app.py

import torch
import spaces
from diffusers import StableDiffusionPipeline, DDIMScheduler, AutoencoderKL
from transformers import AutoFeatureExtractor
from ip_adapter.pipeline_stable_diffusion_extra_cfg import StableDiffusionPipelineCFG

from ip_adapter.ip_adapter_instruct import IPAdapterInstruct
from huggingface_hub import hf_hub_download
import gradio as gr
import cv2

base_model_path = "SG161222/Realistic_Vision_V4.0_noVAE"
vae_model_path = "stabilityai/sd-vae-ft-mse"
image_encoder_path = "laion/CLIP-ViT-H-14-laion2B-s32B-b79K"
ip_ckpt = hf_hub_download(repo_id="CiaraRowles/IP-Adapter-Instruct", filename="ip-adapter-instruct-sd15.bin", repo_type="model")

#safety_model_id = "CompVis/stable-diffusion-safety-checker"
#safety_feature_extractor = AutoFeatureExtractor.from_pretrained(safety_model_id)
#safety_checker = StableDiffusionSafetyChecker.from_pretrained(safety_model_id)
#the model should auto set this

device = "cuda"

noise_scheduler = DDIMScheduler(
    num_train_timesteps=1000,
    beta_start=0.00085,
    beta_end=0.012,
    beta_schedule="scaled_linear",
    clip_sample=False,
    set_alpha_to_one=False,
    steps_offset=1,
)
vae = AutoencoderKL.from_pretrained(vae_model_path).to(dtype=torch.float16)
pipe = StableDiffusionPipelineCFG.from_pretrained(
    base_model_path,
    scheduler=noise_scheduler,
    vae=vae,
    torch_dtype=torch.float16,
#    feature_extractor=safety_feature_extractor,
).to(device)

#pipe.load_lora_weights("h94/IP-Adapter-FaceID", weight_name="ip-adapter-faceid-plusv2_sd15_lora.safetensors")
#pipe.fuse_lora()

ip_model = IPAdapterInstruct(pipe, ip_ckpt, device)

cv2.setNumThreads(1)

@spaces.GPU(enable_queue=True)
def generate_image(image, prompt, negative_prompt, preserve_face_structure, face_strength, scale, nfaa_negative_prompt, progress=gr.Progress(track_tqdm=True)):
    faceid_all_embeds = []
    first_iteration = True

            
    #average_embedding = torch.mean(torch.stack(faceid_all_embeds, dim=0), dim=0)
    
    total_negative_prompt = f"{negative_prompt} {nfaa_negative_prompt}"
    
    print("Generating normal")
    image = ip_model.generate(
        prompt=prompt, negative_prompt=total_negative_prompt, pil_image=image,
        scale=scale, width=512, height=512, num_inference_steps=30
    )

    print(image)
    return image

def change_style(style):
    if style == "Photorealistic":
        return(gr.update(value=True), gr.update(value=1.3), gr.update(value=1.0))
    else:
        return(gr.update(value=True), gr.update(value=0.1), gr.update(value=0.8))

def swap_to_gallery(images):
    return gr.update(value=images, visible=True), gr.update(visible=True), gr.update(visible=False)

def remove_back_to_files():
    return gr.update(visible=False), gr.update(visible=False), gr.update(visible=True)
css = '''
h1{margin-bottom: 0 !important}
'''
with gr.Blocks(css=css) as demo:
    gr.Markdown("# IP-Adapter-Instruct demo")
    gr.Markdown("Demo for the [CiaraRowles/IP-Adapter-Instruct model](https://huggingface.co/CiaraRowles/IP-Adapter-Instruct)")
    with gr.Row():
        with gr.Column():
            files = gr.Files(
                        label="Drag 1 input image",
                        file_types="image"
                    )
            uploaded_files = gr.Gallery(label="Your image", visible=False, columns=5, rows=1, height=125)
            with gr.Column(visible=False) as clear_button:
                remove_and_reupload = gr.ClearButton(value="Remove and upload new ones", components=files, size="sm")
            prompt = gr.Textbox(label="Prompt",
                       info="Try something like 'a photo of a man/woman/person'",
                       placeholder="A photo of a [man/woman/person]...")
            negative_prompt = gr.Textbox(label="Negative Prompt", placeholder="low quality")
            style = gr.Radio(label="Generation type", info="For stylized try prompts like 'a watercolor painting of a woman'", choices=["Photorealistic", "Stylized"], value="Photorealistic")
            submit = gr.Button("Submit")
            with gr.Accordion(open=False, label="Advanced Options"):
                preserve = gr.Checkbox(label="Preserve Face Structure", info="Higher quality, less versatility (the face structure of your first photo will be preserved). Unchecking this will use the v1 model.", value=True)
                face_strength = gr.Slider(label="Face Structure strength", info="Only applied if preserve face structure is checked", value=1.3, step=0.1, minimum=0, maximum=3)
                scale = gr.Slider(label="Scale", value=1.0, step=0.1, minimum=0, maximum=5)
                nfaa_negative_prompts = gr.Textbox(label="Appended Negative Prompts", info="Negative prompts to steer generations towards safe for all audiences outputs", value="naked, bikini, skimpy, scanty, bare skin, lingerie, swimsuit, exposed, see-through")    
        with gr.Column():
            gallery = gr.Gallery(label="Generated Images")
        style.change(fn=change_style,
                    inputs=style,
                    outputs=[preserve, face_strength, scale])
        files.upload(fn=swap_to_gallery, inputs=files, outputs=[uploaded_files, clear_button, files])
        remove_and_reupload.click(fn=remove_back_to_files, outputs=[uploaded_files, clear_button, files])
        submit.click(fn=generate_image,
                    inputs=[files,prompt,negative_prompt,preserve, face_strength, scale, nfaa_negative_prompts],
                    outputs=gallery)
    
    gr.Markdown("This demo includes extra features to mitigate the implicit bias of the model and prevent explicit usage of it to generate content with faces of people, including third parties, that is not safe for all audiences, including naked or semi-naked people.")
    
demo.launch()