Image Models
DUTO supports multiple AI models for image generation and editing. Models are served through WaveSpeed as the primary provider.
Available Models
Nano Banana Pro
Ultra-fast image generation (default model)
google/nano-banana-pro/text-to-image| Attribute | Value |
|---|---|
| Quality | Very Good |
| Speed | Very Fast |
| Cost | 10 credits |
| Provider | Google (via WaveSpeed) |
Strengths:
- Extremely fast generation
- Excellent for iterations
- Good prompt adherence
- Wide aspect ratio support
Best for:
- Rapid prototyping
- Concept testing
- High-volume workflows
- Quick iterations
Aspect Ratios: 1:1, 3:2, 2:3, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9
Nano Banana Pro 4K/8K Ultra
High-resolution output
google/nano-banana-pro/text-to-image-ultra| Attribute | Value |
|---|---|
| Quality | Excellent |
| Speed | Fast |
| Cost | 25 credits (4K), 40 credits (8K) |
| Provider | Google (via WaveSpeed) |
Strengths:
- Ultra-high resolution
- Professional quality
- Fast for resolution
- Print-ready output
Best for:
- Print materials
- Large format displays
- Professional deliverables
- High-detail work
Resolutions: 4K, 8K
Nano Banana 2
Next-generation image generation with editing support
google/nano-banana-2/text-to-image| Attribute | Value |
|---|---|
| Quality | Excellent |
| Speed | Fast |
| Provider | Google (via WaveSpeed) |
Strengths:
- Improved quality over Nano Banana Pro
- Supports both generation and editing modes
- Visual mask-based editing
Best for:
- Higher quality generation
- Image editing with visual masks
- Inpainting tasks
Edit variant: google/nano-banana-2/edit (supports visual mask editing)
SeedDream v4.5
Budget-friendly image generation
bytedance/seedream-v4.5| Attribute | Value |
|---|---|
| Quality | Good |
| Speed | Fast |
| Cost | 8 credits (1024x1024), 15 credits (2048x2048) |
| Provider | ByteDance (via WaveSpeed) |
This is the default Budget Mode image model when auto-selection is enabled.
Strengths:
- Cost-effective
- Reliable quality
- Good detail level
- Consistent results
Best for:
- Budget-conscious projects
- Standard resolution needs
- Bulk generation
- Testing concepts
Resolutions: 1024x1024, 2048x2048
Edit variant: bytedance/seedream-v4.5/edit (image editing capabilities)
SeedDream v5 Lite
Improved SeedDream with editing support
bytedance/seedream-v5.0-lite| Attribute | Value |
|---|---|
| Quality | Good |
| Speed | Fast |
| Provider | ByteDance (via WaveSpeed) |
Strengths:
- Improved over v4.5
- Default edit model for many workflows
- Good for variant generation
Best for:
- Image editing workflows
- Variant generation
- Inpainting with the Angles node
Edit variant: bytedance/seedream-v5.0-lite/edit (default edit model for many nodes)
Flux Kontext
Advanced LoRA-enhanced generation
wavespeed-ai/flux-kontext-dev-lora-ultra-fast| Attribute | Value |
|---|---|
| Quality | Excellent |
| Speed | Very Fast |
| Cost | 12 credits (1024x1024), 20 credits (2048x2048) |
| Provider | WaveSpeed AI |
Strengths:
- Ultra-fast LoRA generation
- Excellent prompt adherence
- Advanced style control
- Great consistency
Best for:
- Brand-consistent content
- Style-preserving generation
- Character consistency
- Professional workflows
Resolutions: 1024x1024, 2048x2048
Model Comparison
| Model | Quality | Speed | Base Cost | Best For |
|---|---|---|---|---|
| Nano Banana Pro | ★★★★ | ★★★★★ | 10 | Quick iterations |
| Nano Banana Ultra | ★★★★★ | ★★★★ | 25-40 | High-res output |
| Nano Banana 2 | ★★★★★ | ★★★★ | - | Generation + editing |
| SeedDream v4.5 | ★★★ | ★★★★ | 8-15 | Budget work |
| SeedDream v5 Lite | ★★★★ | ★★★★ | - | Editing workflows |
| Flux Kontext | ★★★★★ | ★★★★★ | 12-20 | Professional work |
Supported Features
| Feature | Nano Banana Pro | Nano Ultra | Nano Banana 2 | SeedDream v4.5 | SeedDream v5 Lite | Flux Kontext |
|---|---|---|---|---|---|---|
| Text to Image | Yes | Yes | Yes | Yes | Yes | Yes |
| Image Edit | Yes | Yes | Yes | Yes | Yes | - |
| Inpainting | Yes | Yes | Yes | Yes | Yes | - |
| Upscaling | - | Yes | - | - | - | - |
| LoRA Support | - | - | - | - | - | Yes |
Aspect Ratios
All models support various aspect ratios:
| Ratio | Dimensions | Use Case |
|---|---|---|
| 1:1 | 1024x1024 | Social posts, profiles |
| 3:2 | 1536x1024 | Landscape photography |
| 2:3 | 1024x1536 | Portrait photography |
| 3:4 | 1080x1440 | Traditional portrait |
| 4:3 | 1440x1080 | Traditional display |
| 4:5 | 1080x1350 | Instagram portrait |
| 5:4 | 1350x1080 | Fine art |
| 9:16 | 1080x1920 | Stories, vertical |
| 16:9 | 1920x1080 | Landscape, presentations |
| 21:9 | 2048x870 | Ultrawide, cinematic |
Model Selection Tips
Workflow Strategy
- Concept phase - SeedDream v4.5 (8 credits)
- Refinement - Nano Banana Pro (10 credits)
- Final output - Flux Kontext (12 credits) or Nano Ultra (25 credits)
Quality Requirements
| Project Type | Recommended |
|---|---|
| Social media | Nano Banana Pro |
| Print materials | Nano Banana Ultra |
| Client work | Flux Kontext |
| Internal drafts | SeedDream v4.5 |
| Rapid prototyping | Nano Banana Pro |
| Editing / inpainting | SeedDream v5 Lite or Nano Banana 2 |
| Variant generation | SeedDream v5 Lite |
Prompt Tips by Model
Flux Kontext
Handles complex, detailed prompts with LoRA styles:
Professional product photography of a luxury watch,
studio lighting with soft shadows, marble surface,
macro detail on watch face, 4K quality, commercial style
SeedDream v4.5
Keep prompts focused and clear:
Luxury watch on marble, studio lighting, professional photography
Nano Banana Pro
Fast, responsive to clear prompts:
Watch on marble, product photo, studio lighting
Troubleshooting
Quality Issues
Problem: Output looks blurry or low quality
Solutions:
- Use higher quality model (Flux Kontext or Nano Ultra)
- Check aspect ratio settings
- Improve prompt specificity
- Try higher resolution option
Inconsistent Results
Problem: Same prompt gives very different results
Solutions:
- Use seed for reproducibility
- Be more specific in prompt
- Try Flux Kontext for consistency
- Use reference images
Prompt Not Followed
Problem: Output doesn't match prompt
Solutions:
- Simplify prompt
- Use premium model for better adherence
- Break into essential elements
- Add emphasis to key elements