MuscleMapKit

An interactive 3D muscle map for iOS that shades each muscle by how hard it was worked, spins under your finger, and reports taps back by muscle group.

_{If MuscleMapKit is useful to you, a coffee keeps it going.}

Install

.package(url: "https://github.com/haplollc/MuscleMapKit.git", from: "1.0.0")

Try it

import MuscleMapKit

MuscleBody3DView(
    intensities: [.chest: 1.0, .frontDelts: 0.6, .triceps: 0.45]
)
.frame(height: 380)

If this is useful, a star helps others find it.

Built for Cinder, where it answers the question you actually have after a session: not did I train, but what did I actually hit.

Install

.package(url: "https://github.com/haplollc/MuscleMapKit.git", from: "1.0.0")

.target(name: "YourApp", dependencies: ["MuscleMapKit"])

Requires iOS 17+.

Use it

import MuscleMapKit

MuscleBody3DView(
    intensities: [.chest: 1.0, .frontDelts: 0.6, .triceps: 0.45]
)
.frame(height: 380)

intensities is [MuscleGroup: Double] in 0…1. Anything you leave out renders untrained, so you only pass what was worked.

Reacting to taps

MuscleBody3DView(
    intensities: intensities,
    onTapMuscle: { muscle in
        print("tapped \(muscle.displayName)")
    }
)

Inside a scrolling list

Gestures are opt-out, because a body that spins under your finger will fight a ScrollView for the same drag:

MuscleBody3DView(
    intensities: intensities,
    autoRotate: false,
    interactive: false,        // let the scroll view win
    initialYaw: .pi / 5        // rest at a 3/4 angle
)

Turning exercises into muscles

ExerciseMuscleMap maps exercise names to the muscles they work, so you can go straight from a logged workout to a shaded body:

let result = ExerciseMuscleMap.workoutIntensities(
    exercises: [("Bench Press", 4), ("Squat", 5), ("Barbell Row", 3)]
)

MuscleBody3DView(intensities: result.intensities)

Volume drives the shading: more sets on a muscle means a stronger tint.

Names are matched loosely, so "bench press", "Bench Press" and " BENCH PRESS " all resolve to the same thing.

Exercises it doesn't know

The table is deliberately finite, and it tells you what it couldn't place rather than guessing:

let result = ExerciseMuscleMap.workoutIntensities(
    exercises: [("Bench Press", 4), ("Zercher Good Morning", 3)]
)

result.unresolved   // ["Zercher Good Morning"]

Resolve those however you like - a prompt, your own table, a model - and pass them back in:

ExerciseMuscleMap.workoutIntensities(
    exercises: exercises,
    resolved: ["Zercher Good Morning": [.hamstrings: 1.0, .lowerBack: 0.7]]
)

In Cinder this fallback is a local LLM that classifies unknown names in one batched call and caches the answer forever. That model isn't part of this package - the hook is, so you can plug in whatever you already have.

Muscle groups

Seventeen, in gym vocabulary rather than medical nomenclature, because that's how lifters log:

chest · frontDelts · sideDelts · rearDelts · biceps · triceps forearms · traps · lats · upperBack · lowerBack · abs · obliques glutes · quads · hamstrings · calves

Every case has a displayName fit for a label.

How the body is made

It's one continuous mesh, not seventeen separate props.

The MakeHuman base mesh (CC0) is morphed toward an athletic build, normalized to a fixed height and orientation, and then every vertex is classified into one of the seventeen groups using limb-axis frames. That gets baked to a compact binary blob - position, normal, muscle id, and a blend weight per vertex - which ships in the package.

At runtime SceneKit tints by muscle id. Recoloring swaps a vertex-color source and nothing else, so switching muscles costs no geometry work.

Two consequences worth knowing:

• The order of MuscleGroup.allCases is the mesh's id order. Reordering the enum silently repaints the body onto the wrong muscles.
• Blend weights soften the seams, so neighbouring groups fade into each other instead of ending at a hard edge.

Tools/bake_body.py is the baker, kept in the repo so the mesh is reproducible rather than a mystery binary.

Demo

The showcase in the screenshot lives in HaploUI under Muscle Map - presets, an intensity slider, and a toggle per group.

License

MIT for the code. The bundled mesh derives from MakeHuman's CC0 base mesh; see LICENSE.

More from Haplo

More open source from Haplo, built by the same solo developer:

• ThinkingOrbs: dotted, honestly-3D loading indicators for AI and agent interfaces in SwiftUI. Nine hand-tuned designs, two tuned sizes.
• Minted: turn any SVG into a physically-lit 3D gold medallion for SwiftUI.
• Vesalius: a map of your codebase for Claude Code, an on-device code graph with an MCP server, a CLI, a Mac app and a VS Code extension.
• ProcessingField: a SwiftUI progress field, a grid of marks whose sizes describe one soft shape that drifts, folds and breathes while your app is busy.
• Gepetto: a browser agent for iOS and macOS that a person can take over at any time. Real WKWebView, small on-device models, a native SwiftUI panel.
• ClayGlobe: a claymorphic 3D globe for SwiftUI that plants a waving flag on every country you visit.
• Mirage: one-stop on-device diffusion image-generation engine for iOS / macOS / visionOS.
• SiriGlass: the Siri drop from iOS 27, for SwiftUI. Liquid Glass that pours out of the Dynamic Island and lights up with your voice.
• Scrubbers: twelve sliders you will want to touch, for SwiftUI. One view, one enum.
• KoiPond: a living koi pond for SwiftUI. Koi glide over any view, lily pads drift on top, and a Metal shader ripples and refracts it all.