app.py

import spaces
import gradio as gr
import numpy as np
import random
import torch
from PIL import Image
from torchvision import transforms
from diffusers import DiffusionPipeline

# Constants
dtype = torch.bfloat16
device = "cuda" if torch.cuda.is_available() else "cpu"
MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 2048

# Load FLUX model
pipe = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1-schnell", torch_dtype=dtype).to(device)
pipe.enable_model_cpu_offload()
pipe.vae.enable_slicing()
pipe.vae.enable_tiling()

def preprocess_image(image, image_size):
    preprocess = transforms.Compose([
        transforms.Resize((image_size, image_size), interpolation=transforms.InterpolationMode.LANCZOS),
        transforms.ToTensor(),
        transforms.Normalize([0.5], [0.5])
    ])
    image = preprocess(image).unsqueeze(0).to(device, dtype=dtype)
    return image

def check_shapes(latents):
    # Get the shape of the latents
    latent_shape = latents.shape
    print(f"Latent shape: {latent_shape}")

    # Get the expected shape for the transformer input
    expected_shape = (1, latent_shape[1] * latent_shape[2] * latent_shape[3])
    print(f"Expected transformer input shape: {expected_shape}")

    # Try to get the shape of the transformer's weight matrix
    try:
        # Assuming the first layer of the transformer has a linear projection
        if hasattr(pipe.transformer, 'blocks'):
            weight_shape = pipe.transformer.blocks[0].attn.to_q.weight.shape
        else:
            print("Unable to determine transformer weight shape.")
            return
        print(f"Transformer weight shape: {weight_shape}")

        # Check if the shapes are compatible for matrix multiplication
        if expected_shape[1] == weight_shape[1]:
            print("Shapes are compatible for matrix multiplication.")
        else:
            print("Warning: Shapes are not compatible for matrix multiplication.")
            print(f"Expected: {expected_shape[1]}, Got: {weight_shape[1]}")
    except AttributeError as e:
        print(f"Unable to access transformer weights: {e}")

@spaces.GPU()
def infer(prompt, init_image=None, seed=42, randomize_seed=False, width=1024, height=1024, num_inference_steps=4, progress=gr.Progress(track_tqdm=True)):
    if randomize_seed:
        seed = random.randint(0, MAX_SEED)
    generator = torch.Generator(device=device).manual_seed(seed)

    try:
        if init_image is None:
            # text2img case
            image = pipe(
                prompt=prompt,
                height=height,
                width=width,
                num_inference_steps=num_inference_steps,
                generator=generator,
                guidance_scale=0.0
            ).images[0]
        else:
            # img2img case
            init_image = init_image.convert("RGB")
            init_image = preprocess_image(init_image, 1024)  # Using 1024 as FLUX VAE sample size
            
            # Encode the image using FLUX VAE
            latents = pipe.vae.encode(init_image).latent_dist.sample() * 0.18215
            
            # Ensure latents are the correct shape
            latents = torch.nn.functional.interpolate(latents, size=(height // 8, width // 8), mode='bilinear')
            
            # Check shapes before reshaping
            check_shapes(latents)
            
            # Reshape latents to match the expected input shape of the transformer
            latents = latents.reshape(1, -1)

            # Check shapes after reshaping
            check_shapes(latents)

            image = pipe(
                prompt=prompt,
                height=height,
                width=width,
                num_inference_steps=num_inference_steps,
                generator=generator,
                guidance_scale=0.0,
                latents=latents
            ).images[0]

        return image, seed
    except Exception as e:
        print(f"Error during inference: {e}")
        return Image.new("RGB", (width, height), (255, 0, 0)), seed  # Red fallback image

# Gradio interface setup
with gr.Blocks() as demo:
    with gr.Row():
        prompt = gr.Textbox(label="Prompt")
        init_image = gr.Image(label="Initial Image (optional)", type="pil")

    with gr.Row():
        generate = gr.Button("Generate")
        
    with gr.Row():
        result = gr.Image(label="Result")
        seed_output = gr.Number(label="Seed")

    with gr.Accordion("Advanced Settings", open=False):
        seed = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=42)
        randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
        width = gr.Slider(label="Width", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=1024)
        height = gr.Slider(label="Height", minimum=256, maximum=MAX_IMAGE_SIZE, step=32, value=1024)
        num_inference_steps = gr.Slider(label="Number of inference steps", minimum=1, maximum=50, step=1, value=4)

    generate.click(
        infer,
        inputs=[prompt, init_image, seed, randomize_seed, width, height, num_inference_steps],
        outputs=[result, seed_output]
    )

demo.launch()