Skip to content

LTX 2.5 / Image to video

LTX 2.5 image-to-video: write a focused motion brief

An image-to-video prompt should tell the model what changes over time. The image already supplies much of the subject, colour and composition. Rewriting those details with contradictory adjectives can distract from the motion you actually want.

By JollyAIReviewed 2 min read

Separate fixed details from movement

Begin by identifying what should stay: the object, its design and the scene layout. Then state one movement for the subject or camera. If both move, explain their relationship. “The camera follows beside the bicycle” is clearer than independently asking for a rotating camera and a fast bicycle.

Prepare the source for the intended framing and check the preview in the playground. Preset dimensions and source handling can affect the result; do not rely on a resolution phrase embedded in the prompt.

Build a first attempt

  1. Upload a clear source image to LTX 2.5 Image to Video.
  2. Keep the first motion modest enough to compare with the source.
  3. Add a short audio brief if sound matters to the scene.
  4. Watch for shape drift and unintended changes in the background.

Example prompt

PROMPT RECIPE / ADAPT TO YOUR SOURCE

Keep the bicycle, adult rider's protective clothing and trail from the image. The rider moves slowly forward along the visible path while the camera follows from the same side. Stable frame and believable wheel motion. Sound: tyres rolling softly on gravel and light wind.

Refine one variable

If the result is too static, strengthen the action sentence while keeping the camera fixed. If it is unstable, reduce the angle change and simplify the background activity. Save the source and each successful prompt so later iterations have a consistent baseline. A selected result can be useful without demonstrating that every action, image or seed is equally reliable.

REVIEWED JOLLYAI IMPLEMENTATION

The workflow behind the guide

Uploaded source + motion brief → image/text conditioning → configured LTX 2.5 generation graph → decoded video and audio.

Inputs, presets & access
PRO. Prepare the source composition and check the current preset before generation.
Implementation note
JollyAI retains ltx2 workflow identifiers while remapping model components to LTX 2.5. The configured graph contains video and audio generation paths. Access is PRO; available duration and quality presets are shown in the playground. Downloaded-file metadata determines the delivered size and duration.

Configuration reviewed September 20, 2026. Your current playground controls and plans page determine availability. Prompt recipes above are suggestions, not guarantees or claimed test results.

Open Image to video ↗

Sources and useful next steps

The implementation notes describe JollyAI's reviewed local integration. Upstream capabilities and licences may differ from the settings available here.

Found an outdated setting or a reproducible problem? Contact JollyAI with the workflow and job reference. Keep account tokens and private input files out of public reports.

KEEP BUILDING

Continue the workflow.