Skip to content
 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

307 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

EDMG Studio Monolith Repo

This repository is the single authoritative source tree for EDMG Studio.

Studio is not a sidecar anymore. The desktop shell, React frontend, FastAPI backend, vendored EDMG engine packages, setup flow, release packaging, and support tooling all converge on one product path:

  • studio/edmg-studio/

The older standalone-engine installers, duplicate Electron shell, and extra README entrypoints have been retired so the repo presents one product instead of multiple competing workflows.

The legacy standalone web UI prototypes have also been retired. Their planning and audio-reactive capabilities now live inside the Studio app workbenches:

  • studio/edmg-studio/src/workbenches/AiNlpWorkbench.tsx
  • studio/edmg-studio/src/workbenches/AudioReactiveWorkbench.tsx

Canonical launch

Windows:

RUN_ME.bat

macOS/Linux:

./run_me.sh

Those launchers open the Studio dev launcher in studio/edmg-studio/tools/launcher_gui.py, which keeps the UI, backend, and Studio Home storage aligned with the same settings used by the packaged app.

Canonical product layout

  • studio/edmg-studio/ Electron shell, preload, main-process runtime, React/Vite frontend, packaging scripts, release validation, and the only Node/pnpm package root in the repo.
  • studio/edmg-studio/python_backend/ FastAPI backend plus the vendored enhanced_deforum_music_generator and deforum_music engine packages that power planning, analysis, schedules, and EDMG Core support.
  • studio/edmg-studio/tools/launcher_gui.py Shared dev launcher and Studio Home bootstrap flow.
  • studio/edmg-studio/packaging/windows/ Windows-first release orchestration for the packaged Studio app.

Studio Home

Studio separates the app install directory from the heavy runtime storage root. The Studio Home contains:

  • data
  • models
  • cache
  • logs
  • external
  • electron

That keeps large downloads, render caches, and external tools off the app install path and allows migration to another drive or mount such as D:\ on Windows or /mnt/media/EDMG-Studio on Linux.

JS Tooling

  • Run all JS/Electron commands from studio/edmg-studio/.
  • Use Node.js 20.19+ or 22.12+; Node 22 LTS is pinned in studio/edmg-studio/.node-version.
  • The canonical package manager is pnpm@10.33.0, pinned in studio/edmg-studio/package.json.
  • The shipped desktop app version also comes from studio/edmg-studio/package.json#version.

Python tooling

  • Python is pinned to 3.12 in .python-version.
  • uv 0.11.28 manages Python acquisition, the backend environment, locking, tests, linting, and PyInstaller builds.
  • studio/edmg-studio/python_backend/uv.lock is committed release input.
  • Select exactly one accelerator extra: cpu, directml, or cuda; compose it with capability extras such as audio, asr, and internal-video.
  • Packaged Electron applications include the PyInstaller backend and do not require end users to install Python or uv.

See docs/PYTHON_TOOLCHAIN.md for commands and the lock-update policy.

Release, strategy, and operator docs

Test strategy

  • Repo-level tests live under tests/; backend tests live under studio/edmg-studio/python_backend/.
  • Run both scopes from the repo root with uv run --project studio/edmg-studio/python_backend --frozen --extra cpu --extra core --extra audio --group test python scripts/run_pytest_scopes.py.
  • The runner checks the committed lock and performs a frozen CPU-profile sync before executing either scope.

Recommended Local Stack

  • Planner default: NVIDIA Nemotron Ultra via nemotron_cloud (NIM)
  • Local Ollama planner: nemotron-3-ultra:cloud or low-resource qwen3:4b
  • OpenAI-compatible default model: nvidia/llama-3.1-nemotron-ultra-253b-v1
  • Broad still-image default: SDXL Base 1.0
  • Fast still-image option: SD3.5 Large Turbo
  • Reference still guidance: SD3.5 ControlNet Blur, Canny, and Depth
  • Primary HF video backend: Wan2.2 TI2V 5B
  • Short image-to-video fallback: SVD XT Img2Vid

Hardware Tiers

  • Low-spec: qwen3:4b (Ollama) + SDXL Base 1.0
  • Mid-range: Nemotron cloud or qwen3:8b + SDXL Base 1.0 + SD3.5 Large Turbo + SD3.5 Blur/Canny
  • High-end: Nemotron cloud + SDXL Base 1.0 + SD3.5 Large Turbo + SD3.5 Blur/Canny/Depth + Wan2.2 TI2V 5B

Compatibility shims

  • Repo-root sitecustomize.py and the repo-root librosa/ package are source-tree compatibility shims for development and tests.
  • Packaged/backend install flows rely on the declared Python dependencies and do not package those repo-root shims.

Unreal bridge status

  • Studio-side bridge: usable. The repo can preview, export, build an Unreal import plan, and import returned renders back into canonical project outputs.
  • Unreal-side runtime integration: partial. There is no verified in-editor smoke test on this machine, no packaged Unreal plugin/module, no live OSC/WebSocket/Remote Control execution path, no one-click Unreal render job launcher, and no deeper Sequencer build beyond the first importer pass.

Compatibility note

The repo root now keeps only thin launch aliases and monolith-level docs. The runtime code, packaging logic, backend engine packages, and operator tooling live under studio/edmg-studio/.

About

No description, website, or topics provided.

Resources

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages