feat(vto): Add platform links and improve documentation for VTO workflow

- Add 4 platform launch links (Colab, Colab Enterprise, Vertex AI, GitHub)
- Rename notebook files from LJ_GenMedia_Workflow to VTO_GenMedia_Workflow
- Rename images folder to dress for consistency
- Update all references from 'Dress' to lowercase 'dress' throughout
- Add prerequisites section explaining how to copy dress images to GCS
- Document OUTFITS_PREFIX configuration variable
- Remove emoji icons from documentation
- Streamline README by removing unnecessary sections
- Include 4 sample dress images (blue, multi, pink, red)
- Apply linter fixes for consistent formatting

Change-Id: I26a5d9dfe4208be991054203b96cc57f3084fa13
GitOrigin-RevId: 753408edd5c57ae61a1f201e34de5c772d4239d4

Author: layolin

projects/ai/gen-media/notebooks/vto_scale_workflow/README.md

@@ -20,7 +20,7 @@ applications that:
 
 ## Key Features
 
-### 🎯 Multi-Model AI Pipeline
+### Multi-Model AI Pipeline
 
 - **Model Generation**: Creates photorealistic digital models with diverse
   demographics (race, body type, age)
@@ -32,7 +32,7 @@ applications that:
 - **Scalable Architecture**: Processes batch inputs from CSV with parallel
   execution
 
-### 🔧 Technologies Used
+### Technologies Used
 
 - **Google Vertex AI**: Primary platform for all AI operations
 - **Gemini 2.5 Flash**: Orchestration, prompt generation, and quality critique
@@ -45,7 +45,13 @@ applications that:
 
 ```text
 vto_scale_workflow/
-├── LJ_GenMedia_Workflow.ipynb   # Main Jupyter notebook with complete pipeline
+├── VTO_GenMedia_Workflow.ipynb   # Main Jupyter notebook with complete pipeline
+├── VTO_GenMedia_Workflow.nb.py   # Python script version of the notebook
+├── dress/                       # Sample dress images for VTO
+│   ├── blue-front.png
+│   ├── multi-front.png
+│   ├── pink-front.png
+│   └── red-front.png
 ├── requirements.txt              # Python dependencies
 └── README.md                     # This file
 ```
@@ -104,6 +110,19 @@ The pipeline follows a sequential 5-step process:
     - Local: `gcloud auth application-default login`
     - Vertex AI Workbench: Automatic authentication
 
+### Preparing Input Images
+
+Before running the notebook end-to-end, copy the sample dress images from the
+`dress/` folder to your Google Cloud Storage bucket:
+
+```bash
+# Copy all sample dress images to your GCS bucket
+gsutil cp dress/*.png gs://YOUR_BUCKET_NAME/dress/
+```
+
+The notebook configuration uses `OUTFITS_PREFIX = "dress"` to specify where it
+looks for input dress images.
+
 ### Storage Setup
 
 Google Cloud Storage bucket with structure:
@@ -112,14 +131,16 @@ Google Cloud Storage bucket with structure:
 your-bucket/
 ├── Model_Creation.csv        # Generated model definitions
 ├── models/                   # Generated base model images
-├── Dress/                    # Input garment images
-│   ├── dress1.png
-│   ├── dress2.png
+├── dress/                    # Input garment images (copied from dress/ folder)
+│   ├── blue-front.png
+│   ├── multi-front.png
+│   ├── pink-front.png
+│   ├── red-front.png
 │   └── ...
-├── Dress/4tryon/            # VTO output images
-├── Dress/4tryon/final/      # Selected best VTO images
+├── dress/4tryon/            # VTO output images
+├── dress/4tryon/final/      # Selected best VTO images
 │   └── eval_summary.csv     # Critique results
-└── Dress/4tryon/final_motion/ # Generated videos
+└── dress/4tryon/final_motion/ # Generated videos
 ```
 
 ## Installation
@@ -165,7 +186,7 @@ MODEL_VIDEO = "veo-3.0-generate-001"
 
 1.  **Open the Notebook**
 
-- Launch `LJ_GenMedia_Workflow.ipynb` in Jupyter or Vertex AI Workbench
+- Launch `VTO_GenMedia_Workflow.ipynb` in Jupyter or Vertex AI Workbench
 
 1.  **Run Configuration Cell**
 
@@ -178,98 +199,6 @@ MODEL_VIDEO = "veo-3.0-generate-001"
 
 1.  **Access Results**
 
-- Final VTO images: `gs://your-bucket/Dress/4tryon/final/`
-- Motion videos: `gs://your-bucket/Dress/4tryon/final_motion/`
-- Evaluation summary: `gs://your-bucket/Dress/4tryon/final/eval_summary.csv`
-
-## Performance Considerations
-
-- **Parallel Processing**: Utilizes ThreadPoolExecutor for concurrent operations
-- **Retry Mechanism**: Automatic retry for failed VTO attempts (3 attempts by
-  default)
-- **Batch Processing**: Efficient handling of multiple model-outfit combinations
-- **Resource Management**: Configurable worker limits to control API usage
-
-## Output Examples
-
-### Generated Assets
-
-- **Model Images**: Diverse digital models in standardized outfit
-- **VTO Images**: High-quality garment transfers on each model
-- **Motion Videos**: 8-second runway walk showcasing garments
-- **Evaluation CSV**: Detailed critique results with selection reasoning
-
-### Quality Metrics
-
-The AI critique evaluates:
-
-- Garment transfer completeness
-- Fabric texture preservation
-- Fit accuracy and realism
-- Absence of visual artifacts
-- Body proportion maintenance
-
-## Troubleshooting
-
-### Common Issues
-
-1.  **Authentication Errors**
-
-- Ensure proper GCP authentication
-- Verify project permissions
-
-1.  **API Quotas**
-
-- Monitor Vertex AI quotas
-- Adjust `PARALLEL_JOBS_PER_MODEL` if needed
-
-1.  **Storage Access**
-
-- Verify bucket exists and is accessible
-- Check file paths and prefixes
-
-1.  **Model Availability**
-
-- Confirm model versions are available in your region
-- Update model IDs if using newer versions
-
-## Dependencies
-
-Core requirements (see `requirements.txt`):
-
-- `pandas==2.2.2` - Data manipulation
-- `Pillow==11.1.0` - Image processing
-- `google-genai==1.45.0` - Generative AI SDK
-- `google-cloud-storage==2.19.0` - GCS operations
-- `google-cloud-aiplatform==1.74.0` - Vertex AI integration
-
-## License
-
-This project is for demonstration purposes. Please ensure compliance with Google
-Cloud's terms of service and any applicable licensing requirements for
-production use.
-
-## Contributing
-
-This is a demonstration workflow. For production implementations, consider:
-
-- Error handling enhancements
-- Monitoring and logging integration
-- Cost optimization strategies
-- Custom quality assessment metrics
-- Extended diversity parameters
-
-## Support
-
-For issues related to:
-
-- Google Cloud setup: Consult
-  [Google Cloud Documentation](https://cloud.google.com/docs)
-- Vertex AI models: See
-  [Vertex AI Documentation](https://cloud.google.com/vertex-ai/docs)
-- Code issues: Review the notebook comments and inline documentation
-
-## Acknowledgments
-
-Created on 11/12/2025 using Google's suite of Generative AI models on Vertex AI
-platform.
+- Final VTO images: `gs://your-bucket/dress/4tryon/final/`
+- Motion videos: `gs://your-bucket/dress/4tryon/final_motion/`
+- Evaluation summary: `gs://your-bucket/dress/4tryon/final/eval_summary.csv`

…cale_workflow/LJ_GenMedia_Workflow.ipynb …ale_workflow/VTO_GenMedia_Workflow.ipynbprojects/ai/gen-media/notebooks/vto_scale_workflow/LJ_GenMedia_Workflow.ipynb renamed to projects/ai/gen-media/notebooks/vto_scale_workflow/VTO_GenMedia_Workflow.ipynb projects/ai/gen-media/notebooks/vto_scale_workflow/LJ_GenMedia_Workflow.ipynb renamed to projects/ai/gen-media/notebooks/vto_scale_workflow/VTO_GenMedia_Workflow.ipynb

@@ -33,7 +33,32 @@
    "source": [
     "# Gen Media end-to-end Workflow, Virtual Try-on usecase\n",
     "\n",
-    "This Jupyter Notebook outlines a complete, scalable pipeline for generating diverse, photorealistic virtual try-on. The core objective is to use a suite of Google's Generative AI models—Gemini (for orchestration and critique), Gemini Image Generation (for creating diverse base models), Vertex AI Virtual Try-On (VTO) (for garment swapping), and Veo (for adding motion)—to produce a large volume of Virtual Try-On images and short motion videos featuring diverse digital models in various outfits. All these are creatd using one platform Vertex AI!\n",
+    "[![Open in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/GoogleCloudPlatform/the-repo/blob/main/projects/ai/gen-media/notebooks/vto_scale_workflow/VTO_GenMedia_Workflow.ipynb)\n",
+    "[![Open in Colab Enterprise](https://img.shields.io/badge/Open%20in%20Colab%20Enterprise-blue?style=flat-square)](https://console.cloud.google.com/colab/notebooks/github/GoogleCloudPlatform/the-repo/blob/main/projects/ai/gen-media/notebooks/vto_scale_workflow/VTO_GenMedia_Workflow.ipynb)\n",
+    "[![Open in Vertex AI Workbench](https://img.shields.io/badge/Open%20in%20Vertex%20AI%20Workbench-orange?style=flat-square)](https://console.cloud.google.com/vertex-ai/workbench)\n",
+    "[![View on GitHub](https://img.shields.io/badge/View%20on%20GitHub-black?style=flat-square&logo=github)](https://github.com/GoogleCloudPlatform/the-repo/blob/main/projects/ai/gen-media/notebooks/vto_scale_workflow/VTO_GenMedia_Workflow.ipynb)\n",
+    "\n",
+    "This Jupyter Notebook outlines a complete, scalable pipeline for generating diverse, photorealistic virtual try-on. The core objective is to use a suite of Google's Generative AI models—Gemini (for orchestration and critique), Gemini Image Generation (for creating diverse base models), Vertex AI Virtual Try-On (VTO) (for garment swapping), and Veo (for adding motion)—to produce a large volume of Virtual Try-On images and short motion videos featuring diverse digital models in various outfits. All these are created using one platform Vertex AI!\n",
+    "\n",
+    "## Prerequisites - Preparing Your GCS Bucket\n",
+    "\n",
+    "Before running this notebook end-to-end, you need to copy the sample dress images to your Google Cloud Storage bucket:\n",
+    "\n",
+    "1. **Copy Sample Dress Images**: The sample dress images are provided in the `dress/` folder. Copy these files to your GCS bucket under the path specified by `OUTFITS_PREFIX` (which is set to \"dress\" by default):\n",
+    "\n",
+    "   ```bash\n",
+    "   # Example command to copy images from the local dress folder to your GCS bucket\n",
+    "   gsutil cp dress/*.png gs://YOUR_BUCKET_NAME/dress/\n",
+    "   ```\n",
+    "\n",
+    "   The notebook expects dress images to be available at: `gs://YOUR_BUCKET_NAME/dress/`\n",
+    "\n",
+    "   Note: The `OUTFITS_PREFIX = \"dress\"` variable in the configuration section defines where the notebook looks for input dress images.\n",
+    "\n",
+    "2. **Update Configuration**: In the Global Configuration section below, update:\n",
+    "   - `PROJECT_ID`: Your Google Cloud Project ID\n",
+    "   - `BUCKET_NAME`: Your Google Cloud Storage bucket name\n",
+    "   - `LOCATION`: Your preferred region (default: us-central1)\n",
     "\n",
     "Created on 11/12/2025"
    ]
@@ -140,9 +165,7 @@
     "# Global Configuration, UPDATE FOR ANY MODEL CHANGES\n",
     "# --- Project & Location Settings ---\n",
     "# Ensure these match your environment\n",
-    "os.environ[\"GOOGLE_CLOUD_PROJECT\"] = (\n",
-    "    \"PROJECT_ID\"  # update your project\n",
-    ")\n",
+    "os.environ[\"GOOGLE_CLOUD_PROJECT\"] = \"PROJECT_ID\"  # update your project\n",
     "os.environ[\"GOOGLE_CLOUD_LOCATION\"] = \"us-central1\"  # update your location\n",
     "\n",
     "PROJECT_ID = os.environ.get(\"GOOGLE_CLOUD_PROJECT\")\n",
@@ -155,10 +178,10 @@
     "# GCS Paths/Prefixes -> The process will create the subsequent file and folder structure\n",
     "CSV_OBJECT_NAME = \"Model_Creation.csv\"\n",
     "MODELS_PREFIX = \"models\"  # Base images of models\n",
-    "OUTFITS_PREFIX = \"Dress\"  # Input dress images\n",
-    "VTO_OUTPUT_PREFIX = \"Dress/4tryon\"\n",
-    "FINAL_PREFIX = \"Dress/4tryon/final\"\n",
-    "MOTION_OUTPUT_PREFIX = \"Dress/4tryon/final_motion\"\n",
+    "OUTFITS_PREFIX = \"dress\"  # Input dress images\n",
+    "VTO_OUTPUT_PREFIX = \"dress/4tryon\"\n",
+    "FINAL_PREFIX = \"dress/4tryon/final\"\n",
+    "MOTION_OUTPUT_PREFIX = \"dress/4tryon/final_motion\"\n",
     "\n",
     "# --- Model Versions ---\n",
     "# Text/Orchestration Model\n",
@@ -509,7 +532,7 @@
     "#Use Case 3 - Virtual Try-On (Vertex AI VTO)\n",
     "\n",
     "Description: Multi-try-on in one shot using the Vertex AI VTO API\n",
-    "- The process involves pairing the generated model images (from Step 4) with input outfit images (from the GCS Dress prefix).\n",
+    "- The process involves pairing the generated model images (from Step 4) with input outfit images (from the GCS dress prefix).\n",
     "\n",
     "- Use a ThreadPoolExecutor to orchestrate the VTO generation in parallel for multiple model/outfit pairs.\n",
     "\n",
@@ -680,7 +703,7 @@
     "\n",
     "- Execute the critique process in parallel using a ThreadPoolExecutor.\n",
     "\n",
-    "- The winning image for each model/outfit combination is copied to the final GCS folder (Dress/4tryon/final)."
+    "- The winning image for each model/outfit combination is copied to the final GCS folder (dress/4tryon/final)."
    ]
   },
   {
@@ -932,7 +955,7 @@
     "\n",
     "- The VTO image is used as the input image and a prompt (e.g., \"slowly walking on a white runway\") is provided to instruct the motion and environment.\n",
     "\n",
-    "- The generated short video clips are uploaded to the final motion GCS prefix (Dress/4tryon/final_motion)."
+    "- The generated short video clips are uploaded to the final motion GCS prefix (dress/4tryon/final_motion)."
    ]
   },
   {
@@ -1029,7 +1052,7 @@
     "QaVTCIINx8JZ",
     "C57vaI8syqHU"
    ],
-   "name": "LJ_GenMedia_Workflow",
+   "name": "VTO_GenMedia_Workflow",
    "provenance": []
   },
   "kernelspec": {

…cale_workflow/LJ_GenMedia_Workflow.nb.py …ale_workflow/VTO_GenMedia_Workflow.nb.pyprojects/ai/gen-media/notebooks/vto_scale_workflow/LJ_GenMedia_Workflow.nb.py renamed to projects/ai/gen-media/notebooks/vto_scale_workflow/VTO_GenMedia_Workflow.nb.py projects/ai/gen-media/notebooks/vto_scale_workflow/LJ_GenMedia_Workflow.nb.py renamed to projects/ai/gen-media/notebooks/vto_scale_workflow/VTO_GenMedia_Workflow.nb.py

@@ -5,7 +5,7 @@
 #       extension: .py
 #       format_name: percent
 #       format_version: '1.3'
-#       jupytext_version: 1.18.1
+#       jupytext_version: 1.20.0
 #   kernelspec:
 #     display_name: Python 3
 #     language: python
@@ -32,7 +32,32 @@
 # %% [markdown] id="B9SgWpCruX5g"
 # # Gen Media end-to-end Workflow, Virtual Try-on usecase
 #
-# This Jupyter Notebook outlines a complete, scalable pipeline for generating diverse, photorealistic virtual try-on. The core objective is to use a suite of Google's Generative AI models—Gemini (for orchestration and critique), Gemini Image Generation (for creating diverse base models), Vertex AI Virtual Try-On (VTO) (for garment swapping), and Veo (for adding motion)—to produce a large volume of Virtual Try-On images and short motion videos featuring diverse digital models in various outfits. All these are creatd using one platform Vertex AI!
+# [![Open in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/GoogleCloudPlatform/the-repo/blob/main/projects/ai/gen-media/notebooks/vto_scale_workflow/VTO_GenMedia_Workflow.ipynb)
+# [![Open in Colab Enterprise](https://img.shields.io/badge/Open%20in%20Colab%20Enterprise-blue?style=flat-square)](https://console.cloud.google.com/colab/notebooks/github/GoogleCloudPlatform/the-repo/blob/main/projects/ai/gen-media/notebooks/vto_scale_workflow/VTO_GenMedia_Workflow.ipynb)
+# [![Open in Vertex AI Workbench](https://img.shields.io/badge/Open%20in%20Vertex%20AI%20Workbench-orange?style=flat-square)](https://console.cloud.google.com/vertex-ai/workbench)
+# [![View on GitHub](https://img.shields.io/badge/View%20on%20GitHub-black?style=flat-square&logo=github)](https://github.com/GoogleCloudPlatform/the-repo/blob/main/projects/ai/gen-media/notebooks/vto_scale_workflow/VTO_GenMedia_Workflow.ipynb)
+#
+# This Jupyter Notebook outlines a complete, scalable pipeline for generating diverse, photorealistic virtual try-on. The core objective is to use a suite of Google's Generative AI models—Gemini (for orchestration and critique), Gemini Image Generation (for creating diverse base models), Vertex AI Virtual Try-On (VTO) (for garment swapping), and Veo (for adding motion)—to produce a large volume of Virtual Try-On images and short motion videos featuring diverse digital models in various outfits. All these are created using one platform Vertex AI!
+#
+# ## Prerequisites - Preparing Your GCS Bucket
+#
+# Before running this notebook end-to-end, you need to copy the sample dress images to your Google Cloud Storage bucket:
+#
+# 1. **Copy Sample Dress Images**: The sample dress images are provided in the `dress/` folder. Copy these files to your GCS bucket under the path specified by `OUTFITS_PREFIX` (which is set to "dress" by default):
+#
+#    ```bash
+#    # Example command to copy images from the local dress folder to your GCS bucket
+#    gsutil cp dress/*.png gs://YOUR_BUCKET_NAME/dress/
+#    ```
+#
+#    The notebook expects dress images to be available at: `gs://YOUR_BUCKET_NAME/dress/`
+#
+#    Note: The `OUTFITS_PREFIX = "dress"` variable in the configuration section defines where the notebook looks for input dress images.
+#
+# 2. **Update Configuration**: In the Global Configuration section below, update:
+#    - `PROJECT_ID`: Your Google Cloud Project ID
+#    - `BUCKET_NAME`: Your Google Cloud Storage bucket name
+#    - `LOCATION`: Your preferred region (default: us-central1)
 #
 # Created on 11/12/2025
 
@@ -108,9 +133,7 @@
 # Global Configuration, UPDATE FOR ANY MODEL CHANGES
 # --- Project & Location Settings ---
 # Ensure these match your environment
-os.environ["GOOGLE_CLOUD_PROJECT"] = (
-    "PROJECT_ID"  # update your project
-)
+os.environ["GOOGLE_CLOUD_PROJECT"] = "PROJECT_ID"  # update your project
 os.environ["GOOGLE_CLOUD_LOCATION"] = "us-central1"  # update your location
 
 PROJECT_ID = os.environ.get("GOOGLE_CLOUD_PROJECT")
@@ -123,10 +146,10 @@
 # GCS Paths/Prefixes -> The process will create the subsequent file and folder structure
 CSV_OBJECT_NAME = "Model_Creation.csv"
 MODELS_PREFIX = "models"  # Base images of models
-OUTFITS_PREFIX = "Dress"  # Input dress images
-VTO_OUTPUT_PREFIX = "Dress/4tryon"
-FINAL_PREFIX = "Dress/4tryon/final"
-MOTION_OUTPUT_PREFIX = "Dress/4tryon/final_motion"
+OUTFITS_PREFIX = "dress"  # Input dress images
+VTO_OUTPUT_PREFIX = "dress/4tryon"
+FINAL_PREFIX = "dress/4tryon/final"
+MOTION_OUTPUT_PREFIX = "dress/4tryon/final_motion"
 
 # --- Model Versions ---
 # Text/Orchestration Model
@@ -438,7 +461,7 @@ def display_images_in_row(
 # #Use Case 3 - Virtual Try-On (Vertex AI VTO)
 #
 # Description: Multi-try-on in one shot using the Vertex AI VTO API
-# - The process involves pairing the generated model images (from Step 4) with input outfit images (from the GCS Dress prefix).
+# - The process involves pairing the generated model images (from Step 4) with input outfit images (from the GCS dress prefix).
 #
 # - Use a ThreadPoolExecutor to orchestrate the VTO generation in parallel for multiple model/outfit pairs.
 #
@@ -593,7 +616,7 @@ def tryon_worker(
 #
 # - Execute the critique process in parallel using a ThreadPoolExecutor.
 #
-# - The winning image for each model/outfit combination is copied to the final GCS folder (Dress/4tryon/final).
+# - The winning image for each model/outfit combination is copied to the final GCS folder (dress/4tryon/final).
 
 # %% id="SehmdW6_x_q9"
 print("[START] AI Critique")
@@ -829,7 +852,7 @@ def process_critique_group(model_stamp, items_for_model):
 #
 # - The VTO image is used as the input image and a prompt (e.g., "slowly walking on a white runway") is provided to instruct the motion and environment.
 #
-# - The generated short video clips are uploaded to the final motion GCS prefix (Dress/4tryon/final_motion).
+# - The generated short video clips are uploaded to the final motion GCS prefix (dress/4tryon/final_motion).
 
 # %% id="CboQHX_JywYl"
 VIDEO_PROMPT = (