Authentication Required: Sign in and grant access to Google Cloud to use the tool.
Authentication Troubleshooting
If you're having trouble:
Ensure pop-ups are not blocked for this site.
Ensure you select an account with the 'Vertex AI User' role (or appropriate permissions) on project h5g-ai.
Make sure the Vertex AI API is enabled for the project.
Verify the OAuth Client ID is correct and this app's URL (e.g., `https://...pages.dev`) is an Authorized JavaScript origin.
Check the browser console (F12 -> Console) for specific error messages.
Try revoking access and signing in again.
Try a hard refresh (Ctrl+F5 or Cmd+Shift+R).
Check the `_headers` file is present and deployed for COOP/COEP headers if seeing policy errors.
OAuth Status: Not initialized
Generating 0 / 0
Processing...
1. Design
2. Prompts
3. Images
Game Requirements & Inputs
Select model for text generation and/or image analysis. Vision capability required for image analysis.
Applicable only to Gemini 2.5 Flash. Uncheck to potentially reduce latency/cost at the expense of reasoning quality.
Custom Game Symbol Configuration
Choose how many of each symbol type you want for your custom game:
đ¯ Core Symbols
High-Pay (HP)
Medium-Pay (MP)
Low-Pay (LP)
đĄ All LP symbols use same type. 5 LPs: A,K,Q,J,5 | 6 LPs: A,K,Q,J,10,5 | Gemstones: Different cuts & colors
Backgrounds
âšī¸ Each background generates both landscape (1536x768) and portrait (768x1536) versions.
⥠Special Symbols
Wild
Scatter
Bonus
Jackpot
WYSIWYG
Multiplier
Collector
Avatar
Special Feature
đ§ Generation Options
âšī¸ Creates decorative frames for HP, MP, Wild, Bonus, Scatter, Jackpot, and Special symbols. LP symbols don't use frames per design guidelines.
Symbol Summary
Review your selections above
Text Requirements (Theme, Style, General Notes)
Describe your desired game theme, art style, and any general requirements below. If cloning, the required assets will be listed automatically.
Image Input (for Influence Modes)
Drag & Drop up to 10 slot game images here, or click to select files.
Max 10 images, 10MB per image.
In Clone Mode: Images influence overall Theme & Art Style applied to the cloned structure.
In Standard Mode: Used only if an 'Influence...' mode is selected above.
Or click here and paste images (Ctrl+V / Cmd+V)
In Clone Mode: Guides AI interpretation for theme/style.
In Standard Mode: Applied only when an 'Images Influence...' mode is selected.
Or Import Existing:
Symbol & Asset Designs
Your symbol and asset designs will appear here.
Generated Prompts
0Total0Selected
Review, edit, and organize your AI-generated prompts. Select the ones you want to use for image generation.
â ī¸Input design or source changed. Prompts might be outdated.
âšī¸Prompts edited. Existing images might not match.
Selection Controls
Category Filter
Export Options
Prompt Library
đ
â¨
No Prompts Yet
Your generated prompts will appear here once you complete the design step.
Generated Images
Configure settings and generate images. Select ONE image per required asset for Production Download.
â ī¸ Image Generation Failures
Some prompts failed during image generation. You can edit the prompt and retry, or remove the error entry.
Choose from the 9 main generation models. Imagen 4 Ultra recommended for highest quality. Fal.ai models offer fast, high-quality generation with unique features.
Enter your OpenAI API key for gpt-image-1.
Enter your Fal.ai API key in format: UUID:HEX_STRING
Only applies to gpt-image-1. Transparent background will be applied via API parameter.
Number of images generated for each selected prompt. Max 4 for most models.
Controls generation of people/faces.
Adjusts content safety sensitivity.
Let Imagen rewrite prompt (may deviate from Agent 3). 'False' recommended.
Terms to exclude from image generation. Edit as needed.
2.5
Controls how closely the image follows the prompt (1.0-10.0, default: 2.5)
30
Number of denoising steps (10-50, API default: 30). Higher = better quality.
Seed for reproducible results. Leave empty for random generation.
Image output format for Qwen Image model.
Enable or disable Fal.ai's built-in safety checker for Qwen Image.
3.5
CFG scale - how closely to stick to your prompt (1.0-10.0, default: 3.5)
Seed for reproducible results. Leave empty for random generation.
Image output format for FLUX Kontext model.
Safety tolerance level for generated images (1=strict, 6=permissive).
4.5
CFG scale - how closely to stick to your prompt (1.0-10.0, default: 4.5)
50
Number of denoising steps (10-50, optimized for quality: 50). Higher = better quality.
Seed for reproducible results. Leave empty for random generation.
Image output format for FLUX Krea model.
Generation acceleration - higher speed trades some quality for faster results.
Enable or disable Fal.ai's built-in safety checker for Krea model.
5.0
CFG scale - how closely to stick to your prompt (1.0-10.0, default: 5.0)
50
Number of denoising steps (10-100, optimized for quality: 50). Higher = better quality.
Seed for reproducible results. Leave empty for random generation.
Image output format for HiDream model.
Enable or disable Fal.ai's built-in safety checker for HiDream model.
Art style for Recraft V3. All realistic styles, sketches, and children's styles excluded. Default: Digital Illustration (General).
Enable or disable Fal.ai's built-in safety checker for Recraft V3.
Your generated images will appear here. Select one per required asset if downloading for production.
×
Image Caption
Model:-
Symbol:-
Category:-
Target:-
Browse Templates
×
Select a category
0 items
Select a category from the left to view available templates.
đ¨ Customize Assets
Specify what you want specific assets to be. Leave unchecked for system generation. All changes auto-save as you type. The AI will ensure no duplicates and apply your theme/style consistently.