Docs
Generation & Models
Every image and video model, what it's good at, and how generation works.
Generation happens from the prompt bar. Pick a model in the Model Picker (left of the Generate button), write a prompt, and press Generate. Leaving the prompt empty triggers an auto-suggestion based on the selected image.
Image models
| Model | Provider | Best for | Notes |
|---|---|---|---|
| FLUX.2 Klein | Replicate | Fastest and cheapest drafts. | Free tier |
| FLUX.2 Pro | Replicate | Balanced quality up to 4MP. | — |
| FLUX.2 Max | Replicate | Highest consistency FLUX variant. | — |
| Nano Banana Pro | Google (Gemini) | Recommended all-rounder for architectural renders. | ★ Recommended default; Free tier |
| Nano Banana 2 | Google (Gemini) | Fast, capable everyday model. | Free tier |
| Nano Banana 2.1 | Google (Gemini) | Cheaper, faster everyday model with sharper edits. | — |
| GPT Image 2.5 | OpenAI | — | — |
| Seedream 5.0 Pro | Replicate | Precise multi-reference editing with up to 10 input images. | — |
You can generate in two ways: basic (regenerate the whole layer) or brush / inpainting (paint a region to edit just that area — the rest of the image is preserved). Reference images steer style, materials and elements.
Video models
Turn an image into a short clip with the video compositor:
| Model | Provider | Duration | Resolutions | Aspect ratios | Audio |
|---|---|---|---|---|---|
| Kling v3.0 | Replicate | 3–15 s | 720p, 1080p, 4K | 16:9, 9:16, 1:1 | Yes |
| Seedance 2.0 | Replicate | 4–15 s | 480p, 720p, 1080p, 4k | 16:9, 4:3, 1:1, 3:4, 9:16, 21:9, 9:21, adaptive | Yes |
| Wan 2.7 | Replicate | 3–15 s | 720p, 1080p | 16:9, 4:3, 1:1, 3:4, 9:16, 21:9 | No |
| Kling O1 Edit | Replicate | — | 16:9, 9:16, 1:1 | No | |
| Runway Gen-4.5 | Runway | 2–10 s | 16:9, 9:16, 1:1, 4:3, 3:4, 21:9 | No | |
| Veo 3.1 | Google (Gemini) | 4 / 6 / 8 s | 720p, 1080p, 4k | 16:9, 9:16 | Yes |
| Gemini Omni | Google (Gemini) | 3–10 s | 16:9, 9:16 | No |
Camera presets
Prompt-driven camera moves available to the video models:
| Preset | Category |
|---|---|
| Static / Tripod | essential |
| Dolly Forward | essential |
| Dolly Backward | essential |
| Walkthrough POV | essential |
| Pan Left | pan |
| Pan Right | pan |
| Tilt Up | pan |
| Tilt Down | pan |
| Crane Up / Reveal | specialty |
| Orbital / Turntable | specialty |
| Aerial Drone | specialty |
| Tracking Shot | specialty |
| Threshold Crossing | specialty |
| Zoom In (Detail) | zoom |
| Zoom Out (Establish) | zoom |
Notes
Pricing is dynamic. Each generation costs a small amount that depends on the model and settings, resolved live in the app — check the in-app cost shown on the Generate button for the authoritative number. Image upscaling uses Freepik Magnific; 3D worlds use World Labs Marble.