Roadmap

What's next

GMT is actively developed by one person. This roadmap is a rough sketch — priorities shift with feedback, and the "later" column often gets reshuffled. If something here matters to you, say so on r/GMT_fractals.

In progress

Gallery launch + curated seed content

in-progress

The gallery is live with sign-in and public submissions, but still needs a curated batch of launch-day scenes and a proper announcement. First 100 public submissions skip moderation while the seed grows.

Mobile UI refinement

in-progress

Touch-first panel layouts, bigger hit targets, and a control scheme that respects limited screen space. The engine already runs on mobile — the UI needs the pass.

Coming soon

Mesh export workflow

planned

Custom filenames and save locations, plus a batch queue for rendering many variations overnight. Manufacturing workflows often need 20+ iterations to lock a design.

Mesh export unified pipeline

planned

Route mesh-SDF generation through the same feature-injection system as the main renderer, so new features automatically work in mesh export. Internal refactor — eliminates the parallel shader builder.

N-formula hybrid sequences

planned

Generalize the current two-formula Interlace into a full Mandelbulber-style sequence — any number of formulas layered per iteration with per-formula weights. The real feature gap identified by the hybrid-architecture research.

Smooth scene transitions

planned

Crossfade between saved scenes over time instead of hard-cutting. Interpolates camera, parameters, lighting, and coloring together — built for live performance and cinematic pacing.

Controller input (gamepad / drone)

planned

Map Xbox / PlayStation / flight-stick controllers to camera and parameter control. Analog sticks feel radically better than WASD for fractal exploration.

MIDI input

planned

Map MIDI CC messages to any parameter. Perform fractals live with a hardware controller, or sync to a DAW for music-reactive animation.

Spout output (VJ pipelines)

planned

Stream the viewport directly into VDMX, Resolume, MadMapper, and other VJ tools via Spout. Windows desktop-app feature; Syphon equivalent on macOS.

New formulas, ongoing

ongoing

Continued imports from Fragmentarium and the Distance Estimator Compendium. 326 of 494 library entries currently render cleanly; the catalog keeps growing.

Desktop app (Tauri, Windows-first)

planned

Native desktop shell via WebView2 for Chrome-equivalent performance, offline access, and better GPU control. Also the home for Spout output and any other OS-level integrations. macOS and Linux follow the Windows build.

Exploring / later

Ideas that need more thought, demand, or capacity before committing. No promises — included so you know what's being considered.

AI-assisted formula authoring

.frag → .ts conversion helper + GMF round-trip for LLMs. Turns "I want a formula that looks like X" into a working GMT formula without hand-porting GLSL.

User preferences sync

Classroom / education mode

Structured lesson flows for teaching fractal math visually. Presets that walk through concepts step by step.

Formula marketplace

Let community authors share curated formula + preset sets. Depends on the gallery being mature.

VR / stereoscopic viewing

Browser WebXR is maturing — could add minimal-friction VR mode for exploring from inside.

Recently shipped

Online gallery — browse curated and community-submitted scenes, sign in with email or Google, post your own renders, customise your watermark, manage and re-share your submissions
In-app feedback — bug reports, feature requests, and support questions from the ? menu, with optional one-click scene attachment
Audio in the timeline — drop in a track, scrub with playback in sync, and have animation react to audio analysis (peaks, bands, RMS) via the modulation system
Audio in video exports — exports mux the audio track straight into the MP4, so soundtracked animations are one-click to share
Modulation system — wire LFOs and audio analysis to any parameter; per-rule mute switches and a global off-toggle for clean A/B comparisons
Modulation recording — capture every animated value into keyframes, now reliable over long durations
Canvas-based timeline + graph editor — handles thousands of keyframes without lagging; multi-handle bezier drag across multi-selection; flat tracks dim automatically
Timeline UX polish — resizable sidebar (drag the right edge of the track-name column), shift-range select keeps the anchor row, no more duplicate keyframes from double-clicks
Smoother camera controls — eased Q/E roll and drag, Alt for precision, Shift for boost, consistent across orbit and fly modes
Geometric Orbit Trap — renders correctly on Mandelbulb and MandelTerrain, respects iteration coloring
Boot + shader-compile failures surface in the splash screen instead of a blank canvas
Directional light popup no longer crashes on legacy scenes
Bucket render overhaul — explicit W×H output with FHD–8K / print-size presets and a Columns×Rows tile grid that splits huge exports into separate PNGs beyond single-GPU VRAM limits
Preview Region — click-to-zoom a sub-region at full export pixel density with live parameter interactivity
Image sequences with alpha + depth channels for compositing and VJ pipelines
V3/V4 formula pipeline catalog with auto-pick in the Workshop (494 library entries, 326 currently renderable)
Polyhedral fixes — Rhombic Dodecahedron, Rhombic Triacontahedron, Coxeter, Cuboctahedron now render correctly at all iteration levels
Modular graph editor overhaul — correct undo/redo, live condition sliders, DCE uniform fix
Pixel threshold invariant across adaptive scale, DPR, and SSAA bucket upscale
Worker-only rendering architecture — all dual-path code removed
Adaptive resolution — auto-scales render buffer to maintain target FPS, context-aware
GMF format with AI-friendly embedded shader + scene state
Formula Workshop with Fragmentarium (.frag) + Distance Estimator Compendium imports

Release-by-release details live in the repository's dev changelog.