The Video tab offers 6 modes of generation, selectable from a dropdown at the top of the form. As with image, the mode filters the model list, then the chosen model determines which fields appear.
🔺 A quirk of this tab: the Generate Video button is disabled while a generation is running ("Another generation in progress"). It is the only tab that forbids concurrent generations.
Image to video (default mode)
You supply a still image and describe the motion you want. The model "animates" your image while respecting its composition.
- ✨ Animating illustrations: making a dragon's wings beat, a hero's hair float.
- 🏞️ Animated backgrounds: turning a static set into a video backdrop.
- 📸 Animated portraits: a slight animation for dialogue screens.
Text to video
You describe a scene in a prompt (action, mood, style, camera movement…) and the model generates an entirely synthetic video. The negative prompt excludes unwanted elements ("no text, no blur").
- 🎬 Game cinematics: intro or transition sequences.
- 📱 Vertical content (9:16): micro-animations for mobile or short formats.
- 🏫 Training material: dynamically illustrating abstract concepts.
Start and end frame
You supply two images: the first frame (start) and the last (end). The model generates all the intermediate frames to create a smooth transition.
- 🔄 Character morphing, 🚀 storyboard transitions, 🗺️ map zooms, ⚡ reveal effects.
Image and audio to video (AI avatar)
You supply a portrait image and an audio file. The model syncs lip movements and facial expressions to the audio.
- 🎙️ Video NPCs, 📺 AI presenters, 🌍 video localization, 🎭 dubbing existing assets.
Video to video (extension / retake)
You start from an existing video to extend, or to alter from a defined starting point.
- ➕ Extending cinematics, ✏️ fixing video endings, 🔁 building loops.
Motion Control
You combine a character image and a motion reference video. The model applies the reference movements to your character.
- 💃 Character animation, 🥋 action scenes, 🎓 instructional characters.
Settings you will meet most often
- Prompt and negative prompt.
- Duration: either a list of durations offered by the model, or a free number.
- Aspect ratio: 16:9, 9:16, 4:3, 1:1…
- Resolution and FPS: depending on the model.
- Audio: depending on the model, either a simple "generate with audio" switch, or an audio file to supply (mandatory in avatar mode).
- Reference image, first frame, last frame, source video, start time, character orientation: depending on mode and model.
🔺 All files (reference image, audio, source video) are picked from the project files: import them first.