LTX-2 Image-to-Video (I2V) | Make an LTX-2 Video from a Reference Image
Learn how to generate an LTX-2 video from an image. Best image types, motion prompts, camera moves, and settings for clean, consistent results.
Use image-to-video (I2V) when you want a specific look, identity, or composition to stay consistent. This guide shows how to turn a single reference image into controlled motion.
If you are using the LTX-2 video generator for I2V, the tips below help keep identity stable and motion clean.
When to use image-to-video
I2V is ideal when you need:
- Consistent character or product identity
- A specific visual style or lighting
- Less randomness than text-only generation
Step 1 - Choose a good reference image
Great I2V inputs usually have:
- A clear subject that is not too small
- Stable lighting (avoid heavy glare)
- Minimal text overlays
- Little or no motion blur
Recommended: single subject, clean background, cinematic framing.
Step 2 - Write a motion prompt (the key to I2V)
For I2V, your prompt should describe motion, not just appearance.
Motion prompt template:
- Subject motion: turns head slightly, walks forward, fabric sways
- Camera: slow pan, gentle dolly-in, locked shot
- Environment: wind, dust particles, soft rain
- Style constraints: keep identity consistent, do not change outfit
Example prompt:
Animate the reference image. The character stays consistent.
Subtle breathing motion, hair gently sways.
Slow dolly-in camera. Soft floating dust in warm light.
Do not change clothing or face.Step 3 - Pick a safe camera move
If you want clean results:
- Best: slow push-in, slow pan, static shot
- Risky: fast whip pans, big zooms, extreme rotation
Step 4 - Settings that usually help
Start with:
- Duration: short (increase later)
- Quality: Balanced, then High once stable
- Motion strength: Low or Medium at first
Common issues and fixes
Identity drift (face or outfit changes)
Fix: reduce motion strength, use shorter duration, add constraints like "keep identity consistent."
Warping around edges
Fix: simplify camera motion, use a cleaner reference image, increase quality after the prompt is stable.
Video barely moves
Fix: add explicit movement verbs, increase motion strength slightly, describe environment motion.
FAQ
Q: Can I use product images?
A: Yes. Best results come from well-lit product shots with clear edges.
Q: Should I still include style words (cinematic, anime)?
A: Yes, but keep them short. Motion guidance matters most in I2V.