Motion generation guide
Describe an action as full-body 3D motion
JST-2 text-to-motion generates a body performance from a written action. The useful output is movement for a compatible character, rather than a finished video with lighting, editing and sound.
Write a performance the body can follow
Start with an action, its direction and a clear finish. Describe movement rather than camera cuts or visual effects. A prompt with one short sequence gives you a concrete way to assess whether the result follows the intended action and preserves its order.
Editorial prompt suggestion, not a tested output: “A person takes two small steps forward, stops with both feet planted, raises the right hand to wave, then lowers it.” In the studio, keep the prompt within 400 characters and choose a duration of 3–60 whole seconds.
Check contact and timing, not just the silhouette
Watch the complete movement before focusing on details. Does the person finish the requested action? Inspect weight shifts, planted feet and transitions between poses. A smooth-looking clip can still slide across the floor or skip a meaningful pause in the prompt.
Compare motion on a neutral character before attributing a problem to appearance. Then test the intended character, especially at the shoulders, hands and hips. These checks are practical evaluation suggestions; they are not claims that every generated motion will satisfy production requirements.
Apply the result to a compatible character
Viggle says JST-2 motions apply directly to JST-2 characters without a separate retargeting step and are natively compatible with standard 3D character rigs. Its developer overview describes skeletal animation export for use in games and animation applications.
The studio exposes a GLB motion asset when returned and lets you select completed character and motion assets for preview. Confirm scale, orientation and compatibility before using the files elsewhere. Motion generation does not by itself provide an environment, a rendered film or a soundtrack.
Change one movement instruction at a time
If the action feels wrong, identify whether the issue is direction, pace, pose or sequence. Revise the relevant phrase while keeping the rest of the brief stable. This makes the comparison useful even when repeated generations produce different results.
For choreography that must match an existing performance, video-to-motion is a different starting point: it reconstructs movement from footage. Text-to-motion is appropriate when you want to describe an action without filming it. Neither workflow guarantees physically accurate contact with arbitrary scene objects.
Questions & answers
Is text-to-motion the same as an AI animation video generator?
No. This guide concerns full-body 3D motion for a compatible character. A finished animation video also needs character appearance, scene setup and rendering. Viggle offers other products, but their video workflows should not be confused with this motion task.
Can I request several actions in one prompt?
You can describe a sequence where the selected interface permits it, but a short ordered brief is easier to evaluate. Start with one action or a simple transition. Check that all requested stages appear before increasing the complexity.
Will the character correctly touch a prop or another person?
A body motion does not guarantee contact with a particular scene object or performer. Inspect the resulting alignment in your target scene. JST-2 scene collisions and object interactions remain rule-based rather than a general learned-physics guarantee.
Check the original sources.
Reviewed . Model claims are attributed to Viggle; practical suggestions are editorial guidance.