🪑
Queue up your products
Add product images on the left. Each generates 4 separate views — front, side, isometric, and luxury interior scene. Finished batches appear here to download.
Completed Batches
Saved render URLs from Supabase and this browser
PDF + ZIP workflow
Bulk Upload
Extraction Review
Analyze the catalog and ZIP on the left, then add selected extracted products to the Bulk Render page for the final queue review.
📋 Matched Products Review
Queue review
Bulk Render
Ready to add to render queue
📦 Batch Processor
Catalog → Queue
1
Upload PDF Catalog + Images ZIP
Analyzing...
2
Extracted Products
Review and edit product details below
📷
Images in ZIP
All images extracted from the ZIP archive
3
Match Results
Accept, reject, or pick different images for each product
📋 Product Features — View Prompts
Click Edit on a view before changing its prompt
💡 Tip: Be specific about material, color, shape, and style. The more detail you add, the more accurate the AI output will be.
📡 API Monitor
Health checks for all configured external services
Last checked: —
🖼️ Matched Images
Saved original product images with matched descriptions from PDF+ZIP uploads
Product detail & actions
ℹ️ About — Product Image Studio
AI-powered product photography studio
📖
How to Use
- Select Images — Click the upload box in the sidebar to add product photos (PNG, JPG, WEBP, up to 10MB each).
- Add Description — (Optional) Describe materials, colors, and style to help the AI preserve key details from the reference image.
- Set Brand Reference — (Optional) Enter a furniture brand name (e.g. Minotti, Poliform) so the AI matches the brand's design language and aesthetic.
- Choose Resolution — Select output resolution: 0.5K (512px, fastest), 1K (1024px, recommended), 2K, or 4K.
- Select AI Provider — Choose between Gemini 3, OpenAI GPT Image, or Stability AI.
- Add to Queue — Click "Add Selected to Queue" to add items. Toggle Auto-Queue to start rendering immediately after adding.
- Start Queue — Click "▶ Start Queue" to begin processing. The background worker on the VPS processes up to 5 items concurrently.
- Download Results — Once complete, each batch card shows all 4 generated views. Click individual images to preview & download, or use "Download All" to get a ZIP.
🤖
AI Models & Specifications
Gemini 3 — Flash / Pro
Provider: Google
Model: gemini-3.0-flash-001 / gemini-3.0-pro-001
Capability: Image-to-image generation
Max Resolution: 4K
Best for: Fast, high-quality renders with strong brand/style adherence
Auto-selects Flash for standard views, Pro for interior scenes (higher detail). Brand reference is injected directly into the prompt for style matching.
OpenAI — GPT Image
Provider: OpenAI
Model: gpt-image-1-mini by default
Capability: Image editing & generation
Max Resolution: 4K
Best for: Photorealistic renders, e-commerce quality
Uses image-to-image editing API. Brand reference is included in the prompt to guide style and material choices.
Stability AI — SDXL
Provider: Stability AI
Model: sdxl-v1.0 / sdxl-v1.0-turbo
Capability: Image-to-image generation
Max Resolution: 4K target sizing
Best for: Fast drafts and alternate visual treatments
Uses the server-side STABILITY_API_KEY. Cheap/mini provider variants use the turbo model when configured.
🔑
API Key Resources
Gemini API
GEMINI_API_KEY
Open Gemini 3 docs
OpenAI API
OPENAI_API_KEY
Open OpenAI API page
Stability AI
STABILITY_API_KEY
Open Stability platform
⚙️
Workflow & Architecture
1
Upload
Images are uploaded to Supabase Storage for persistent access.
2
Queue
Items are saved to Supabase database with status tracking (waiting → active → done/error).
3
Process
A background worker on the VPS polls every 5 seconds, picks up queued items, and generates all 4 views in parallel using the selected AI provider.
4
Store
Generated images are saved to Supabase Storage and the database is updated with status and URLs.
5
Drive Upload
(Optional) If Google Drive is configured, completed renders are automatically uploaded to a folder named after the product.
6
Download
Results appear in the main panel. Download individual images or all 4 views as a ZIP file.
🎨
The 4 Generated Views
1
Front View
Straight-on product shot. Best for thumbnails, e-commerce listings, and hero images. White/clean background.
2
Side View
Profile perspective showing product depth, proportions, and silhouette. Clean background.
3
Isometric View
Angled 3D-feel shot. Best for marketing materials, social media, and catalog spreads. Clean background.
4
Interior Scene
Lifestyle render placing the product in a luxury interior setting. Best for social media, catalog hero images, and brand storytelling.
💡
Tips for Best Results
- Use high-quality reference images with good lighting and clear product visibility.
- Write detailed descriptions — mention materials (e.g. "tan leather, brushed brass legs"), colors, and textures.
- Set a brand reference to match the product's design language (e.g. "Minotti" for warm Italian luxury, "Vitra" for modern design).
- For interior scenes, Gemini 3 Pro is recommended for the best detail and composition.
- Use 0.5K resolution for quick drafts and testing prompts, then switch to 1K or 2K for final renders.
- Enable Auto-Queue when processing multiple items to save clicks.
- Customize the AI prompts via the 📋 Features panel to fine-tune how each view is generated.
📜 Queue Processing Log
0 entries
No queue activity yet.
Start the queue to see processing history here.
Start the queue to see processing history here.
📋 Failure Log
0 entries
No failures logged yet.