Update app.py

app.py

@@ -1,154 +1,90 @@
-import gradio as gr
-import numpy as np
-import random
-
-# import spaces #[uncomment to use ZeroGPU]
-from diffusers import DiffusionPipeline
+import io
 import torch
-
-device = "cuda" if torch.cuda.is_available() else "cpu"
-model_repo_id = "stabilityai/sdxl-turbo"  # Replace to the model you would like to use
-
-if torch.cuda.is_available():
-    torch_dtype = torch.float16
-else:
-    torch_dtype = torch.float32
-
-pipe = DiffusionPipeline.from_pretrained(model_repo_id, torch_dtype=torch_dtype)
-pipe = pipe.to(device)
-
-MAX_SEED = np.iinfo(np.int32).max
-MAX_IMAGE_SIZE = 1024
-
-
-# @spaces.GPU #[uncomment to use ZeroGPU]
-def infer(
-    prompt,
-    negative_prompt,
-    seed,
-    randomize_seed,
-    width,
-    height,
-    guidance_scale,
-    num_inference_steps,
-    progress=gr.Progress(track_tqdm=True),
+from fastapi import FastAPI, UploadFile, File, Form
+from fastapi.responses import StreamingResponse
+from PIL import Image
+from diffusers import StableDiffusionPipeline, DDIMScheduler
+from diffusers.utils import load_image
+from ip_adapter.ip_adapter_faceid import IPAdapterFaceID
+from insightface.app import FaceAnalysis
+
+app = FastAPI(title="IP-Adapter-FaceID API")
+
+# Variables globales para los modelos
+app_insight = None
+ip_model = None
+
+@app.on_event("startup")
+def load_models():
+    global app_insight, ip_model
+    
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+    
+    # 1. Inicializar InsightFace para detectar rostros
+    app_insight = FaceAnalysis(name="buffalo_l", providers=['CUDAExecutionProvider', 'CPUExecutionProvider'])
+    app_insight.prepare(ctx_id=0, det_size=(640, 640))
+    
+    # 2. Inicializar Stable Diffusion v1.5
+    base_model_path = "SG161222/Realistic_Vision_V4.0_noVAE" # O el modelo base que prefieras
+    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,
+    )
+    
+    pipe = StableDiffusionPipeline.from_pretrained(
+        base_model_path,
+        torch_dtype=torch.float16 if device == "cuda" else torch.float32,
+        scheduler=noise_scheduler
+    ).to(device)
+    
+    # 3. Cargar las capas de IP-Adapter FaceID
+    ip_ckpt = "h94/IP-Adapter-FaceID/ip-adapter-faceid_sd15.bin"
+    ip_model = IPAdapterFaceID(pipe, ip_ckpt, device, num_tokens=4)
+
+@app.get("/")
+def read_root():
+    return {"status": "running", "info": "Envía un POST a /predict para generar imágenes"}
+
+@app.post("/predict")
+async def predict(
+    prompt: str = Form(...),
+    negative_prompt: str = Form("monochrome, lowres, bad anatomy, worst quality, low quality"),
+    num_inference_steps: int = Form(30),
+    face_image: UploadFile = File(...)
 ):
-    if randomize_seed:
-        seed = random.randint(0, MAX_SEED)
-
-    generator = torch.Generator().manual_seed(seed)
-
-    image = pipe(
+    # Leer la imagen enviada desde el cliente (Kotlin, etc.)
+    face_bytes = await face_image.read()
+    image = Image.open(io.BytesIO(face_bytes)).convert("RGB")
+    
+    # Extraer el rostro (Embedding) usando InsightFace
+    import numpy as np
+    cv_img = np.array(image)
+    faces = app_insight.get(cv_img)
+    
+    if len(faces) == 0:
+        return {"error": "No se detectó ningún rostro en la imagen provista."}
+        
+    face_id_embed = torch.from_numpy(faces[0].normed_embedding).unsqueeze(0).unsqueeze(0)
+    
+    # Generar la nueva imagen mezclando el prompt + identidad del rostro
+    images = ip_model.generate(
         prompt=prompt,
         negative_prompt=negative_prompt,
-        guidance_scale=guidance_scale,
+        face_id_embed=face_id_embed,
+        num_samples=1,
         num_inference_steps=num_inference_steps,
-        width=width,
-        height=height,
-        generator=generator,
-    ).images[0]
-
-    return image, seed
-
-
-examples = [
-    "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k",
-    "An astronaut riding a green horse",
-    "A delicious ceviche cheesecake slice",
-]
-
-css = """
-#col-container {
-    margin: 0 auto;
-    max-width: 640px;
-}
-"""
-
-with gr.Blocks(css=css) as demo:
-    with gr.Column(elem_id="col-container"):
-        gr.Markdown(" # Text-to-Image Gradio Template")
-
-        with gr.Row():
-            prompt = gr.Text(
-                label="Prompt",
-                show_label=False,
-                max_lines=1,
-                placeholder="Enter your prompt",
-                container=False,
-            )
-
-            run_button = gr.Button("Run", scale=0, variant="primary")
-
-        result = gr.Image(label="Result", show_label=False)
-
-        with gr.Accordion("Advanced Settings", open=False):
-            negative_prompt = gr.Text(
-                label="Negative prompt",
-                max_lines=1,
-                placeholder="Enter a negative prompt",
-                visible=False,
-            )
-
-            seed = gr.Slider(
-                label="Seed",
-                minimum=0,
-                maximum=MAX_SEED,
-                step=1,
-                value=0,
-            )
-
-            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
-
-            with gr.Row():
-                width = gr.Slider(
-                    label="Width",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,  # Replace with defaults that work for your model
-                )
-
-                height = gr.Slider(
-                    label="Height",
-                    minimum=256,
-                    maximum=MAX_IMAGE_SIZE,
-                    step=32,
-                    value=1024,  # Replace with defaults that work for your model
-                )
-
-            with gr.Row():
-                guidance_scale = gr.Slider(
-                    label="Guidance scale",
-                    minimum=0.0,
-                    maximum=10.0,
-                    step=0.1,
-                    value=0.0,  # Replace with defaults that work for your model
-                )
-
-                num_inference_steps = gr.Slider(
-                    label="Number of inference steps",
-                    minimum=1,
-                    maximum=50,
-                    step=1,
-                    value=2,  # Replace with defaults that work for your model
-                )
-
-        gr.Examples(examples=examples, inputs=[prompt])
-    gr.on(
-        triggers=[run_button.click, prompt.submit],
-        fn=infer,
-        inputs=[
-            prompt,
-            negative_prompt,
-            seed,
-            randomize_seed,
-            width,
-            height,
-            guidance_scale,
-            num_inference_steps,
-        ],
-        outputs=[result, seed],
+        seed=42
     )
-
-if __name__ == "__main__":
-    demo.launch()
+    
+    # Convertir la imagen generada a bytes para retornarla por HTTP
+    output_image = images[0]
+    img_byte_arr = io.BytesIO()
+    output_image.save(img_byte_arr, format='PNG')
+    img_byte_arr.seek(0)
+    
+    return StreamingResponse(img_byte_arr, media_type="image/png")