app.py

import gradio as gr
import jax
import numpy as np
import jax.numpy as jnp
from flax.jax_utils import replicate
from flax.training.common_utils import shard
from PIL import Image
from diffusers import FlaxStableDiffusionControlNetPipeline, FlaxControlNetModel
import cv2

# load control net and stable diffusion v1-5
controlnet, controlnet_params = FlaxControlNetModel.from_pretrained(
    "Nahrawy/controlnet-VIDIT-FAID", dtype=jnp.bfloat16, revision="615ba4a457b95a0eba813bcc8caf842c03a4f7bd"
)
pipe, params = FlaxStableDiffusionControlNetPipeline.from_pretrained(
    "runwayml/stable-diffusion-v1-5", controlnet=controlnet, revision="flax", dtype=jnp.bfloat16
)

def create_key(seed=0):
    return jax.random.PRNGKey(seed)

def process_mask(image):
    mask = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    mask = cv2.resize(mask,(512,512))
    return mask



def infer(prompts, negative_prompts, image):
    params["controlnet"] = controlnet_params
    
    num_samples = 1 #jax.device_count()
    rng = create_key(0)
    rng = jax.random.split(rng, jax.device_count())
    im = process_mask(image)
    mask = Image.fromarray(im)
    
    prompt_ids = pipe.prepare_text_inputs([prompts] * num_samples)
    negative_prompt_ids = pipe.prepare_text_inputs([negative_prompts] * num_samples)
    processed_image = pipe.prepare_image_inputs([mask] * num_samples)
    
    p_params = replicate(params)
    prompt_ids = shard(prompt_ids)
    negative_prompt_ids = shard(negative_prompt_ids)
    processed_image = shard(processed_image)
    print(processed_image[0].shape)
    output = pipe(
        prompt_ids=prompt_ids,
        image=processed_image,
        params=p_params,
        prng_seed=rng,
        num_inference_steps=50,
        neg_prompt_ids=negative_prompt_ids,
        jit=True,
    ).images
    
    output_images = pipe.numpy_to_pil(np.asarray(output.reshape((num_samples,) + output.shape[-3:])))
    return output_images

e_images = ['0.png',
            '0.png',
            '0.png',
            '0.png',
            '0.png',
            '2.png',
            '2.png',
            '2.png',
            '2.png',]
e_prompts = ['a dog in the middle of the road, shadow on the ground,light direction north-east',
             'a dog in the middle of the road, shadow on the ground,light direction north-west',
             'a dog in the middle of the road, shadow on the ground,light direction south-west',
             'a dog in the middle of the road, shadow on the ground,light direction south-east',
            'a red rural house, shadow on the ground, light direction north',
            'a red rural house, shadow on the ground, light direction east',
            'a red rural house, shadow on the ground, light direction south',
            'a red rural house, shadow on the ground, light direction west']
e_negative_prompts = ['monochromatic, unrealistic, bad looking, full of glitches',
                     'monochromatic, unrealistic, bad looking, full of glitches',
                     'monochromatic, unrealistic, bad looking, full of glitches',
                     'monochromatic, unrealistic, bad looking, full of glitches',
                     'monochromatic, unrealistic, bad looking, full of glitches',
                     'monochromatic, unrealistic, bad looking, full of glitches',
                     'monochromatic, unrealistic, bad looking, full of glitches',
                     'monochromatic, unrealistic, bad looking, full of glitches']
examples = []
for image, prompt, negative_prompt in zip(e_images, e_prompts, e_negative_prompts):
  examples.append([prompt, negative_prompt, image])
    
title = " # ControlLight: Light control through ControlNet and Depth Maps conditioning"
info = '''
# ControlLight: Light control through ControlNet and Depth Maps conditioning
We propose a ControlNet using depth maps conditioning that is capable of controlling the light direction in a scene while trying to maintain the scene integrity.
The model was trained on [VIDIT dataset](https://huggingface.co/datasets/Nahrawy/VIDIT-Depth-ControlNet) and [
A Dataset of Flash and Ambient Illumination Pairs from the Crowd](https://huggingface.co/datasets/Nahrawy/FAID-Depth-ControlNet) as a part of the [Jax Diffusers Event](https://huggingface.co/jax-diffusers-event).

Due to the limited available data the model is clearly overfit, but it serves as a proof of concept to what can be further achieved using enough data.

A large part of the training data is synthetic so we encourage further training using synthetically generated scenes, using Unreal engine for example.

The WandB training logs can be found [here](https://wandb.ai/hassanelnahrawy/controlnet-VIDIT-FAID), it's worth noting that the model was left to overfit for experimentation and it's advised to use the 8K steps weights or prior weights.

This project is a joint work between [ParityError](https://huggingface.co/ParityError) and [Nahrawy](https://huggingface.co/Nahrawy).

'''
with gr.Blocks() as demo:
    gr.Markdown(title)
    prompts = gr.Textbox(label='prompts')
    negative_prompts = gr.Textbox(label='negative_prompts')
    with gr.Row():
      with gr.Column():
        in_image = gr.Image(label="Depth Map Conditioning")
      with gr.Column():
        out_image = gr.Gallery(label="Generated Image")
    with gr.Row():
        btn = gr.Button("Run")
    with gr.Row():
        gr.Markdown(info)
        gr.Examples(examples=examples,
                   inputs=[prompts,negative_prompts, in_image],
                   outputs=out_image,
                   fn=infer,
                   cache_examples=True)
    btn.click(fn=infer, inputs=[prompts,negative_prompts, in_image] , outputs=out_image)
    
demo.launch()