Upload 5 files

Dockerfile

@@ -0,0 +1,30 @@
+FROM python:3.10-slim
+
+# System dependencies
+RUN apt-get update && apt-get install -y git && rm -rf /var/lib/apt/lists/*
+
+WORKDIR /workspace
+
+# Install Python dependencies
+COPY requirements.txt .
+RUN pip install --upgrade pip && pip install --no-cache-dir -r requirements.txt
+
+# Download models and adapters at build time
+# You can replace MODEL_ID or adapter repos as needed
+
+# -- Download Stable Diffusion v1-5 weights --
+RUN python -c "from diffusers import StableDiffusionImg2ImgPipeline; StableDiffusionImg2ImgPipeline.from_pretrained('runwayml/stable-diffusion-v1-5', cache_dir='./models', torch_dtype='float32')"
+
+# -- Download IP-Adapter weights (official Hugging Face repo) --
+RUN mkdir -p ./models/ip_adapter && \
+    wget -O ./models/ip_adapter/ip-adapter_sd15.bin https://huggingface.co/h94/IP-Adapter/resolve/main/models/ip-adapter_sd15.bin
+
+# Copy app code
+COPY . .
+
+# Set environment (disable gradio analytics, useful for Spaces)
+ENV GRADIO_ANALYTICS_ENABLED="False"
+
+EXPOSE 7860
+
+CMD ["python", "app.py"]

README.md

@@ -1,10 +1,26 @@
----
-title: Sticker Diffusion
-emoji: 👁
-colorFrom: yellow
-colorTo: pink
-sdk: docker
-pinned: false
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# AI Sticker Generator (Stable Diffusion + IP-Adapter/InstantID/ControlNet, CPU-ready)
+
+## 🚀 Features
+- Generate emoji/sticker-style faces from your own photos
+- Uses Stable Diffusion + one adapter (IP-Adapter, InstantID, or ControlNet)
+- Runs on CPU
+
+## 🖥️ Quickstart: Hugging Face Spaces
+
+1. **Create a new Space:**  
+   Go to [https://huggingface.co/spaces](https://huggingface.co/spaces), click **"Create new Space"**.  
+   Select **"Gradio"** as the SDK, choose **CPU** for hardware.
+
+2. **Upload these files.**
+
+3. **Add model weights:**
+   - For IP-Adapter: Download weights from the official [IP-Adapter repo](https://github.com/tencent-ailab/IP-Adapter) or Hugging Face model hub.
+   - Place them in the `models/` directory.
+   - Alternatively, adjust the code to download them at runtime.
+
+4. **Requirements are auto-installed from `requirements.txt` on Hugging Face.**
+
+5. **Hit "Run".**
+
+## 🛠️ Local testing
+- Create a virtual environment and install dependencies:

app.py

@@ -0,0 +1,21 @@
+import gradio as gr
+from utils import generate_sticker
+
+def predict(image, prompt):
+    result_img = generate_sticker(image, prompt)
+    return result_img
+
+with gr.Blocks() as demo:
+    gr.Markdown("# 🦄 AI Sticker Generator (Stable Diffusion + IP-Adapter)")
+    with gr.Row():
+        image_input = gr.Image(type="pil", label="Upload your photo")
+        prompt_input = gr.Textbox(
+            label="Prompt (style or mood for emoji)",
+            value="cartoon emoji, white outline, clean background",
+        )
+    output_image = gr.Image(label="Sticker Output")
+    run_btn = gr.Button("Generate Sticker")
+    run_btn.click(predict, inputs=[image_input, prompt_input], outputs=output_image)
+
+if __name__ == "__main__":
+    demo.launch()

requirements.txt

@@ -0,0 +1,6 @@
+# Only install what's needed for Stable Diffusion, IP-Adapter, and CPU inference
+diffusers==0.28.0
+transformers
+torch  # Version will depend on your platform
+accelerate
+Pillow

utils.py

@@ -0,0 +1,36 @@
+import torch
+from diffusers import StableDiffusionImg2ImgPipeline
+from PIL import Image
+
+# --- Place any download or path setup here ---
+MODEL_ID = "runwayml/stable-diffusion-v1-5"  # Can swap for custom path if using IP-Adapter
+DEVICE = "cpu"
+MODEL_CACHE = "./models"
+
+# (Optional) Download IP-Adapter weights and patch pipeline if desired
+
+def generate_sticker(input_image, prompt):
+    """
+    Given a user image and a prompt, generates a sticker/emoji-style portrait.
+    """
+    # Load the model (download if not present)
+    pipe = StableDiffusionImg2ImgPipeline.from_pretrained(
+        MODEL_ID,
+        torch_dtype=torch.float32,
+        cache_dir=MODEL_CACHE,
+        safety_checker=None,  # Disable for demo/testing
+    ).to(DEVICE)
+
+    # Preprocess the image (resize, etc)
+    init_image = input_image.convert("RGB").resize((512, 512))
+
+    # Run inference (low strength for identity preservation)
+    result = pipe(
+        prompt=prompt,
+        image=init_image,
+        strength=0.65,
+        guidance_scale=7.5,
+        num_inference_steps=30
+    )
+    # Return the generated image (as PIL)
+    return result.images[0]