Upload app.py

app.py

@@ -0,0 +1,47 @@
+import gradio as gr
+from transformers import pipeline
+from diffusers import StableDiffusion3Pipeline 
+from diffusers import DiffusionPipeline
+import torch
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+model_repo_id = "stabilityai/stable-diffusion-3.5-medium"
+image_style = "pixel art"
+torch_dtype = torch.float32
+
+if torch.cuda.is_available():
+    torch_dtype = torch.bfloat16
+
+def generate_description(image):
+    model = pipeline("image-to-text", model="nlpconnect/vit-gpt2-image-captioning")
+    return model(image)[0]['generated_text']
+
+def generate_image_by_description(description):
+    pipe = DiffusionPipeline.from_pretrained(model_repo_id, torch_dtype=torch_dtype)
+    pipe = pipe.to(device)
+    prompt = (
+        f"Generate a high-quality, detailed image of a {image_style} of a pigeon. "
+        f"The description of the pigeon is: {description}. "
+        "Make it visually appealing with clear textures and distinct colors."
+    )
+    image = pipe(
+        prompt,
+        num_inference_steps=40,
+        guidance_scale=4.5,
+    ).images[0]
+    return image
+
+with gr.Blocks() as demo:
+    selected_image = gr.Image(type="filepath", label="Upload an Image of the Pigeon")
+    generate_button = gr.Button("Generate Avatar", variant="primary")
+    generated_image = gr.Image(type="numpy", label="Generated Avatar")
+
+    # Function chaining: generate description, then generate image without displaying text
+    def process_and_generate(image):
+        description = generate_description(image)
+        return generate_image_by_description(description)
+
+    generate_button.click(process_and_generate, inputs=selected_image, outputs=generated_image)
+
+
+demo.launch()