Gesture and pose tests
Explore a wave, turn or short sequence with an illustrated or realistic character.
Use one image for the character’s appearance and one video for movement. Choose orientation and background guidance, then review the new generated performance.
Tips for better results
Reference video
MP4 or MOV, 3s to 10s for `image` orientation, up to 30s for `video`.
Reference image
JPG or PNG, max 10MB, minimum 340px, aspect ratio between 2:5 and 5:2.
Track previews, status, and downloads in one place.
Prepare a six-second, single-shot video of someone waving and turning slightly. Keep head and hands visible. Pair it with a character image framed at a similar scale; avoid asking a cropped portrait to supply a full-body action.
Sign in, choose orientation and upload both files. Wait until they are ready. In optional settings, choose which input should guide the background. A short prompt can describe the desired scene without contradicting the reference movement.
Generate, then review the result in history. Watch the hand, face, clothing edges and turn throughout the clip. If the body breaks or the subject disappears, improve the source pairing before submitting another paid request. Download a useful take and assemble it in Video Edit AI. For a portrait driven by an audio recording, choose Lip Sync. See pricing or all applications.
Explore a wave, turn or short sequence with an illustrated or realistic character.
Use a clear movement reference to test a character concept. Match the visible body area in the image and video.
Explore a mascot or character gesture before adding the chosen take to a larger edit. Check recognizable details and proportions.
Use a prompt or still references when you do not need motion from a video.
Change a clip’s appearance instead of transferring its performance.
Edit the still image before using it as an animation reference.
Use a clear subject with the required limbs visible. A full-body performance needs a full-body image; a headshot gives little information for leg movement. Similar framing helps, but exact identity or movement is not guaranteed.
Image orientation asks the character to stay aligned with the photo’s facing direction and accepts a 3–10-second reference. Video orientation follows the reference’s turns and angles and accepts 3–30 seconds. This is character orientation, not portrait-versus-landscape output. Choose it before preparing the clip.
Reference video: MP4/MOV up to 100 MiB. Character image: JPG/PNG up to 10 MiB. Both must be readable, at least 340 pixels on each side, with a width-to-height ratio between 2:5 and 5:2. Prepare a shorter source elsewhere if needed; there is no range trimmer here.
Prefer one continuously visible subject, clear gestures and few cuts or obstructions. Select the video background or image background in optional settings to guide the scene. This requests a generated treatment; it does not promise an unchanged background or perfect movement transfer.
This page has no audio, resolution or export-format selector. Preview the generated video and listen to any audio before downloading. The result is a video, not an editable 3D rig or motion-capture data file.
Yes. Selecting valid inputs after sign-in uploads them for cloud processing. Generation uses account credits; consult current pricing. Refresh checks an existing task. There is no task Retry button here, and another Generate request can consume credits again. Completion time varies.