uuu

app.py

@@ -27,10 +27,26 @@ token = os.getenv("HF_TOKEN")
 login(token=token)
 
 # Model and Pipeline Setup
+
 model_path = 'stabilityai/stable-diffusion-3.5-large'
 ip_adapter_path = './ip-adapter.bin'
 image_encoder_path = "google/siglip-so400m-patch14-384"
 
+transformer = SD3Transformer2DModel.from_pretrained(
+    model_path, subfolder="transformer", torch_dtype=torch.bfloat16
+)
+
+pipe = StableDiffusion3Pipeline.from_pretrained(
+    model_path, transformer=transformer, torch_dtype=torch.bfloat16
+).to("cuda")
+
+pipe.init_ipadapter(
+    ip_adapter_path=ip_adapter_path,
+    image_encoder_path=image_encoder_path,
+    nb_token=64,
+)
+
+
 # Load transformer and pipeline
 transformer = SD3Transformer2DModel.from_pretrained(
     model_path, subfolder="transformer", torch_dtype=torch.bfloat16
@@ -48,22 +64,23 @@ pipe.init_ipadapter(
 
 
 @spaces.GPU
-def gui_generation(text, num_imgs, width, height):
+def gui_generation(prompt, ref_img):
     """
     Generate images using Stable Diffusion 3.5
     """
-    images = pipe(
-        prompt=text,
-        width=width,
-        height=height,
-        num_images_per_prompt=num_imgs,
+    image = pipe(
+        width=1024,
+        height=1024,
+        prompt=prompt,
         negative_prompt="lowres, low quality, worst quality",
         num_inference_steps=24,
         guidance_scale=5.0,
         generator=torch.Generator("cuda").manual_seed(42),
-    ).images
+        clip_image=ref_img,
+        ipadapter_scale=0.5,
+    ).images[0]
 
-    return images
+    return image
 
 
 # Create Gradio interface
@@ -72,19 +89,16 @@ with gr.Blocks() as demo:
 
     with gr.Row():
         prompt_box = gr.Textbox(label="Prompt", placeholder="Enter your image generation prompt")
-        number_slider = gr.Slider(1, 30, value=2, step=1, label="Batch size")
 
     with gr.Row():
-        width_slider = gr.Slider(256, 1536, value=1024, step=64, label="Width")
-        height_slider = gr.Slider(256, 1536, value=1024, step=64, label="Height")
-
-    gallery = gr.Gallery(columns=[3], rows=[1], object_fit="contain", height="auto")
+        ref_img = gr.Image(type="pil", label="Upload Reference Image")
+    gallery = gr.Image(type="pil", label="Generated Image")
 
     generate_btn = gr.Button("Generate")
 
     generate_btn.click(
         fn=gui_generation,
-        inputs=[prompt_box, number_slider, width_slider, height_slider],
+        inputs=[prompt_box, ref_img],
         outputs=gallery
     )
 demo.launch()