Kai’Sa AI Dance Photo Guide: Preparation and Common Fixes
The source image is the part of a fixed dance template you can control most directly. Choose it for readable identity and body shape. A beautiful portrait can still be a difficult dance input if the face is hidden or most of the body is missing.
Start with a visible face and body
For a full-body dance, prefer an image showing the head, torso, arms and legs. Leave a margin around the subject so hands and feet are not already touching the image edges. The model must invent any parts of the body or outfit absent from the source.
Compare two typical candidates: a close selfie may show the face clearly but provide no information about the lower outfit; a sharp full-body photo provides both. Prefer the full-body version when its face is still large and clear enough to recognize.
Use even light and natural detail
Choose an image with readable facial features rather than heavy shadows across the eyes. Avoid motion blur, strong smoothing filters and very small screenshots. Enlarging a blurry image does not restore the details that were never captured.
Make the silhouette easy to read
A single subject against a simple background reduces ambiguity about which figure should dance. Keep arms distinguishable from the torso where possible. A group photograph or a mirror scene with several visible figures can leave the intended subject unclear.
Distinct clothing can help define a look, but tiny logos, dense patterns and fine jewelry may change across frames. If exact outfit details matter, review those areas in the result rather than assuming they will remain identical.
- One clearly identifiable subject.
- Feet and hands visible where practical.
- No large objects covering the body.
Prepare original character artwork
Original character art can be explored when you have permission to use it. Prefer a complete, coherent figure over a collage or a character sheet showing several poses. A single front-facing or slightly angled view makes the intended subject easier to understand.
Check uploads and diagnose a poor result
Save the image as JPG, PNG or WebP, within the 10 MB upload limit. If your phone exports another format, convert it to a supported format rather than only renaming the extension. Recheck the converted image for rotation and cropping.
If the face changes during the dance, try a sharper source with no hands, sunglasses or shadows covering it. If limbs or clothing look wrong, check whether those areas were visible in the original image. Fast turns and occlusion can still cause errors even with a good source.
If a hand or foot disappears only in another player or editor, compare it with the original downloaded MP4: a crop or zoom may be hiding it. For a new generation, keep the settings fixed and change one input at a time so the comparison is useful.
Frequently asked questions
Can I use a selfie?
You can try a clear selfie, but a tight face crop leaves the body and outfit for AI to invent. A readable full-body image is a better starting point for this dance.
Will a better photo prevent every distortion?
No. Better input can reduce ambiguity, but fast motion and occlusion can still cause inconsistent faces, hands or clothing.
What if a supported image still will not upload?
Check that it is no larger than 10 MB, inspect the displayed error and verify your connection. Keep a copy before reloading. If the problem persists, contact support with the error details.