Scenario Maya Animation logo

Scenario Maya Animation

Organization
scenario-labs
scenario-maya-animation

Use when animating a character or prop in Maya through Python: a walk or run cycle, a jump or landing, a heavy lift, throw or punch, an acting or dialogue shot, lip sync, eye darts and blinks, stepped blocking, splining, polish, Graph Editor tangents, animation layers, mocap cleanup or retargeting, game clips and root motion, or reviewing a playblast. Also when motion looks floaty, weightless, mushy, poppy, posey, sliding or chattery, or feet slap or skate.

Overview

Publisherscenario-labs
Repositoryskills
Skill namescenario-maya-animation
Stars
970
Forks
114
Bundled files
6
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 6 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by scenario-labs on GitHub. Read the source before you install it.

Installation

Install the Scenario Maya Animation AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/scenario-labs/skills/tree/main/skills/dcc/maya/scenario-maya-animation
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/scenario-labs/skills.git /tmp/skills
mkdir -p .claude/skills
cp -r /tmp/skills/skills/dcc/maya/scenario-maya-animation .claude/skills/scenario-maya-animation
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Scenario Maya Animation in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Scenario Maya Animation on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Scenario Maya Animation is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Maya animation (animator)

Expert animation means every frame is a choice: poses that read as drawings from the camera, holds and snaps placed on purpose, designed spacing, weight shown by resistance and a caught stop. Work big to small (story, body, face, detail); never add detail to a stage that does not read yet. An agent cannot scrub, so it measures (world and screen tracks, spacing, holds, contacts, a ballistic fit, overlap lags), then looks at a playblast or a headless sheet before calling anything done. If a sibling skill named here is missing from your available skills, ask the user to install it (npx skills add scenario-labs/skills --skill <name>); unattended, proceed from tool schemas and flag the gap.

REQUIRED BACKGROUND: scenario-maya-expert (execution channel, review loop, 2027 version traps).

Toolkit: scripts/mx_anim.py. Its analysis and image layer is pure Python and ran offline.

Stance (the expert delta)

  • Pick the workflow per shot; blocking ends when the information is in. Neistadt: pose to pose for acting, straight ahead or extremes for flow; blocking is done when all five key categories exist (main, anticipation, overshoot and follow-through, breakdown, hold). Elver splines physical shots from the start; Camporota proves timing on two boxes first.
  • Maya is "the worst in-betweener in the world" (Newman). Feed it keys: ones on fast actions, 2 to 4 frames apart on slow parts. Spline readiness is a test: spline the range, play; mush means back to stepped with more poses. Native Set Key reshapes neighboring curves (Wade): insert keys, prove it with a curve diff, and share keys across every control before splining.
  • Weight is resistance, then release, then a caught stop. Wade: show what the character is up against (a delay, acceleration from rest) before its strength; strength is a frame count. Elver: more up and down reads as a balloon, not weight; weight is a fall caught abruptly, a bigger hold in the squash, residual jiggle. Heavy air is gravity-true [added physics]; Wade's weighted apex adds hang, for light or cartoony bodies.
  • Feet and spine sell every weight shift (jump, hop, landing, lift). Camporota: animate the foot roll in every crouch, it sells the compression; in a big drop the spine drags behind the hips and only the arms drop; first passes overdo the anticipation, so dial it back; a touch-down goes flat "almost in one frame". Elver: on the first airborne frame the toe tip points back to where it left the ground.
  • Judge in camera space, at speed. Elver: a motion trail can show a perfect 3D arc that fails on screen. Lazare: the eye goes to the biggest motion, so the non-focus character stays "appropriately still". Project tracks through the shot camera; read the playblast before any metric.
  • The base layer is the animation; mocap is precious. Newman: a layer is a time-boxed experiment with marker keys and a protective key; never clean curves on it, merge with Smart Bake, then clean (Elver: only a late global note or an additive tremble, then bake). On mocap fix only the defect asked for, or the cascade ends "not hand key and not mocap"; after a merge, Euler filter, then delete keys down to extremes.
  • Lip sync animates sounds, not words. Santos: jaw only first (the muppet pass), then the big five (jaw, two corners, top and bottom lip) for 90%; lead by place of articulation, bilabials on the frame. Lazare: shift every mouth key 1 to 2 frames earlier only when the whole line plays late.
  • Polish is triage; review is top down. Elver ranks the notes, spends on the ones that pay, owns every contact frame. Newman grades Intent, Mechanics, Performance, Integration; a playblast does not prove game feel.

Establish first

InputWhy it changes the planDefault when the brief is silent
Style and destinationsquash, stops, closeness to reference: cartoony pushes; live action needs pose squash and no dead stops (Elver); VFX stays near reference (Neistadt)realistic feature quality
Frame ratetiming numbers, currentUnit24 fps (film)
Weight and strengthframes of resistance and of overcome (Wade: decide both numbers first)heavy: resistance beat of 2+ frames, no hang
Rigcontrol names, IK/FK, spaces, foot roll attribute (map once with listAttr(keyable=True))hands in IK, no IK/FK or space switches inside a shot (Newman)
Cameraa locked shot camera is the only judge; games also need every angle and the enginebuild the shot camera, lock it
Reference, audiolandmarks (contacts, takeoff, apex, landing, blinks, darts) give rhythm, then push (Neistadt)beat list from the brief
Game metricsspeeds and heights agreed in round numbers before animating (Newman); root motion or in place (Epic)in place, metrics sheet requested

Workflow

  1. Plan in numbers. Beats with frames, the leading part per beat, key categories, contacts, windows (crouch, push, air, impacts), style; A.store_plan(plan). Physical shots: A.air_frames(rise, drop) budgets the air. Dialogue: phrases with subtext and an action verb, ruled sounds (M, B, P, F, V, L, T, D), 2 to 4 tentpoles (Santos, Wade). GATE: frame counts add up.
  2. Set up. fps, range, sound, shot camera locked; query the tangent defaults (2027 notes). GATE: mx_validate.validate(profile="shot", rules={"fps": 24}) has no fail.
  3. Proxy pass (big, fast or camera-bound moves). A.proxy_boxes(), key translate and lean, air with A.ballistic_keys(), rotation 1 to 2 frames behind (Camporota). The first anticipation is usually too big: A.scale_curve(root, "translateY", 0.8, rest_y, time=crouch). GATE: A.evaluate_gates() on the proxy (ballistic, catch, resistance, push_fastest); proxy_curves.png looked at.
  4. Blocking, stepped. A.set_key_defaults("blocking"); A.key_pose(pose, f, fill=ctrls) keys every control on every pose; passes: main, anticipations, overshoots, breakdowns (favor a neighbor, never 50/50), holds (A.key_hold()); A.share_keys(ctrls). Never act from the neutral rig pose (Neistadt). Crouch poses: foot roll keyed, spine dragging, arms hanging (Camporota). GATE: every key category present; stepped playblast (GUI) or tracks.png; write the story it reads, then check silhouettes.
  5. Spline pass 1, the core. snap = A.snapshot(ctrls) (the buffer curve); A.to_spline() on COG, spine, legs only; face, eyes, fingers stay stepped (Camporota, Santos). Air gravity-true; broken tangent into a hard landing (Newman); planted feet flat. Mush: A.restore(snap), add poses. Every later key: A.insert_key(), then A.curve_diff() on sampled snapshots is ok outside its two segments (P4). GATE: A.run_gates(plan) ballistic, hang, catch, slide, penetration, foot_roll, spine_drag, heel_to_flat pass; no A.overshoots() through holds.
  6. Spline pass 2, overlap. Arms, head, secondary: correlated chains edited together first (Newman), then A.offset_keys() by 1 to 2 frames down the chain. GATE: overlap lags not all zero; pose_to_pose_tell low; tracks clean in camera.
  7. Face, if the shot has one. Body first, then the mask (eyes, brows), then the mouth: muppet jaw keyed at A.mouth_keys(phonemes), big five, tongue, connection by A.copy_curve() (nostrils 1 frame behind the corners, dampened). Eyes linear, 2-frame darts; blinks close faster than they open. GATE: A.jaw_report(), A.lip_lag() against A.audio_envelope(), A.blink_report(), A.transitions(); close-ups on tentpoles.
  8. Polish, by triage. Fix the top-ranked notes, one section at a time. Contacts and fast moves on ones; peel-off on every takeoff; impacts get a pushed squash and a bigger hold; pops become 3 frames on an arc plus an overshoot; A.retime() tightens a section; moving holds; a motion-blurred render before final (all Elver). GATE: every gate passes (peel_off, impact_hold included), layers merged, no sub-frames, Euler filtered, critique.md on the final playblast.

Variants. Cycle: poses every 3 frames, loop with A.cycle() before polish, offsets inside the cycle keep the period. Game clip: metrics, A.cycle_speed(), loop seam, root rules, then scenario-maya-pipeline-scripting. Mocap: procedures P12; retarget: P13; heavy interactions: P9.

Numbers

ValueRelative toSource
Gravity 1.703 cm/frame^2 (981/576)24 fps, scene in cmphysics [added]
Walk keys every 3 frames: contact 1, down 4, pass 7, up 10, contact 13; loop 1 to 2524-frame cycleCamporota
Touch-down to flat foot about 1 frame, at most 2; walk arm offsets 0, 1, 2, head 1every step and landingCamporota; body digest
Rotation 1 to 2 frames behind translationproxy and bodyCamporota
Keys on ones on fast actions; 2 to 4 frames apart on slow onesblockingNewman
Blink close 2 / open 4 (or 3 / 5; big eyes one more)framesElver
Eye dart 2 frames, linear; brows lead lids by 1 to 2, never 5framesElver
Snappy move: 3 frames on an arc plus a small overshoot, never 1framesElver
Global lip shift 1 to 2 frames earlier; opens near full on frame 1when late at speedLazare
Big five = 90% of lip sync; jaw plus corners 50 to 60%controlsSantos, Elver
Game speeds: walk 2, run 4 to 6, sprint 12 to 15 m/sstylized playerNewman
Playblast default scale 0.5: pass percent=100active viewMaya 2027 Help

Quality gates

Measurable (A.run_gates(plan); pure logic A.evaluate_gates), for any jump, hop, landing or weight shift in the plan: fps; heavy air g_ratio 0.8 to 1.25, hang excess at most 1 frame, constant horizontal speed; catch (60%+ of the fall speed gone in 2 frames, decel shorter than recovery), then a 3+ frame hold at the bottom; push fastest; heavy resistance beat of 3+ still frames; every crouch: foot roll moves, spine lags the hips; touch-downs flat within 2 frames; lift-off toe within 30 degrees of the takeoff spot; planted feet at most 0.2 cm/frame, penetration under 0.5 cm; pre-landing stretch; no pops outside impacts; overlap lags; moving end hold; tell under 90% after splining; no sub-frames; no layers at polish. [added] thresholds are toolkit defaults. Visual: A.review_images(plan, out) (headless: curves.png stacked lanes, tracks.png camera-space tracks with onion-skin poses); in the GUI A.playblast_pack() (shot camera plus a no-lights silhouette) and a stepped-preview playblast. Read frames in time order; write the first-view read before the numbers (critique.md).

Common mistakes

MistakeWhat it looks likeFix
Splining before the poses are inmush, arms tweened through the bodyA.restore(snap), add breakdowns stepped
Even spacing, whole body on the same framesfloaty, "hitting poses"favor neighbors, holds, offset_keys down the chains
First-pass anticipation keptthe crouch upstages the actionA.scale_curve(..., time=crouch) about the rest value
Torso locked to the hips, arms swung back in the crouchstiff, generic anticipationspine drags behind the hips, only the arms drop
Feet flat in the crouch, pointing forward at takeoff, slow to flatten on landingweak compression, no push, slapping feetfoot roll; toe aimed back at the takeoff spot; flat within 1 to 2 frames
Stylized hang on a heavy bodyfloaty apexballistic_keys, flat apex, broken contact tangent
No resistance before the effortstyrofoam weight2+ frames of strain before the push
Symmetric dip, or a bounce straight out of the squashlight landingsudden catch, bigger hold, one overshoot, slow recovery
S or bare setKeyframe on a splined curveneighbors reshape, holds driftA.insert_key, then A.curve_diff
Spline through a holddrifting holdkey_hold pillars, flat or plateau sides
Arcs checked on a 3D trailclean trail, broken arc on screencamera_track pixel tracks
Lip sync keyed on the soundlate at speedper-sound lead, then a global shift

Handoffs

Receives from scenario-maya-rigging and scenario-maya-deformation: the rig referenced (never imported), validate(profile="rig") clean, the control list with IK/FK, space and foot roll attributes, a range-of-motion playblast. Rig problems (knee pop, flipping space) go back as a note with frames. Delivers shots to scenario-maya-lighting-rendering: a new saved version, validate(profile="shot") clean, layers merged, helpers deleted (A.delete_helpers()), run_gates JSON with no fail, the final playblast the agent looked at, what was not verified. Game clips to scenario-maya-pipeline-scripting: skeleton baked per clip, clip names and ranges (bookmarks), root rule (in place or root motion), loop seam and speed reports.

Maya 2027 notes

  • Default tangent: Auto Span (Legacy) on one doc page, Clamped on another: query keyTangent(q=True, g=True, itt=True). Lock/Free Tangent Weight is now Tangent Length; Butterworth is Smooth Filter (Butterworth).
  • Time Slider > Enable Stepped Preview; A.stepped_preview() is the scripted twin. Dope Sheet Time Snap is off since 2025.1 and scaleKey leaves sub-frames: A.snap_subframes().
  • Ghost Editor and Motion Trail Editor sit under Visualize; ghosting a skinned mesh needs Cached Playback (disabled by motion blur, Trax, classic dynamics).
  • Playblast needs a model panel (GUI via mx_bridge); in mayapy use review_images or Arnold stills.
  • Sequencer (ex Camera Sequencer): lock every camera. HumanIK bakes from its own window, not Bake Simulation. Time Editor clips never mix with animation layers. Blue Pencil replaced Grease Pencil.

References

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Scenario Maya Animation AI skill do?

Use when animating a character or prop in Maya through Python: a walk or run cycle, a jump or landing, a heavy lift, throw or punch, an acting or dialogue shot, lip sync, eye darts and blinks, stepped blocking, splining, polish, Graph Editor tangents, animation layers, mocap cleanup or retargeting, game clips and root motion, or reviewing a playblast. Also when motion looks floaty, weightless, mushy, poppy, posey, sliding or chattery, or feet slap or skate.

Why use Scenario Maya Animation on TypingMind?

Because you install it once and use it with any model. Scenario Maya Animation is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Scenario Maya Animation in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/scenario-labs/skills/tree/main/skills/dcc/maya/scenario-maya-animation. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Scenario Maya Animation?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Scenario Maya Animation?

As many as you like. As long as a model supports skills, you can use Scenario Maya Animation with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Scenario Maya Animation AI skill free?

Yes. It is published on GitHub by scenario-labs under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇