βThe image-plus-video setup makes runway motion easier to test than writing body movement from scratch. Cleaner full-body references gave me the most usable drafts.β
Maya Chen
Fashion motion draft
Upload a character image, add one motion reference video, and use Kling 3 Motion Control to transfer pose, gesture, body timing, and camera rhythm while preserving the subject's identity and style.
Kling 3 Motion Control is a Kuaishou Kling video model for driving a still character image with movement from a reference video. The image supplies the subject to preserve, while the motion video supplies the action, body timing, pose sequence, and camera rhythm.
Use it when exact movement matters more than prompt-only direction: dance transfer, fashion movement, presenter gestures, avatar motion, product-in-hand demos, and social video variations. On Toolaze, this flow uses one character image, one motion reference video, selectable Image or Video character orientation, and 720p or 1080p output.
AI Video Model
| Performance Metric | Toolaze Specification |
|---|---|
| Input mode | Image-to-video motion control |
| Required inputs | 1 character image plus 1 motion reference video |
| Character image | 1 JPEG, PNG, or JPG image, max 10MB; image size must be greater than 300px with aspect ratio from 2:5 to 5:2. |
| Motion reference video | 1 MP4 or QuickTime motion reference video, max 100MB. Image orientation supports 3-10 second references; Video orientation supports 3-30 seconds. |
| Prompt | Prompt is optional; use it to preserve identity, outfit, lighting, framing, and style. Keep it under 2500 characters. |
| Character orientation | Choose Image to match the character image orientation, or Video to follow the character orientation in the reference video. |
| Duration | Output duration follows the uploaded motion reference video duration instead of a fixed manual duration. |
| Motion control source | Pose, gesture, walk cycle, hand movement, body timing, and camera rhythm come from the uploaded video. |
| Resolution | 720p or 1080p |
| Credit guidance | Credits scale with the uploaded motion reference video duration; current Toolaze rate is 40 credits per output second at 720p or 54 credits per output second at 1080p. |
Start with Kling 3 Motion Control selected so the generator expects both a character image and a motion reference video.
Use one clear JPEG, PNG, or JPG image that shows the face, outfit, body shape, product, or style you want to preserve.
Upload one MP4 or QuickTime clip with the pose, gesture, walk, dance, hand action, or camera rhythm you want to transfer. Choose Image or Video orientation based on which source should define character direction.
Choose 720p or 1080p. Output duration and total credits follow the uploaded reference-video length, and the optional prompt can protect identity, outfit, lighting, framing, and style.
Apply a reference dance, walk, turn, or gesture to a character image while preserving identity, clothing, and body proportions.
Turn a still outfit or editorial portrait into a runway-style video guided by a clean motion reference.
Create front-facing explainer, greeting, reaction, or profile clips where posture and hand timing follow supplied footage.
Use a reference gesture to guide how a product is held, rotated, lifted, or shown to camera.
Draft vertical character clips for Reels, TikTok, Shorts, and ads when you already have motion footage to match.
Check whether a pose, gesture, or body movement works before producing a longer campaign, scene, or creator video.
Choose Motion Control when a reference video should drive the action. Choose a standard Kling model when you do not have motion footage or need broader text-to-video settings.
| Decision Point | Kling 3 Motion Control | Kling 2.6 Pro Motion Control | Kling 3.0 |
|---|---|---|---|
| Toolaze model | Kling 3 Motion Control | Kling 2.6 Pro Motion Control | Kling 3.0 |
| Input modes | Image-to-video motion control with 1 character image + 1 motion reference video | Image-to-video motion control with 1 character image + 1 motion reference video | Prompt or multi-image video generation, up to 4 images |
| Video reference support | Required: 1 motion reference video | Required: 1 motion reference video | Does not use an uploaded motion reference video |
| Motion control style | Reference-video motion transfer with optional prompt guidance for identity, outfit, framing, and style | Reference-video motion transfer with optional prompt guidance for identity and style | Prompt-described subject action, camera path, shot sequence, and scene style |
| Reference duration | Image orientation supports 3-10 seconds; Video orientation supports 3-30 seconds; output duration follows the reference video | Image orientation supports 3-10 seconds; Video orientation supports 3-30 seconds; output duration follows the reference video | Manual duration choices in the standard Kling 3.0 selector |
| Resolution | 720p or 1080p | 720p or 1080p | 720p, 1080p, or 4K |
| Credit guidance | 40 credits per output second at 720p or 54 credits per output second at 1080p | 22 credits per output second in the current Toolaze pricing table | Starts at 28 credits per output second for standard 720p |
| Choose it when | You want the newer Kling motion-control model to follow a specific action video | You need lower-cost motion transfer with the Kling 2.6 motion-control flow | You want broader prompt-driven generation, more image references, 4K, or longer scene planning |
Pick the model by the input you already have: motion reference footage, a cinematic prompt, or a polished visual draft.
| Decision Point | Kling 3 Motion Control | Wan 2.7 | Seedance 2.5 |
|---|---|---|---|
| Good fit | Motion transfer from one character image and one motion reference video | Short cinematic clips from text, image, or reference-led prompts | High-detail video drafts with strong visual finish and scene continuity |
| What to check first | Whether your motion video is clean, visible, within the accepted duration range, and close to the character framing | Whether your idea needs Wan-style text/image/video reference direction rather than exact body-motion transfer | Whether the selected Seedance setup supports your input type, duration, and quality target |
| Prompt style | Preserve identity, outfit, proportions, lighting, camera framing, and style while following the reference motion | Scene setup, shot rhythm, motion direction, camera movement, and output setting | Subject, style, scene change, lighting, pacing, and quality cues |
| Reference guidance | One image gives the subject; one video gives the movement | Reference media can guide the scene, but motion is not copied as directly as a driving video | Reference support depends on selected Seedance setup and availability |
| When to choose another model | Choose another video model when you do not have a motion reference video | Choose Motion Control when exact pose, gesture, or dance transfer matters most | Choose Motion Control when matching a specific action video matters more than visual polish alone |
βThe image-plus-video setup makes runway motion easier to test than writing body movement from scratch. Cleaner full-body references gave me the most usable drafts.β
Maya Chen
Fashion motion draft
βI used a hand-motion reference to plan how a small device should turn in frame. Keeping the prompt focused on shape, logo clarity, and lighting made the draft easier to review.β
Jordan Lee
Product video planning
βFor presenter clips, matching the crop between the portrait and motion video mattered more than adding a long style prompt. The gesture stayed clearer when the reference was simple.β
Nina Patel
Avatar gesture test
It is an image-to-video generator flow for applying motion from one reference video to one uploaded character image while preserving identity, outfit, framing, and visual style.
No. Kling 3 Motion Control requires both a character image and a motion reference video. Choose a standard video model if you do not have motion footage.
No. The character image and motion reference video are required; the prompt is optional and useful for preserving identity, outfit, lighting, framing, and style.
Toolaze supports one character image and one motion reference video for this flow. The image can be JPEG, PNG, or JPG up to 10MB; the motion video can be MP4 or QuickTime up to 100MB.
Image orientation supports 3-10 second reference videos. Video orientation supports 3-30 second reference videos. The generated output duration follows the uploaded reference video duration.
Image orientation matches the character direction from the uploaded image. Video orientation follows the character direction from the uploaded motion reference video.
Kling 3 Motion Control on Toolaze supports 720p and 1080p output options.
Credits scale with the uploaded reference-video duration. The current Toolaze rate is 40 credits per output second at 720p or 54 credits per output second at 1080p.
Video generation uses Toolaze credits. Starter credits, available settings, reference-video length, and final cost can vary by account quota and selected resolution.
Both use one character image and one motion reference video. Kling 3 Motion Control uses the newer Kling 3 motion-control model, while Kling 2.6 Motion Control remains a lower-cost nearby option on Toolaze.
Choose Motion Control when you have a motion reference video and need a character image to follow it. Choose standard Kling 3.0 when you want broader prompt-driven generation, more image references, 4K, or longer scene planning.
Keep it short and protective: preserve identity, outfit, body proportions, lighting, camera framing, and style while following the uploaded motion reference. Remove extra scene changes that compete with the video.
Create Wan 2.7 videos from text, images, video references, or audio direction. Use multi-shot prompts, first/last-frame transitions, continuation scenes, and synchronized sound for short cinematic clips at 720p or 1080p.

Create Wan 2.6 videos from text prompts or one uploaded image. Use multi-shot structure, cinematic camera direction, synchronized sound cues, 5/10/15 second timing, and 720p or 1080p output for polished short clips.
Create Wan 2.5 videos from a written prompt or one reference image. Choose text-to-video or image-to-video, set 16:9, 9:16, or 1:1, pick 5 or 10 seconds at 720p/1080p, and prompt for motion, camera, materials, and synchronized sound.