Document generating multiple images

[kendallroth]

Generating multiple images per prompt is ideal for comparison purposes, but does not seem supported. I am not sure whether this is just some models, or whether all Replicate models only generate a single output. However, it would be beneficial to have a documented approach to generating multiple images, as I am having trouble wading through attempts at such.

I first thought I could create multiple "background" predictions at the same time, but am not sure how to wait until they have all completed processing. Likely would involve the wait() function on all started predictions, but not sure how to go about this.

Is this a process that could be documented if already known? If not, could I get pointed in the right direction (if this is even possible)?

[mattt]

Hi @kendallroth. #193 added support for async_run, which makes it easy to generate a bunch of images in parallel and wait for them all to finish.

import asyncio
import replicate

# https://replicate.com/stability-ai/sdxl
model_version = "stability-ai/sdxl:39ed52f2a78e934b3ba6e2a89f5b1c712de7dfea535525255b1aa35c5565e08b"
prompts = [
    f"A chariot pulled by a team of {count} rainbow unicorns"
    for count in ["two", "four", "six", "eight"]
]

async with asyncio.TaskGroup() as tg:
    tasks = [
        tg.create_task(replicate.async_run(model_version, input={"prompt": prompt}))
        for prompt in prompts
    ]

results = await asyncio.gather(*tasks)
print(results)

[kendallroth]

Thanks! I will admit I am no longer working with replicate anymore, but hope/trust that other will find this useful as well!