DanceFlow ecosystem map connecting StemLab, DanceRudiments and DanceMoves, with supporting tools KeywordMoves, RasterMoves and PixelCue.

DanceFlow

DanceFlow · open creative tools

From musical evidence to deliberate motion.

DanceFlow is a family of independent tools for understanding music and turning that evidence into motion-aware visual work. The ecosystem also covers audio timelines, search evidence, visual identity analysis, visual-media preparation, cached thumbnails and web metadata. Use one part on its own, or connect the pieces that fit your workflow.

The core path

Listen. Describe. Move.

The tools exchange explicit timing and analysis data rather than hiding everything inside one application. That keeps the evidence inspectable and lets each component remain useful by itself.

StemLab

Understand the track

Local, evidence-led analysis for tempo, beats, structure, stems, harmony, lyrics, MIDI transcription and loopable sections. Model routes and uncertainty stay explicit.

Explore StemLab →

DanceRudiments

Describe the movement

Deterministic rhythmic position functions with a C++ core, musical-time authoring and Python, TypeScript/WebAssembly and C# bindings for creative applications.

Explore DanceRudiments →

DanceMoves

Publish the response

A browser and WordPress runtime for BPM-, cue- and lyric-timed effects, with accessibility and performance fallbacks.

Explore DanceMoves →

Supporting tools

The work around the motion matters too.

These tools support discovery, visual preparation, reusable interfaces and publication. Each has its own job and installation route. Their connections are explicit; sharing an ecosystem does not mean every pair is automatically integrated.

PixelCue

A local visual-media analysis service and desktop browser for tagging images and sampled video or archive content, with selectable models, persistent analysis and request-scoped progress. It also supplies the shared image measurements and optional model interpretations used by PrismMoves.

Explore PixelCue →

ThumbMoves

An independent, cache-only Python library for existing Windows and freedesktop thumbnails, with explicit cache misses. PixelCue uses it to reuse previews before reopening original media.

Explore ThumbMoves →

PrismMoves

Understand an image before you extend its visual language. PrismMoves brings PixelCue measurements of colour, composition and texture together with optional model interpretations, then turns them into readable image profiles and evidence reports. Use it to review artwork and visual references before deciding what to preserve or change.

Measurements, model hypotheses and unknowns stay separate. One image supports an individual profile; recurring identity rules need comparison evidence and visual review.

Original synthetic artwork beside actual PixelCue palette, composition and texture measurements
Actual PixelCue 0.9.0 measurement output from an original synthetic illustration. This example explains the report; it is not a model-accuracy evaluation.

Explore PrismMoves on GitHub →

MediaSlide

A responsive React media browser with coordinated list, details, thumbnail and slide views, responsive selection detail, extensible renderers and bidirectional pagination.

Explore MediaSlide →

WorkflowMoves

The guarded workflow layer for DanceFlow: versioned rules, task guides, portable agent skills, evidence gates and additive project integration. Preview exact route changes, preserve project-owned instructions and verify adoption separately.

Explore WorkflowMoves →

Try and inspect

The useful bit is what you can test.

The source, demonstrations and public results are available separately. A working example is evidence of that example—not a promise that every tool or every track behaves identically.

See a motion-aware EPK

The Arcadians press page is one public example of DanceMoves responding to a track’s structure and pulse.

Open Arcadians →

Working principles

Clear boundaries make better creative tools.

Independent by designInstall one component without adopting the whole workflow.
Evidence stays labelledA model output, observation and verified result are not treated as the same thing.
Motion stays boundedIntermittent audio accents stay within defined subjects or regions. Continuous motion has explicit scene bounds.
Accessible fallbacksReduced motion, keyboard use and readable static states are part of the design.
Reproducible outputsHashes, versions, settings and source evidence travel with the result where practical. Tool improvements are tested, pushed to GitHub and published as new releases.
External services stay explicitLocal and hosted routes are kept distinct, including their costs, permissions and limitations.

Start with the problem in front of you.

If one of the tools is useful, use that one. If the whole path fits, the pieces are designed to pass evidence between them without blurring what each part actually knows.

Get in touch

DanceFlow is a software and creative-tool project. It is separate from the My Songs catalogue.

Source and package coverage

Explore the DanceFlow tools.

Seventeen independently useful components, with the DanceFlow toolkit providing discovery, portable skills and supported adapters. Source, stable release, installed runtime and a working host are checked separately.

Potbot and DanceFlow for Unity publish preview releases. Potbot consuming DanceFlow MCP remains planned. The Unity WebGL example is a deployed build; it does not establish every Editor, GPU, rig or song-alignment route.

Latest DanceFlow toolkit · How the supported handoffs fit together