Character creation guide
Generate a 3D character from a reference image
Turn a character image into an animation-ready 3D asset with JST-2. This studio supports image-to-character generation; Viggle’s broader model release also describes generating characters from text.
Start with a readable character reference
In the studio, select character mode and upload a PNG, JPEG or WebP image up to 3 MiB. A clearly visible subject gives you a stronger reference for checking identity, clothing and proportions. The 30-credit character request is separate from a scene or motion request.
JST-2 can complete cropped or partially occluded characters according to the release. Hidden features are inferred, so a plausible result may differ from the design you intended. Supply artwork or photographs you have permission to process, especially when the subject is a recognizable person.
What animation-ready means in this model
JST-2 represents character parts as persistent Physics-Native Tokens with position, attributes and learned dynamics. Viggle says those dynamics describe how parts deform during motion, allowing its character workflow to animate without a separate manual rigging and skinning stage.
The studio exposes the returned rigged VSPLAT character asset. That does not guarantee a particular polygon topology or a ready-made export for every downstream tool. Check the actual asset and the destination application’s requirements before planning a production pipeline around it.
Inspect the shape before adding a complex action
Review the character from multiple viewpoints. Check the face against the source, then inspect hands, limb proportions and clothing boundaries. A strong front view can conceal a problem at the back or beneath an arm. Judge the full asset rather than its preview alone.
Editorial evaluation suggestion, not a tested output: begin with a slow turn and a simple arm raise. These movements make it easier to inspect joint alignment and clothing deformation before trying a fast dance or a full-body action.
Keep appearance and performance as separate decisions
Viggle says JST-2 motions can be applied directly to JST-2 characters without a separate retargeting step. This makes it useful to judge character identity separately from motion timing. A different action will not necessarily repair a shape problem in the original character.
The launch article identifies improvements to complex clothing deformation and hair motion as upcoming. Test those areas in the movements your project needs. A character that looks convincing while standing still may need a different reference or further work for an extreme pose.
Questions & answers
Can JST-2 create a full character from a cropped image?
Viggle describes support for cropped and partially occluded inputs. The missing body regions are generated interpretations, not recovered ground truth. Compare the result with your intended design and use a fuller reference when those details matter.
Do I need to rig a JST-2 character manually?
The official JST-2 workflow is designed to provide animation-ready characters without a separate rigging and skinning step. Moving the asset into another tool can still require compatibility checks and adjustments for that tool’s animation pipeline.
Is a character the same output as a scene SPZ?
No. This studio’s scene workflow returns SPZ environment assets, while character generation exposes a rigged VSPLAT asset when returned. Character generation uses 30 site credits. Check format support in the application where you want to animate the character.
Check the original sources.
Reviewed . Model claims are attributed to Viggle; practical suggestions are editorial guidance.