app.py

import torch
import torch.nn.functional as F

from diffusers import AutoencoderKL, LMSDiscreteScheduler, UNet2DConditionModel

# For video display:
from PIL import Image
from torchvision import transforms as tfms
from tqdm.auto import tqdm
from transformers import CLIPTextModel, CLIPTokenizer, logging, AutoProcessor, CLIPVisionModel

import os, glob
from pathlib import Path
import gradio as gr

torch.manual_seed(1)

def pil_to_latent(input_im):
    # Single image -> single latent in a batch (so size 1, 4, 64, 64)
    with torch.no_grad():
        latent = vae.encode(tfms.ToTensor()(input_im).unsqueeze(0).to(torch_device)*2-1) # Note scaling
    return 0.18215 * latent.latent_dist.sample()

def latents_to_pil(latents):
    # bath of latents -> list of images
    latents = (1 / 0.18215) * latents
    with torch.no_grad():
        image = vae.decode(latents).sample
    image = (image / 2 + 0.5).clamp(0, 1)
    image = image.detach().cpu().permute(0, 2, 3, 1).numpy()
    images = (image * 255).round().astype("uint8")
    pil_images = [Image.fromarray(image) for image in images]
    return pil_images

# Prep Scheduler
def set_timesteps(scheduler, num_inference_steps):
    scheduler.set_timesteps(num_inference_steps)
    scheduler.timesteps = scheduler.timesteps.to(torch.float32) # minor fix to ensure MPS compatibility, fixed in diffusers PR 3925

def get_output_embeds(input_embeddings):
    # CLIP's text model uses causal mask, so we prepare it here:
    bsz, seq_len = input_embeddings.shape[:2]
    causal_attention_mask = text_encoder.text_model._build_causal_attention_mask(bsz, seq_len, dtype=input_embeddings.dtype)

    # Getting the output embeddings involves calling the model with passing output_hidden_states=True
    # so that it doesn't just return the pooled final predictions:
    encoder_outputs = text_encoder.text_model.encoder(
        inputs_embeds=input_embeddings,
        attention_mask=None, # We aren't using an attention mask so that can be None
        causal_attention_mask=causal_attention_mask.to(torch_device),
        output_attentions=None,
        output_hidden_states=True, # We want the output embs not the final output
        return_dict=None,
    )

    # We're interested in the output hidden state only
    output = encoder_outputs[0]

    # There is a final layer norm we need to pass these through
    output = text_encoder.text_model.final_layer_norm(output)

    # And now they're ready!
    return output


# Defined a latent loss that is purely based on one image instead of multiple images
# This is used and working
# Need to look at if I can have the latents of multiple images merged together to give a thought
# Rather than a image - So that the thought is more on the design rather than pure image

def latent_loss(latent, conditioning_image):
    # How far are the image embeds from lossembeds:
    # image = Image.open(conditioning_image)
    r_image = conditioning_image.resize((512,512))
    r_latent = pil_to_latent(r_image)
    error = F.mse_loss(0.5*latent,0.5*r_latent)
    return error
#Generating an image with these modified embeddings


def generate_with_embs(text_input, text_embeddings, conditioning_image):
    height = 512                        # default height of Stable Diffusion
    width = 512                         # default width of Stable Diffusion
    num_inference_steps = 30            # Number of denoising steps
    guidance_scale = 7.5                # Scale for classifier-free guidance
    
    generator = torch.manual_seed(torch.seed())   # Seed generator to create the inital latent noise
    batch_size = 1
    loss_scale = 100 #@param
    imageCondSteps = 5

    max_length = text_input.input_ids.shape[-1]
    uncond_input = tokenizer(
      [""] * batch_size, padding="max_length", max_length=max_length, return_tensors="pt"
    )
    with torch.no_grad():
        uncond_embeddings = text_encoder(uncond_input.input_ids.to(torch_device))[0]
    text_embeddings = torch.cat([uncond_embeddings, text_embeddings])

    # Prep Scheduler
    set_timesteps(scheduler, num_inference_steps)

    # Prep latents
    latents = torch.randn(
    (batch_size, unet.in_channels, height // 8, width // 8),
    generator=generator,
    )
    latents = latents.to(torch_device)
    latents = latents * scheduler.init_noise_sigma

    # Loop
    for i, t in tqdm(enumerate(scheduler.timesteps), total=len(scheduler.timesteps)):
        # expand the latents if we are doing classifier-free guidance to avoid doing two forward passes.
        latent_model_input = torch.cat([latents] * 2)
        sigma = scheduler.sigmas[i]
        latent_model_input = scheduler.scale_model_input(latent_model_input, t)

        # predict the noise residual
        with torch.no_grad():
            noise_pred = unet(latent_model_input, t, encoder_hidden_states=text_embeddings)["sample"]

        # perform guidance
        noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
        noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)

        #### ADDITIONAL GUIDANCE ###
        if conditioning_image:
            if i%imageCondSteps == 0:
                # Requires grad on the latents
                latents = latents.detach().requires_grad_()

                # Get the predicted x0:
                latents_x0 = latents - sigma * noise_pred

                # Decode to image space
                # denoised_images = vae.decode((1 / 0.18215) * latents_x0).sample / 2 + 0.5 # range (0, 1)

                # Calculate loss
                # loss = flag_loss(denoised_images, lossEmbeds) * blue_loss_scale
                loss = latent_loss(latents_x0, conditioning_image) * loss_scale
                # loss = blue_loss(denoised_images) * blue_loss_scale

                print(i, 'loss item:', loss.item())

                # Get gradient
                cond_grad = torch.autograd.grad(loss, latents)[0]
                # print("cond_grad:", cond_grad)

                # Modify the latents based on this gradient
                latents = latents.detach() - cond_grad * sigma**2

        # compute the previous noisy sample x_t -> x_t-1
        latents = scheduler.step(noise_pred, t, latents).prev_sample

    return latents_to_pil(latents)[0]


def getImageWithStyle(prompt, style_name, conditioning_image):
    prompt = prompt + ' in the style of puppy'

    style_loc = "styles/" + style_name + '/learned_embeds.bin'
    style_embed = torch.load(style_loc)
    # print(style_embed)
    # print(style_embed.keys(), style_embed[style_embed.keys()[0]].shape)
    new_style = list(style_embed.keys())[0]

    # Tokenize
    text_input = tokenizer(prompt, padding="max_length", max_length=tokenizer.model_max_length, truncation=True, return_tensors="pt")
    # print("text_input:", text_input)
    input_ids = text_input.input_ids.to(torch_device)
    # print("Input Ids:", input_ids)
    print("Input Ids shape:", input_ids.shape)

    token_emb_layer = text_encoder.text_model.embeddings.token_embedding
    # Get token embeddings
    token_embeddings = token_emb_layer(input_ids)
    print("Token Embeddings shape:", token_embeddings.shape)

    # The new embedding - our special style word
    replacement_token_embedding = style_embed[new_style].to(torch_device)

    # Insert this into the token embeddings
    token_embeddings[0, torch.where(input_ids[0]==6829)] = replacement_token_embedding.to(torch_device)

    pos_emb_layer = text_encoder.text_model.embeddings.position_embedding
    position_ids = text_encoder.text_model.embeddings.position_ids
    print("position_ids shape:", position_ids.shape)
    position_embeddings = pos_emb_layer(position_ids)
    print("Position Embeddings shape:", token_embeddings.shape)

    # Combine with pos embs
    input_embeddings = token_embeddings + position_embeddings

    #  Feed through to get final output embs
    modified_output_embeddings = get_output_embeds(input_embeddings)

    # And generate an image with this:
    image = generate_with_embs(text_input, modified_output_embeddings, conditioning_image)
    return image
    # name = "./Outputs/" + filename+".jpg"
    
    # image.save(name)


# Supress some unnecessary warnings when loading the CLIPTextModel
logging.set_verbosity_error()

# Set device
torch_device = "cuda" if torch.cuda.is_available() else "mps" if torch.backends.mps.is_available() else "cpu"
if "mps" == torch_device: os.environ['PYTORCH_ENABLE_MPS_FALLBACK'] = "1"

# Load the autoencoder model which will be used to decode the latents into image space.
vae = AutoencoderKL.from_pretrained("CompVis/stable-diffusion-v1-4", subfolder="vae")

# Load the tokenizer and text encoder to tokenize and encode the text.
tokenizer = CLIPTokenizer.from_pretrained("openai/clip-vit-large-patch14")
text_encoder = CLIPTextModel.from_pretrained("openai/clip-vit-large-patch14")
image_encoder = CLIPVisionModel.from_pretrained("openai/clip-vit-large-patch14")
image_processor = AutoProcessor.from_pretrained("openai/clip-vit-large-patch14")

# The UNet model for generating the latents.
unet = UNet2DConditionModel.from_pretrained("CompVis/stable-diffusion-v1-4", subfolder="unet")

# The noise scheduler
scheduler = LMSDiscreteScheduler(beta_start=0.00085, beta_end=0.012, beta_schedule="scaled_linear", num_train_timesteps=1000)

# To the GPU we go!
vae = vae.to(torch_device)
text_encoder = text_encoder.to(torch_device)
unet = unet.to(torch_device)
image_encoder = image_encoder.to(torch_device)

# Used puppy as a placeholder here since the token is known
# Will replace with some other word that is better
# prompt = 'A farm'
# style_name = 'birb'
# conditioning_image_folder = './conditioning_images/'
# style_folder = './styles/'

# seedlist = [*range(0, 10000, 500)]
# print(seedlist)

# stylelist = ['birb', ]

# i = 0
# for style_path in glob.glob(os.path.join(style_folder, '*')):
#     seed = seedlist[i]
#     i = i + 1
#     getImageWithStyle(prompt, style_path, None, seed)
#     for conditioning_image in glob.glob(os.path.join(conditioning_image_folder, '*.jpg')):
#         print("style_path:", style_path, "conditioning_image:", conditioning_image)
#         getImageWithStyle(prompt, style_path, conditioning_image, seed)

def generateOutput(prompt, style_name, conditioning_image):
    outputImage = getImageWithStyle(prompt, style_name, conditioning_image)
    return outputImage
    
title = "Stable Diffusion SD Styles along with image conditioning"
description = "Shows the Stable Diffusion usage with SD Styles as well as ways to condition using different loss aspects"
examples = [["A farm", 'midjourney', 'conditioning_images/indianflag.jpg'],["A playground", 'lineart', None]]
style_options = ['birb', 'moebius', 'midjourney', 'cute-game-style', 'depthmap', 'hitokomoru', 'lineart', 'madhubani']
demo = gr.Interface(
    generateOutput, 
    inputs = [
        gr.Textbox(),
        gr.Dropdown(choices=style_options, label="Choose the Style you want"),
        gr.Image(width=256, height=256, label="Image to use for Conditioning", type='pil'),
        ], 
    outputs = [
        gr.Image(width=256, height=256, label="Output"),
        ],
    title = title,
    description = description,
    examples = examples,
    cache_examples=False
)
demo.launch()