std) + self.cut_out_size = ( + feature_extractor.size + if isinstance(feature_extractor.size, int) + else feature_extractor.size["shortest_edge"] + ) + self.make_cutouts = MakeCutouts(self.cut_out_size) - accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys()) - extra_step_kwargs = {} - if accepts_eta: - extra_step_kwargs["eta"] = eta + set_requires_grad(self.text_encoder, False) + set_requires_grad(self.clip_model, False) - # check if the scheduler accepts generator - accepts_generator = "generator" in set(inspect.signature(self.scheduler.step).parameters.keys()) - if accepts_generator: - extra_step_kwargs["generator"] = generator - return extra_step_kwargs + def enable_attention_slicing(self, slice_size: Optional[Union[str, int]] = "auto"): + if slice_size == "auto": + # half the attention head size is usually a good trade-off between + # speed and memory + slice_size = self.unet.config.attention_head_dim // 2 + self.unet.set_attention_slice(slice_size) - def check_inputs( - self, prompt, strength, callback_steps, negative_prompt=None, prompt_embeds=None, negative_prompt_embeds=None - ): - if strength < 0 or strength > 1: - raise ValueError(f"The value of strength should in [0.0, 1.0] but is {strength}") + def disable_attention_slicing(self): + self.enable_attention_slicing(None) - if (callback_steps is None) or ( - callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0) - ): - raise ValueError( - f"`callback_steps` has to be a positive integer but is {callback_steps} of type" - f" {type(callback_steps)}." - ) + def freeze_vae(self): + set_requires_grad(self.vae, False) - if prompt is not None and prompt_embeds is not None: - raise ValueError( - f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to" - " only forward one of the two." - ) - elif prompt is None and prompt_embeds is None: - raise ValueError( - "Provide either `prompt` or `prompt_embeds`. Cannot leave both `prompt` and `prompt_embeds` undefined." - ) - elif prompt is not None and (not isinstance(prompt, str) and not isinstance(prompt, list)): - raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}") + def unfreeze_vae(self): + set_requires_grad(self.vae, True) - if negative_prompt is not None and negative_prompt_embeds is not None: - raise ValueError( - f"Cannot forward both `negative_prompt`: {negative_prompt} and `negative_prompt_embeds`:" - f" {negative_prompt_embeds}. Please make sure to only forward one of the two." - ) + def freeze_unet(self): + set_requires_grad(self.unet, False) - if prompt_embeds is not None and negative_prompt_embeds is not None: - if prompt_embeds.shape != negative_prompt_embeds.shape: - raise ValueError( - "`prompt_embeds` and `negative_prompt_embeds` must have the same shape when passed directly, but" - f" got: `prompt_embeds` {prompt_embeds.shape} != `negative_prompt_embeds`" - f" {negative_prompt_embeds.shape}." - ) + def unfreeze_unet(self): + set_requires_grad(self.unet, True) def get_timesteps(self, num_inference_steps, strength, device): # get the original timestep using init_timestep @@ -527,7 +211,7 @@ def prepare_latents(self, image, timestep, batch_size, num_images_per_prompt, dt if isinstance(generator, list): init_latents = [ - self.vae.encode(image[i : i + 1]).latent_dist.sample(generator[i]) for i in range(batch_size) + self.vae.encode(image[i: i + 1]).latent_dist.sample(generator[i]) for i in range(batch_size) ] init_latents = torch.cat(init_latents, dim=0) else: @@ -562,136 +246,208 @@ def prepare_latents(self, image, timestep, batch_size, num_images_per_prompt, dt return latents + @torch.enable_grad() + def cond_fn( + self, + latents, + timestep, + index, + text_embeddings, + noise_pred_original, + text_embeddings_clip, + clip_guidance_scale, + num_cutouts, + use_cutouts=True, + ): + latents = latents.detach().requires_grad_() + + if isinstance(self.scheduler, LMSDiscreteScheduler): + sigma = self.scheduler.sigmas[index] + # the model input needs to be scaled to match the continuous ODE formulation in K-LMS + latent_model_input = latents / ((sigma ** 2 + 1) ** 0.5) + else: + latent_model_input = latents + + # predict the noise residual + noise_pred = self.unet(latent_model_input, timestep, encoder_hidden_states=text_embeddings).sample + + if isinstance(self.scheduler, (PNDMScheduler, DDIMScheduler)): + alpha_prod_t = self.scheduler.alphas_cumprod[timestep] + beta_prod_t = 1 - alpha_prod_t + # compute predicted original sample from predicted noise also called + # "predicted x_0" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf + pred_original_sample = (latents - beta_prod_t ** (0.5) * noise_pred) / alpha_prod_t ** (0.5) + + fac = torch.sqrt(beta_prod_t) + sample = pred_original_sample * (fac) + latents * (1 - fac) + elif isinstance(self.scheduler, LMSDiscreteScheduler): + sigma = self.scheduler.sigmas[index] + sample = latents - sigma * noise_pred + else: + raise ValueError(f"scheduler type {type(self.scheduler)} not supported") + + sample = 1 / self.vae.config.scaling_factor * sample + image = self.vae.decode(sample).sample + image = (image / 2 + 0.5).clamp(0, 1) + + if use_cutouts: + image = self.make_cutouts(image, num_cutouts) + else: + image = transforms.Resize(self.cut_out_size)(image) + image = self.normalize(image).to(latents.dtype) + + image_embeddings_clip = self.clip_model.get_image_features(image) + image_embeddings_clip = image_embeddings_clip / image_embeddings_clip.norm(p=2, dim=-1, keepdim=True) + + if use_cutouts: + dists = spherical_dist_loss(image_embeddings_clip, text_embeddings_clip) + dists = dists.view([num_cutouts, sample.shape[0], -1]) + loss = dists.sum(2).mean(0).sum() * clip_guidance_scale + else: + loss = spherical_dist_loss(image_embeddings_clip, text_embeddings_clip).mean() * clip_guidance_scale + + grads = -torch.autograd.grad(loss, latents)[0] + + if isinstance(self.scheduler, LMSDiscreteScheduler): + latents = latents.detach() + grads * (sigma ** 2) + noise_pred = noise_pred_original + else: + noise_pred = noise_pred_original - torch.sqrt(beta_prod_t) * grads + return noise_pred, latents + @torch.no_grad() - @replace_example_docstring(EXAMPLE_DOC_STRING) def __call__( - self, - prompt: Union[str, List[str]] = None, - image: Union[torch.FloatTensor, PIL.Image.Image] = None, - strength: float = 0.8, - num_inference_steps: Optional[int] = 50, - guidance_scale: Optional[float] = 7.5, - negative_prompt: Optional[Union[str, List[str]]] = None, - num_images_per_prompt: Optional[int] = 1, - eta: Optional[float] = 0.0, - generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None, - prompt_embeds: Optional[torch.FloatTensor] = None, - negative_prompt_embeds: Optional[torch.FloatTensor] = None, - output_type: Optional[str] = "pil", - return_dict: bool = True, - callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None, - callback_steps: int = 1, + self, + prompt: Union[str, List[str]], + height: Optional[int] = 512, + width: Optional[int] = 512, + image: Union[torch.FloatTensor, PIL.Image.Image] = None, + strength: float = 0.8, + num_inference_steps: Optional[int] = 50, + guidance_scale: Optional[float] = 7.5, + num_images_per_prompt: Optional[int] = 1, + eta: float = 0.0, + clip_guidance_scale: Optional[float] = 100, + clip_prompt: Optional[Union[str, List[str]]] = None, + num_cutouts: Optional[int] = 4, + use_cutouts: Optional[bool] = True, + generator: Optional[torch.Generator] = None, + latents: Optional[torch.FloatTensor] = None, + output_type: Optional[str] = "pil", + return_dict: bool = True, ): - r""" - Function invoked when calling the pipeline for generation. - - Args: - prompt (`str` or `List[str]`, *optional*): - The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`. - instead. - image (`torch.FloatTensor` or `PIL.Image.Image`): - `Image`, or tensor representing an image batch, that will be used as the starting point for the - process. - strength (`float`, *optional*, defaults to 0.8): - Conceptually, indicates how much to transform the reference `image`. Must be between 0 and 1. `image` - will be used as a starting point, adding more noise to it the larger the `strength`. The number of - denoising steps depends on the amount of noise initially added. When `strength` is 1, added noise will - be maximum and the denoising process will run for the full number of iterations specified in - `num_inference_steps`. A value of 1, therefore, essentially ignores `image`. - num_inference_steps (`int`, *optional*, defaults to 50): - The number of denoising steps. More denoising steps usually lead to a higher quality image at the - expense of slower inference. This parameter will be modulated by `strength`. - guidance_scale (`float`, *optional*, defaults to 7.5): - Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598). - `guidance_scale` is defined as `w` of equation 2. of [Imagen - Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale > - 1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`, - usually at the expense of lower image quality. - negative_prompt (`str` or `List[str]`, *optional*): - The prompt or prompts not to guide the image generation. If not defined, one has to pass - `negative_prompt_embeds`. instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` - is less than `1`). - num_images_per_prompt (`int`, *optional*, defaults to 1): - The number of images to generate per prompt. - eta (`float`, *optional*, defaults to 0.0): - Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to - [`schedulers.DDIMScheduler`], will be ignored for others. - generator (`torch.Generator`, *optional*): - One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html) - to make generation deterministic. - prompt_embeds (`torch.FloatTensor`, *optional*): - Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not - provided, text embeddings will be generated from `prompt` input argument. - negative_prompt_embeds (`torch.FloatTensor`, *optional*): - Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt - weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input - argument. - output_type (`str`, *optional*, defaults to `"pil"`): - The output format of the generate image. Choose between - [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`. - return_dict (`bool`, *optional*, defaults to `True`): - Whether or not to return a [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] instead of a - plain tuple. - callback (`Callable`, *optional*): - A function that will be called every `callback_steps` steps during inference. The function will be - called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`. - callback_steps (`int`, *optional*, defaults to 1): - The frequency at which the `callback` function will be called. If not specified, the callback will be - called at every step. - Examples: - - Returns: - [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] or `tuple`: - [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] if `return_dict` is True, otherwise a `tuple. - When returning a tuple, the first element is a list with the generated images, and the second element is a - list of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work" - (nsfw) content, according to the `safety_checker`. - """ - # 1. Check inputs. Raise error if not correct - self.check_inputs(prompt, strength, callback_steps, negative_prompt, prompt_embeds, negative_prompt_embeds) - - # 2. Define call parameters - if prompt is not None and isinstance(prompt, str): + if isinstance(prompt, str): batch_size = 1 - elif prompt is not None and isinstance(prompt, list): + elif isinstance(prompt, list): batch_size = len(prompt) else: - batch_size = prompt_embeds.shape[0] - device = self._execution_device - # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2) - # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1` - # corresponds to doing no classifier free guidance. - do_classifier_free_guidance = guidance_scale > 1.0 + raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}") - # 3. Encode input prompt - prompt_embeds = self._encode_prompt( + if height % 8 != 0 or width % 8 != 0: + raise ValueError(f"`height` and `width` have to be divisible by 8 but are {height} and {width}.") + + # get prompt text embeddings + text_input = self.tokenizer( prompt, - device, - num_images_per_prompt, - do_classifier_free_guidance, - negative_prompt, - prompt_embeds=prompt_embeds, - negative_prompt_embeds=negative_prompt_embeds, + padding="max_length", + max_length=self.tokenizer.model_max_length, + truncation=True, + return_tensors="pt", ) - - # 4. Preprocess image - image = preprocess(image) - - # 5. set timesteps - self.scheduler.set_timesteps(num_inference_steps, device=device) - timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength, device) + text_embeddings = self.text_encoder(text_input.input_ids.to(self.device))[0] + # duplicate text embeddings for each generation per prompt + text_embeddings = text_embeddings.repeat_interleave(num_images_per_prompt, dim=0) + + # set timesteps + accepts_offset = "offset" in set(inspect.signature(self.scheduler.set_timesteps).parameters.keys()) + extra_set_kwargs = {} + if accepts_offset: + extra_set_kwargs["offset"] = 1 + + self.scheduler.set_timesteps(num_inference_steps, **extra_set_kwargs) + # Some schedulers like PNDM have timesteps as arrays + # It's more optimized to move all timesteps to correct device beforehand + timesteps_tensor = self.scheduler.timesteps.to(self.device) + + timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength, self.device) latent_timestep = timesteps[:1].repeat(batch_size * num_images_per_prompt) - # 6. Prepare latent variables + # Preprocess image + image = preprocess(image, width, height) latents = self.prepare_latents( - image, latent_timestep, batch_size, num_images_per_prompt, prompt_embeds.dtype, device, generator + image, latent_timestep, batch_size, num_images_per_prompt, text_embeddings.dtype, self.device, generator ) - # 7. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline - extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta) + if clip_guidance_scale > 0: + if clip_prompt is not None: + clip_text_input = self.tokenizer( + clip_prompt, + padding="max_length", + max_length=self.tokenizer.model_max_length, + truncation=True, + return_tensors="pt", + ).input_ids.to(self.device) + else: + clip_text_input = text_input.input_ids.to(self.device) + text_embeddings_clip = self.clip_model.get_text_features(clip_text_input) + text_embeddings_clip = text_embeddings_clip / text_embeddings_clip.norm(p=2, dim=-1, keepdim=True) + # duplicate text embeddings clip for each generation per prompt + text_embeddings_clip = text_embeddings_clip.repeat_interleave(num_images_per_prompt, dim=0) + + # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2) + # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1` + # corresponds to doing no classifier free guidance. + do_classifier_free_guidance = guidance_scale > 1.0 + # get unconditional embeddings for classifier free guidance + if do_classifier_free_guidance: + max_length = text_input.input_ids.shape[-1] + uncond_input = self.tokenizer([""], padding="max_length", max_length=max_length, return_tensors="pt") + uncond_embeddings = self.text_encoder(uncond_input.input_ids.to(self.device))[0] + # duplicate unconditional embeddings for each generation per prompt + uncond_embeddings = uncond_embeddings.repeat_interleave(num_images_per_prompt, dim=0) + + # For classifier free guidance, we need to do two forward passes. + # Here we concatenate the unconditional and text embeddings into a single batch + # to avoid doing two forward passes + text_embeddings = torch.cat([uncond_embeddings, text_embeddings]) + + # get the initial random noise unless the user supplied it + + # Unlike in other pipelines, latents need to be generated in the target device + # for 1-to-1 results reproducibility with the CompVis implementation. + # However this currently doesn't work in `mps`. + latents_shape = (batch_size * num_images_per_prompt, self.unet.in_channels, height // 8, width // 8) + latents_dtype = text_embeddings.dtype + if latents is None: + if self.device.type == "mps": + # randn does not work reproducibly on mps + latents = torch.randn(latents_shape, generator=generator, device="cpu", dtype=latents_dtype).to( + self.device + ) + else: + latents = torch.randn(latents_shape, generator=generator, device=self.device, dtype=latents_dtype) + else: + if latents.shape != latents_shape: + raise ValueError(f"Unexpected latents shape, got {latents.shape}, expected {latents_shape}") + latents = latents.to(self.device) + + # scale the initial noise by the standard deviation required by the scheduler + latents = latents * self.scheduler.init_noise_sigma + + # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature + # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers. + # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502 + # and should be between [0, 1] + accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys()) + extra_step_kwargs = {} + if accepts_eta: + extra_step_kwargs["eta"] = eta + + # check if the scheduler accepts generator + accepts_generator = "generator" in set(inspect.signature(self.scheduler.step).parameters.keys()) + if accepts_generator: + extra_step_kwargs["generator"] = generator - # 8. Denoising loop - num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order with self.progress_bar(total=num_inference_steps) as progress_bar: for i, t in enumerate(timesteps): # expand the latents if we are doing classifier free guidance @@ -699,37 +455,44 @@ def __call__( latent_model_input = self.scheduler.scale_model_input(latent_model_input, t) # predict the noise residual - noise_pred = self.unet(latent_model_input, t, encoder_hidden_states=prompt_embeds).sample + noise_pred = self.unet(latent_model_input, t, encoder_hidden_states=text_embeddings).sample - # perform guidance + # perform classifier free guidance if do_classifier_free_guidance: noise_pred_uncond, noise_pred_text = noise_pred.chunk(2) noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond) + # perform clip guidance + if clip_guidance_scale > 0: + text_embeddings_for_guidance = ( + text_embeddings.chunk(2)[1] if do_classifier_free_guidance else text_embeddings + ) + noise_pred, latents = self.cond_fn( + latents, + t, + i, + text_embeddings_for_guidance, + noise_pred, + text_embeddings_clip, + clip_guidance_scale, + num_cutouts, + use_cutouts, + ) + # compute the previous noisy sample x_t -> x_t-1 latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs).prev_sample - # call the callback, if provided - if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0): - progress_bar.update() - if callback is not None and i % callback_steps == 0: - callback(i, t, latents) - - # 9. Post-processing - image = self.decode_latents(latents) + # scale and decode the image latents with vae + latents = 1 / self.vae.config.scaling_factor * latents + image = self.vae.decode(latents).sample - # 10. Run safety checker - image, has_nsfw_concept = self.run_safety_checker(image, device, prompt_embeds.dtype) + image = (image / 2 + 0.5).clamp(0, 1) + image = image.cpu().permute(0, 2, 3, 1).numpy() - # 11. Convert to PIL if output_type == "pil": image = self.numpy_to_pil(image) - # Offload last model to CPU - if hasattr(self, "final_offload_hook") and self.final_offload_hook is not None: - self.final_offload_hook.offload() - if not return_dict: - return (image, has_nsfw_concept) + return (image, None) - return StableDiffusionPipelineOutput(images=image, nsfw_content_detected=has_nsfw_concept) + return StableDiffusionPipelineOutput(images=image, nsfw_content_detected=None) \ No newline at end of file From f08bb97f7dc11532cc14387758acb35dff139ab8 Mon Sep 17 00:00:00 2001 From: njindal Date: Fri, 17 Mar 2023 20:06:46 +0530 Subject: [PATCH 3/4] [Img2Img]: img2img pipeline --- .../clip_guided_stable_diffusion_img2img.py | 90 ++++++++++--------- 1 file changed, 46 insertions(+), 44 deletions(-) diff --git a/examples/community/clip_guided_stable_diffusion_img2img.py b/examples/community/clip_guided_stable_diffusion_img2img.py index 43a2a229954f..c9d2bc6e5931 100644 --- a/examples/community/clip_guided_stable_diffusion_img2img.py +++ b/examples/community/clip_guided_stable_diffusion_img2img.py @@ -17,11 +17,13 @@ PNDMScheduler, UNet2DConditionModel, ) +from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipelineOutput from diffusers.utils import ( PIL_INTERPOLATION, + deprecate, randn_tensor, ) -from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipelineOutput + EXAMPLE_DOC_STRING = """ Examples: @@ -109,7 +111,7 @@ def forward(self, pixel_values, num_cutouts): size = int(torch.rand([]) ** self.cut_power * (max_size - min_size) + min_size) offsetx = torch.randint(0, sideX - size + 1, ()) offsety = torch.randint(0, sideY - size + 1, ()) - cutout = pixel_values[:, :, offsety: offsety + size, offsetx: offsetx + size] + cutout = pixel_values[:, :, offsety : offsety + size, offsetx : offsetx + size] cutouts.append(F.adaptive_avg_pool2d(cutout, self.cut_size)) return torch.cat(cutouts) @@ -132,14 +134,14 @@ class CLIPGuidedStableDiffusion(DiffusionPipeline): """ def __init__( - self, - vae: AutoencoderKL, - text_encoder: CLIPTextModel, - clip_model: CLIPModel, - tokenizer: CLIPTokenizer, - unet: UNet2DConditionModel, - scheduler: Union[PNDMScheduler, LMSDiscreteScheduler, DDIMScheduler], - feature_extractor: CLIPFeatureExtractor, + self, + vae: AutoencoderKL, + text_encoder: CLIPTextModel, + clip_model: CLIPModel, + tokenizer: CLIPTokenizer, + unet: UNet2DConditionModel, + scheduler: Union[PNDMScheduler, LMSDiscreteScheduler, DDIMScheduler], + feature_extractor: CLIPFeatureExtractor, ): super().__init__() self.register_modules( @@ -211,7 +213,7 @@ def prepare_latents(self, image, timestep, batch_size, num_images_per_prompt, dt if isinstance(generator, list): init_latents = [ - self.vae.encode(image[i: i + 1]).latent_dist.sample(generator[i]) for i in range(batch_size) + self.vae.encode(image[i : i + 1]).latent_dist.sample(generator[i]) for i in range(batch_size) ] init_latents = torch.cat(init_latents, dim=0) else: @@ -248,23 +250,23 @@ def prepare_latents(self, image, timestep, batch_size, num_images_per_prompt, dt @torch.enable_grad() def cond_fn( - self, - latents, - timestep, - index, - text_embeddings, - noise_pred_original, - text_embeddings_clip, - clip_guidance_scale, - num_cutouts, - use_cutouts=True, + self, + latents, + timestep, + index, + text_embeddings, + noise_pred_original, + text_embeddings_clip, + clip_guidance_scale, + num_cutouts, + use_cutouts=True, ): latents = latents.detach().requires_grad_() if isinstance(self.scheduler, LMSDiscreteScheduler): sigma = self.scheduler.sigmas[index] # the model input needs to be scaled to match the continuous ODE formulation in K-LMS - latent_model_input = latents / ((sigma ** 2 + 1) ** 0.5) + latent_model_input = latents / ((sigma**2 + 1) ** 0.5) else: latent_model_input = latents @@ -309,7 +311,7 @@ def cond_fn( grads = -torch.autograd.grad(loss, latents)[0] if isinstance(self.scheduler, LMSDiscreteScheduler): - latents = latents.detach() + grads * (sigma ** 2) + latents = latents.detach() + grads * (sigma**2) noise_pred = noise_pred_original else: noise_pred = noise_pred_original - torch.sqrt(beta_prod_t) * grads @@ -317,24 +319,24 @@ def cond_fn( @torch.no_grad() def __call__( - self, - prompt: Union[str, List[str]], - height: Optional[int] = 512, - width: Optional[int] = 512, - image: Union[torch.FloatTensor, PIL.Image.Image] = None, - strength: float = 0.8, - num_inference_steps: Optional[int] = 50, - guidance_scale: Optional[float] = 7.5, - num_images_per_prompt: Optional[int] = 1, - eta: float = 0.0, - clip_guidance_scale: Optional[float] = 100, - clip_prompt: Optional[Union[str, List[str]]] = None, - num_cutouts: Optional[int] = 4, - use_cutouts: Optional[bool] = True, - generator: Optional[torch.Generator] = None, - latents: Optional[torch.FloatTensor] = None, - output_type: Optional[str] = "pil", - return_dict: bool = True, + self, + prompt: Union[str, List[str]], + height: Optional[int] = 512, + width: Optional[int] = 512, + image: Union[torch.FloatTensor, PIL.Image.Image] = None, + strength: float = 0.8, + num_inference_steps: Optional[int] = 50, + guidance_scale: Optional[float] = 7.5, + num_images_per_prompt: Optional[int] = 1, + eta: float = 0.0, + clip_guidance_scale: Optional[float] = 100, + clip_prompt: Optional[Union[str, List[str]]] = None, + num_cutouts: Optional[int] = 4, + use_cutouts: Optional[bool] = True, + generator: Optional[torch.Generator] = None, + latents: Optional[torch.FloatTensor] = None, + output_type: Optional[str] = "pil", + return_dict: bool = True, ): if isinstance(prompt, str): batch_size = 1 @@ -367,7 +369,7 @@ def __call__( self.scheduler.set_timesteps(num_inference_steps, **extra_set_kwargs) # Some schedulers like PNDM have timesteps as arrays # It's more optimized to move all timesteps to correct device beforehand - timesteps_tensor = self.scheduler.timesteps.to(self.device) + self.scheduler.timesteps.to(self.device) timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, strength, self.device) latent_timestep = timesteps[:1].repeat(batch_size * num_images_per_prompt) @@ -448,7 +450,7 @@ def __call__( if accepts_generator: extra_step_kwargs["generator"] = generator - with self.progress_bar(total=num_inference_steps) as progress_bar: + with self.progress_bar(total=num_inference_steps): for i, t in enumerate(timesteps): # expand the latents if we are doing classifier free guidance latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents @@ -495,4 +497,4 @@ def __call__( if not return_dict: return (image, None) - return StableDiffusionPipelineOutput(images=image, nsfw_content_detected=None) \ No newline at end of file + return StableDiffusionPipelineOutput(images=image, nsfw_content_detected=None) From 5ea2662f00c81fee95a38ae7af3ebb5b579e9904 Mon Sep 17 00:00:00 2001 From: njindal Date: Fri, 17 Mar 2023 20:18:44 +0530 Subject: [PATCH 4/4] [Img2Img]: img2img pipeline --- examples/community/README.md | 58 ++++++++++++++++-------------------- 1 file changed, 25 insertions(+), 33 deletions(-) diff --git a/examples/community/README.md b/examples/community/README.md index d6e689e3fff0..ba0cc0344643 100644 --- a/examples/community/README.md +++ b/examples/community/README.md @@ -6,31 +6,31 @@ Please have a look at the following table to get an overview of all community examples. Click on the **Code Example** to get a copy-and-paste ready code example that you can try out. If a community doesn't work as expected, please open an issue and ping the author on it. -| Example | Description | Code Example | Colab | Author | -|:--------------------------------------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-----------------------------------------------------------:| -| CLIP Guided Stable Diffusion | Doing CLIP guidance for text to image generation with Stable Diffusion | [CLIP Guided Stable Diffusion](#clip-guided-stable-diffusion) | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/CLIP_Guided_Stable_diffusion_with_diffusers.ipynb) | [Suraj Patil](https://github.com/patil-suraj/) | -| One Step U-Net (Dummy) | Example showcasing of how to use Community Pipelines (see https://github.com/huggingface/diffusers/issues/841) | [One Step U-Net](#one-step-unet) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) | -| Stable Diffusion Interpolation | Interpolate the latent space of Stable Diffusion between different prompts/seeds | [Stable Diffusion Interpolation](#stable-diffusion-interpolation) | - | [Nate Raw](https://github.com/nateraw/) | -| Stable Diffusion Mega | **One** Stable Diffusion Pipeline with all functionalities of [Text2Image](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py), [Image2Image](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py) and [Inpainting](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py) | [Stable Diffusion Mega](#stable-diffusion-mega) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) | -| Long Prompt Weighting Stable Diffusion | **One** Stable Diffusion Pipeline without tokens length limit, and support parsing weighting in prompt. | [Long Prompt Weighting Stable Diffusion](#long-prompt-weighting-stable-diffusion) | - | [SkyTNT](https://github.com/SkyTNT) | -| Speech to Image | Using automatic-speech-recognition to transcribe text and Stable Diffusion to generate images | [Speech to Image](#speech-to-image) | - | [Mikail Duzenli](https://github.com/MikailINTech) -| Wild Card Stable Diffusion | Stable Diffusion Pipeline that supports prompts that contain wildcard terms (indicated by surrounding double underscores), with values instantiated randomly from a corresponding txt file or a dictionary of possible values | [Wildcard Stable Diffusion](#wildcard-stable-diffusion) | - | [Shyam Sudhakaran](https://github.com/shyamsn97) | +| Example | Description | Code Example | Colab | Author | +|:---------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-----------------------------------------------------------:| +| CLIP Guided Stable Diffusion | Doing CLIP guidance for text to image generation with Stable Diffusion | [CLIP Guided Stable Diffusion](#clip-guided-stable-diffusion) | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/CLIP_Guided_Stable_diffusion_with_diffusers.ipynb) | [Suraj Patil](https://github.com/patil-suraj/) | +| One Step U-Net (Dummy) | Example showcasing of how to use Community Pipelines (see https://github.com/huggingface/diffusers/issues/841) | [One Step U-Net](#one-step-unet) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) | +| Stable Diffusion Interpolation | Interpolate the latent space of Stable Diffusion between different prompts/seeds | [Stable Diffusion Interpolation](#stable-diffusion-interpolation) | - | [Nate Raw](https://github.com/nateraw/) | +| Stable Diffusion Mega | **One** Stable Diffusion Pipeline with all functionalities of [Text2Image](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py), [Image2Image](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py) and [Inpainting](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py) | [Stable Diffusion Mega](#stable-diffusion-mega) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) | +| Long Prompt Weighting Stable Diffusion | **One** Stable Diffusion Pipeline without tokens length limit, and support parsing weighting in prompt. | [Long Prompt Weighting Stable Diffusion](#long-prompt-weighting-stable-diffusion) | - | [SkyTNT](https://github.com/SkyTNT) | +| Speech to Image | Using automatic-speech-recognition to transcribe text and Stable Diffusion to generate images | [Speech to Image](#speech-to-image) | - | [Mikail Duzenli](https://github.com/MikailINTech) +| Wild Card Stable Diffusion | Stable Diffusion Pipeline that supports prompts that contain wildcard terms (indicated by surrounding double underscores), with values instantiated randomly from a corresponding txt file or a dictionary of possible values | [Wildcard Stable Diffusion](#wildcard-stable-diffusion) | - | [Shyam Sudhakaran](https://github.com/shyamsn97) | | [Composable Stable Diffusion](https://energy-based-model.github.io/Compositional-Visual-Generation-with-Composable-Diffusion-Models/) | Stable Diffusion Pipeline that supports prompts that contain "|" in prompts (as an AND condition) and weights (separated by "|" as well) to positively / negatively weight prompts. | [Composable Stable Diffusion](#composable-stable-diffusion) | - | [Mark Rich](https://github.com/MarkRich) | -| Seed Resizing Stable Diffusion | Stable Diffusion Pipeline that supports resizing an image and retaining the concepts of the 512 by 512 generation. | [Seed Resizing](#seed-resizing) | - | [Mark Rich](https://github.com/MarkRich) | -| Imagic Stable Diffusion | Stable Diffusion Pipeline that enables writing a text prompt to edit an existing image| [Imagic Stable Diffusion](#imagic-stable-diffusion) | - | [Mark Rich](https://github.com/MarkRich) | -| Multilingual Stable Diffusion | Stable Diffusion Pipeline that supports prompts in 50 different languages. | [Multilingual Stable Diffusion](#multilingual-stable-diffusion-pipeline) | - | [Juan Carlos Piñeros](https://github.com/juancopi81) | -| Image to Image Inpainting Stable Diffusion | Stable Diffusion Pipeline that enables the overlaying of two images and subsequent inpainting| [Image to Image Inpainting Stable Diffusion](#image-to-image-inpainting-stable-diffusion) | - | [Alex McKinney](https://github.com/vvvm23) | -| Text Based Inpainting Stable Diffusion | Stable Diffusion Inpainting Pipeline that enables passing a text prompt to generate the mask for inpainting| [Text Based Inpainting Stable Diffusion](#image-to-image-inpainting-stable-diffusion) | - | [Dhruv Karan](https://github.com/unography) | -| Bit Diffusion | Diffusion on discrete data | [Bit Diffusion](#bit-diffusion) | - |[Stuti R.](https://github.com/kingstut) | -| K-Diffusion Stable Diffusion | Run Stable Diffusion with any of [K-Diffusion's samplers](https://github.com/crowsonkb/k-diffusion/blob/master/k_diffusion/sampling.py) | [Stable Diffusion with K Diffusion](#stable-diffusion-with-k-diffusion) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) | -| Checkpoint Merger Pipeline | Diffusion Pipeline that enables merging of saved model checkpoints | [Checkpoint Merger Pipeline](#checkpoint-merger-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | - Stable Diffusion v1.1-1.4 Comparison | Run all 4 model checkpoints for Stable Diffusion and compare their results together | [Stable Diffusion Comparison](#stable-diffusion-comparisons) | - | [Suvaditya Mukherjee](https://github.com/suvadityamuk) | - MagicMix | Diffusion Pipeline for semantic mixing of an image and a text prompt | [MagicMix](#magic-mix) | - | [Partho Das](https://github.com/daspartho) | -| Stable UnCLIP | Diffusion Pipeline for combining prior model (generate clip image embedding from text, UnCLIPPipeline `"kakaobrain/karlo-v1-alpha"`) and decoder pipeline (decode clip image embedding to image, StableDiffusionImageVariationPipeline `"lambdalabs/sd-image-variations-diffusers"` ). | [Stable UnCLIP](#stable-unclip) | - |[Ray Wang](https://wrong.wang) | -| UnCLIP Text Interpolation Pipeline | Diffusion Pipeline that allows passing two prompts and produces images while interpolating between the text-embeddings of the two prompts | [UnCLIP Text Interpolation Pipeline](#unclip-text-interpolation-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | -| UnCLIP Image Interpolation Pipeline | Diffusion Pipeline that allows passing two images/image_embeddings and produces images while interpolating between their image-embeddings | [UnCLIP Image Interpolation Pipeline](#unclip-image-interpolation-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | -| DDIM Noise Comparative Analysis Pipeline | Investigating how the diffusion models learn visual concepts from each noise level (which is a contribution of [P2 weighting (CVPR 2022)](https://arxiv.org/abs/2204.00227)) | [DDIM Noise Comparative Analysis Pipeline](#ddim-noise-comparative-analysis-pipeline) | - |[Aengus (Duc-Anh)](https://github.com/aengusng8) | -| CLIP Guided Img2Img Stable Diffusion Pipeline | Doing CLIP guidance for image to image generation with Stable Diffusion | [CLIP Guided Stable Diffusion](#clip-guided-img2img-stable-diffusion) | - | [Nipun Jindal](https://github.com/nipunjindal/) | +| Seed Resizing Stable Diffusion| Stable Diffusion Pipeline that supports resizing an image and retaining the concepts of the 512 by 512 generation. | [Seed Resizing](#seed-resizing) | - | [Mark Rich](https://github.com/MarkRich) | +| Imagic Stable Diffusion | Stable Diffusion Pipeline that enables writing a text prompt to edit an existing image| [Imagic Stable Diffusion](#imagic-stable-diffusion) | - | [Mark Rich](https://github.com/MarkRich) | +| Multilingual Stable Diffusion| Stable Diffusion Pipeline that supports prompts in 50 different languages. | [Multilingual Stable Diffusion](#multilingual-stable-diffusion-pipeline) | - | [Juan Carlos Piñeros](https://github.com/juancopi81) | +| Image to Image Inpainting Stable Diffusion | Stable Diffusion Pipeline that enables the overlaying of two images and subsequent inpainting| [Image to Image Inpainting Stable Diffusion](#image-to-image-inpainting-stable-diffusion) | - | [Alex McKinney](https://github.com/vvvm23) | +| Text Based Inpainting Stable Diffusion | Stable Diffusion Inpainting Pipeline that enables passing a text prompt to generate the mask for inpainting| [Text Based Inpainting Stable Diffusion](#image-to-image-inpainting-stable-diffusion) | - | [Dhruv Karan](https://github.com/unography) | +| Bit Diffusion | Diffusion on discrete data | [Bit Diffusion](#bit-diffusion) | - |[Stuti R.](https://github.com/kingstut) | +| K-Diffusion Stable Diffusion | Run Stable Diffusion with any of [K-Diffusion's samplers](https://github.com/crowsonkb/k-diffusion/blob/master/k_diffusion/sampling.py) | [Stable Diffusion with K Diffusion](#stable-diffusion-with-k-diffusion) | - | [Patrick von Platen](https://github.com/patrickvonplaten/) | +| Checkpoint Merger Pipeline | Diffusion Pipeline that enables merging of saved model checkpoints | [Checkpoint Merger Pipeline](#checkpoint-merger-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | +Stable Diffusion v1.1-1.4 Comparison | Run all 4 model checkpoints for Stable Diffusion and compare their results together | [Stable Diffusion Comparison](#stable-diffusion-comparisons) | - | [Suvaditya Mukherjee](https://github.com/suvadityamuk) | +MagicMix | Diffusion Pipeline for semantic mixing of an image and a text prompt | [MagicMix](#magic-mix) | - | [Partho Das](https://github.com/daspartho) | +| Stable UnCLIP | Diffusion Pipeline for combining prior model (generate clip image embedding from text, UnCLIPPipeline `"kakaobrain/karlo-v1-alpha"`) and decoder pipeline (decode clip image embedding to image, StableDiffusionImageVariationPipeline `"lambdalabs/sd-image-variations-diffusers"` ). | [Stable UnCLIP](#stable-unclip) | - |[Ray Wang](https://wrong.wang) | +| UnCLIP Text Interpolation Pipeline | Diffusion Pipeline that allows passing two prompts and produces images while interpolating between the text-embeddings of the two prompts | [UnCLIP Text Interpolation Pipeline](#unclip-text-interpolation-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | +| UnCLIP Image Interpolation Pipeline | Diffusion Pipeline that allows passing two images/image_embeddings and produces images while interpolating between their image-embeddings | [UnCLIP Image Interpolation Pipeline](#unclip-image-interpolation-pipeline) | - | [Naga Sai Abhinay Devarinti](https://github.com/Abhinay1997/) | +| DDIM Noise Comparative Analysis Pipeline | Investigating how the diffusion models learn visual concepts from each noise level (which is a contribution of [P2 weighting (CVPR 2022)](https://arxiv.org/abs/2204.00227)) | [DDIM Noise Comparative Analysis Pipeline](#ddim-noise-comparative-analysis-pipeline) | - |[Aengus (Duc-Anh)](https://github.com/aengusng8) | +| CLIP Guided Img2Img Stable Diffusion Pipeline | Doing CLIP guidance for image to image generation with Stable Diffusion | [CLIP Guided Img2Img Stable Diffusion](#clip-guided-img2img-stable-diffusion) | - | [Nipun Jindal](https://github.com/nipunjindal/) | @@ -1085,21 +1085,17 @@ The following code requires roughly 12GB of GPU RAM. ```python from io import BytesIO - import requests import torch from diffusers import DiffusionPipeline from PIL import Image from transformers import CLIPFeatureExtractor, CLIPModel - feature_extractor = CLIPFeatureExtractor.from_pretrained( "laion/CLIP-ViT-B-32-laion2B-s34B-b79K" ) clip_model = CLIPModel.from_pretrained( "laion/CLIP-ViT-B-32-laion2B-s34B-b79K", torch_dtype=torch.float16 ) - - guided_pipeline = DiffusionPipeline.from_pretrained( "CompVis/stable-diffusion-v1-4", # custom_pipeline="clip_guided_stable_diffusion", @@ -1110,14 +1106,10 @@ guided_pipeline = DiffusionPipeline.from_pretrained( ) guided_pipeline.enable_attention_slicing() guided_pipeline = guided_pipeline.to("cuda") - prompt = "fantasy book cover, full moon, fantasy forest landscape, golden vector elements, fantasy magic, dark light night, intricate, elegant, sharp focus, illustration, highly detailed, digital painting, concept art, matte, art by WLOP and Artgerm and Albert Bierstadt, masterpiece" - url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg" - response = requests.get(url) init_image = Image.open(BytesIO(response.content)).convert("RGB") - image = guided_pipeline( prompt=prompt, num_inference_steps=30, @@ -1137,4 +1129,4 @@ Init Image Output Image -![img2img_clip_guidance](https://huggingface.co/datasets/njindal/images/resolve/main/clip_guided_img2img.jpg) \ No newline at end of file +![img2img_clip_guidance](https://huggingface.co/datasets/njindal/images/resolve/main/clip_guided_img2img.jpg)