[sd3] make sure height and size are divisible by 16

[yiyixuxu]

for 1920 X 1080 should get an input error and ask user to resize to 1920 x 1072 (on current main will get an error on scheduler step)

ValueError: `height` and `width` have to be divisible by 16 but are 1920 and 1080.You can use height 1920 and width 1072.

import torch
import argparse
from diffusers import StableDiffusion3Pipeline

def main(height, width):
    device = torch.device("cuda:1")

    pipe = StableDiffusion3Pipeline.from_pretrained("stabilityai/stable-diffusion-3-medium-diffusers", torch_dtype=torch.float16)
    pipe = pipe.to(device)
    generator = torch.Generator(device=device).manual_seed(0)

    image = pipe(
        "A cat holding a sign that says hello world",
        negative_prompt="",
        num_inference_steps=28,
        height=height,
        width=width,
        guidance_scale=7.0,
        generator=generator,
    ).images[0]

    print(f"{height}x{width} -> image: {image.size}")
    image.save(f"yiyi_test_1_out_{height}_{width}.png")

if __name__ == "__main__":
    parser = argparse.ArgumentParser(description="Generate an image with Stable Diffusion 3")
    parser.add_argument("--height", type=int, default=1920, help="Height of the output image")
    parser.add_argument("--width", type=int, default=1080, help="Width of the output image")
    
    args = parser.parse_args()
    
    main(args.height, args.width)

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[asomoza]

thanks, LGTM