DeepMotion - AI Motion Capture & Body Tracking
DeepMotion · 圖片 · 影片
DeepMotion is a browser-based AI motion capture and body tracking platform that converts ordinary video, or a written prompt, into 3D character animation. Its Animate 3D tool tracks full-body movement, faces and hands from phone or camera footage with no suit and no special hardware, while SayMotion generates motion from text. Both export to FBX, BVH, GLB and MP4, which makes DeepMotion a practical entry point for indie animators, game developers and anyone who needs mocap without renting a studio.

關於 DeepMotion - AI Motion Capture & Body Tracking
What Is DeepMotion
DeepMotion is a web platform for markerless motion capture and 3D animation, built by a team that has worked on motion technology since 2014. The pitch is simple: you shouldn't need a suit, a dedicated capture stage or a full animator's toolkit to get believable character movement. You upload a video or type a description of the motion you want, and the AI returns a rigged animation you can drop into your project.
The company runs two products. Animate 3D handles video-to-3D animation and tracks body, face and hand motion, including up to 16 people in a single clip on the top plans. SayMotion goes the other direction and generates 3D motion from a text prompt, with editing tools that let you loop, blend and refine the result. Both live in the same browser portal and share a library of custom characters, so you can retarget one animation onto a different character without reprocessing everything.
The limits matter as much as the features. Animate 3D clips are short: 20 seconds on the free tier, 30 seconds on Innovator, and only Professional and above allow anything close to feature-length shots. Video processing is asynchronous, so a job queues and finishes later rather than instantly. SayMotion is still labeled Open Beta, is desktop-only, and currently supports full-body animation only, with hand and face motion listed as future work.
So who is it not for? If you need frame-perfect actor performance for a film, this isn't that tool. If you need a lot of usable animation cheaply, it probably is.
Getting Started
- Create a free account at deepmotion.com and sign in to the motion portal. No credit card is needed for the Freemium tier.
- Pick a product: Animate 3D for video-based motion capture, or SayMotion for text-to-3D animation.
- In Animate 3D, upload a clip and, if you want a specific rig, a custom FBX, GLB or VRM character. In SayMotion, type a prompt that describes the motion.
- Choose settings before running the job. Face and hand tracking, physics simulation, foot locking and motion smoothing all change the output, and your plan determines which of them are available.
- Review the result in the portal, polish it with the Rotoscope Pose Editor if needed, then export as FBX, BVH, GLB or MP4.
產品資訊
快速了解 DeepMotion - AI Motion Capture & Body Tracking 的定價、支援平台與效能。
適合對象
這項工具最適合的使用者、任務與情境。
使用者
- Indie animators and solo creators
- Game developers prototyping character movement
- XR and virtual production artists
- Students learning animation
任務
- Turning phone footage into 3D motion
- Generating animation from a text prompt
- Facial and hand capture
- Multi-person scenes
- Retargeting existing motion onto a new rig
情境
- Building a library of animations on a budget
- Prototyping a game or film sequence before committing to production-quality work.
- Social media and short-form content
- Architectural or product visualisation
主要功能
Markerless Motion Capture From Video
Animate 3D extracts full-body motion from a single ordinary video, no markers, no suit, no depth camera. You can film on a phone or pull a clip from an archive, and the AI handles tracking and retargeting onto a rigged character. That removes the two biggest barriers to mocap for small teams: equipment cost and studio time. No suit to rent. No stage to book.
Face and Hand Tracking
The platform tracks facial expressions and hand gestures from the same footage it uses for the body, so lip movement and finger articulation come through without extra hardware. Hand tracking is precise enough for sign language and detailed gestures. Very fast finger motion still produces occasional artifacts, which the Rotoscope Pose Editor is built to clean up.
Text to 3D Animation in SayMotion
SayMotion generates 3D motion from a written prompt, using DeepMotion's motion dataset as the source. You describe the action, pick a character and get multiple variants to choose from. It's the part of the platform that feels genuinely different from a traditional mocap pipeline, since you can produce a movement that never existed on camera. Most motion capture software makes you supply the performance; this tool lets you describe it.
AI Editing Tools: Inpainting, Loop, Merge and Refine
Animations rarely come out perfect, and SayMotion's editing layer is where most of the practical value sits. Inpainting inserts new motion into an existing animation, Loop matches the first and last frames so a cycle repeats without a visible jump, Merge combines variants into longer sequences and Refine applies a creativity scale to adjust motion detail. For a walk cycle, Loop alone can save hours of manual keyframing.
Rotoscope Pose Editor
The Rotoscope Pose Editor is a frame-by-frame correction tool that overlays your animation on the source video so you can drag a joint back where it belongs. It supports multi-joint selection, interpolation between two poses and keyboard shortcuts, and edits auto-save as you go. DeepMotion holds a patent pending on it, and the Free Animation Credit Program pays you back credits for using it.
Physics Simulation and Motion Smoothing
Physics Simulation reduces impossible movement, like a hand passing through a leg, and Motion Smoothing uses filters to remove jitter. Both are configurable per job, and both matter more than they sound: raw AI tracking wobbles, and the difference between a stiff result and a convincing one often comes down to these two settings.
Custom Characters and Automatic Retargeting
Upload your own FBX, GLB or VRM character, or build one with the integrated Ready Player Me and Avaturn creators. The AI retargets animation onto whatever character you choose, and higher plans raise the number of custom characters you can store, from 3 on the free tier up to 100 on Enterprise.
Export Formats and Pipeline Compatibility
Output covers FBX, BVH, GLB and MP4, plus DMPE on Professional plans and above, along with JPG, PNG and GIF for poses. That covers Blender, Maya, Unity, Unreal Engine, iClone and Roblox workflows, so DeepMotion slots into an existing pipeline instead of replacing it. Camera angles, environment backgrounds and transparent image export are available for the video and pose outputs.
Credit-Based Pricing With an Earn-Back Program
Every job consumes credits, and the Free Animation Credit Program lets you earn up to 100% back by labeling your own uploaded videos with text descriptions or by correcting animations in the Rotoscope Pose Editor. It's an unusual model, and it's effectively paid work in exchange for platform time. Heavy users on the Studio plan get unlimited credits, with the first 7,200 each month processed at the highest priority. Credits reset monthly and don't carry over.
優缺點
優點
- No hardware or suit needed, so mocap becomes an option for individuals rather than only studios.
- Both video-to-3D and text-to-3D animation in one portal, with a shared character library and export formats that fit Blender, Unity, Unreal Engine and Roblox.
- The free tier is real: 25 credits a month, 1080p uploads and personal-use exports. No card required.
- AI editing tools (Loop, Inpaint, Merge, Refine) and the Rotoscope Pose Editor cover the gap between a rough AI result and something usable.
- REST APIs exist for both Animate 3D and SayMotion, so animation can be generated inside your own application.
缺點
- Clip length is capped at 20 seconds on the free plan and 30 seconds on Innovator, which rules out continuous long takes without stitching. Professional plans raise the ceiling to 120 seconds and Studio to 360 seconds.
- Jobs run asynchronously through a priority queue, so turnaround depends on your plan tier. You can't get instant results.
- SayMotion is Open Beta and desktop-only, and currently animates the body only. Face and hand generation are listed as coming later.
- Video-based capture works best with clean footage. Motion toward and away from the camera, occlusion and cluttered backgrounds all degrade tracking quality.
常見問題
It generates 3D character animation from video or text. The typical use is producing mocap-quality movement for games, film, XR and social content without renting a capture stage or hiring a performer for a traditional suit-based session.
相關內容
探索與 DeepMotion - AI Motion Capture & Body Tracking 相關的工具、技能與文章。
DeepMotion - AI Motion Capture & Body Tracking 替代方案
X Ray Interpreter
X-ray Interpreter · 圖片X Ray Interpreter 是一款以網頁為基礎的 AI 放射科工具,能把 X 光、電腦斷層、磁振造影、超音波與 PET 影像變成白話報告。你上傳一張掃描影像,片刻之間就得到初步的 X 光判讀結果,之後若有需要解釋的地方,還能繼續追問。它的定位是第二意見與學習輔助,而非醫療診斷。
Aieasypic
AIEasyPic · 圖片AIEasyPic 是一款 AI 圖片生成器,能在幾秒內把單純的文字提示變成完成的藝術作品。你也可以用自己照片訓練自訂模型、在現有圖片中替換臉部,並從文字建立短影片片段,全部都能在瀏覽器中完成。它適合想要快速視覺素材的休閒創作者,也適合想打造個人模型、又不想碰任何程式的業餘玩家。
Chargen
Chargen · 圖片Chargen 是一款 AI 角色生成器與世界觀建構工具,專為《龍與地下城》與其他桌上角色扮演遊戲打造。它能將一句構想變成手繪風格的肖像,在同一場工作階段中生出 NPC、怪物、地圖與遭遇,並把每隻生物的細節放在手邊。桌上角色扮演美術這部分靠一套名為 Gold 的點數系統運作,而文字生成器則對所有人保持免費。
