Spatial control just landed in @MagicMotionAI. 🪄🚨
Place waypoints. Draw paths. Keyframe full-body poses and hand/foot targets using control rig then hit Generate.
Available now for all registered users, and when you are ready, also for your agents via API.
#3D#AI#Animation
Come watch live as Savi, the gamedev agent built into @Spawn generates the animations for an entire fighting game using only the @MagicMotion API. https://t.co/MfLAYvQFmL
When you need precise control over your character animation text prompts alone won't do. Getting started with your first directed motion only takes a few seconds.
Using the new MCP your agent can also direct motion for you the same way.
Your agent can now direct 3D character motion down to the frame.🪄🤖🎬
Motion creation superpowers for any agent, one MCP URL:
Infinite chaining
Mix and match various constraint types
Image to pose
Frame-accurate game mechanics and cinematics
We'll be pushing an update to the MMAI hosted mcp sometime later today to improve the instructions, so your agents have a better idea of how to unleash its full potential right out of the gate when you connect to it.
Including using image 2 pose fed right into pose constraints.
We gave Savi, @Spawn's game-dev agent, our API and pointed it at a fighting game we're building, and now your agent can do the same thing easily with our hosted MCP server.
It made full animation sets that match each other. Nobody touched an animation by hand.
Awesome experiments will be shared soon where a gamedev agent uses the @MagicMotionAI constraint based API directly and learned how to create incredibly intricate animations with it.
Better animation than Opus5.5/Astra alone for a fraction of the time and tokens.
@Awesome_O_AI Appreciate it! The experiments are out now, and anyone's agent can do the same through our hosted MCP server. Full thread with the clips:
https://t.co/MuBUaVJiiQ
Magic Motion AI now runs a hosted MCP server.
Add https://t.co/iBOt9YrXCg to your MCP client, sign in, and your agent can generate 3D character motion that hits the mark: floor marks, paths and poses.
🪄🤖
To celebrate the MCP launch, we just gave every @MagicMotionAI account 60 free credits 🎉
New here? Sign up and get 60 credits too. We've removed CC based trials.
Add our hosted MCP to your agent, sign in, and have it start directing 3D character motion in less than a minute.
@luccacerf We think it's going to fill a major gap for gamedev agents and hope it allows you to get much more out of the platform especially long term as we improve it further.
Please don't hesitate to reach out with any questions, feedback, suggestions.
Magic Motion AI now runs a hosted MCP server.
Add https://t.co/iBOt9YrXCg to your MCP client, sign in, and your agent can generate 3D character motion that hits the mark: floor marks, paths and poses.
🪄🤖
@luccacerf We've found that most SoTA LLMs are great at retargeting. The MCP server explains the MMAI rig to your agent which makes very easy for them to retarget generated motions to just about any rig you can throw at it for just a few tokens. Even custom-built ones. 🚀
3D humanoid motion was the one thing agents couldn't do well.
Now it's a tool call.
Add our hosted MCP to your agent and let it animate.
Docs: https://t.co/4iAmzoa1OG
There's nothing to install and no GPU to run.
It's just one URL you connect any MCP client, sign in, and your agent is directing 3D character motion within minutes. Every take comes back as FBX, ready for @Blender, @unity, @UnrealEngine, @threejs, and more.
Two characters, one scene. Your agent stages both in 3D space with paths, waypoints, marks and headings: each walks to their mark and they bow in sync. Takes chain into longer sequences.
It's a fraction of the tokens and time of having an LLM hand-key it.
Your agent can build a whole animation state machine: idle, walk, run, attack, hit, recover just through the MMAI MCP server.
Pose constraints pin where each clip starts and ends, so the set is designed to chain together instead of getting stitched up after the fact.
An agent that can direct and generate 3D motion quickly and accurately can:
build a full move set from a design doc
stage previz with characters on their marks
keep a whole cast's animation consistent
Savi didn't need reference images either or hand poser either.
It generated its own pose boards, turned them into pose constraints using our image to pose endpoint, and generated the whole set from there.
You describe the move. Now with our MCP easily let your agent make it.
We gave Savi, @Spawn's game-dev agent, our API and pointed it at a fighting game we're building, and now your agent can do the same thing easily with our hosted MCP server.
It made full animation sets that match each other. Nobody touched an animation by hand.
There is a better way than either of these methods.
Give your agent our API, save it from burning tons of tokens while also allowing it to have nearly unlimited generative humanoid (not for long) motion capabilities.
🪄🧐
Haters were right: it makes zero sense to burn LLM tokens animating humanoids
instead of draining my Opus 5.5 or Astra limits, i switched to Mixamo and got way smoother zombies in 10 minutes
exact pipeline to copy:
1 - make a mesh on Tripo or Meshy (both have free tokens)
2 -don't rig it, just download the raw mesh
3 - send the mesh to Sol / Astra / Argon / Opus / Sonnet / DeepSeek / Grok
4- ask the LLM to set up Mixamo with your model (pick manually or use an existing pack)
in a couple of minutes = ready-to-use rigged and animated 3D model