A local Windows desktop application for upscaling, frame-interpolating, and re-encoding anime episodes with first-class preservation of subtitles, chapters, attachments, fonts, and metadata.
- 100% local. No cloud inference, no telemetry.
- Optimized for NVIDIA RTX 3000-series and newer (CUDA / NCNN-Vulkan) with support for all GPUs as well as CPU for video processing and inference.
- Anime-tuned defaults (Real-CUGAN, Real-ESRGAN anime, RIFE).
- MKV-first, with mkvmerge/mkvpropedit for advanced Matroska preservation.
- Stage-based, resumable, deterministic pipeline (probe → plan → scene-detect → batched decode/upscale/interpolate/postprocess/encode → mux → validate).
Current release: 1.0.0-beta4. See CHANGELOG.md for what's in the
box and known limitations.
The main branch tracks post-release development after 1.0.0-beta4.
Download the latest AEP-Setup-1.0.0-beta4.exe from the
GitHub Releases page and run
it. The installer is small (~75 MB); on first launch the GUI offers to download and
verify the pinned third-party binaries (FFmpeg, MKVToolNix, NCNN-Vulkan tools — about
2 GB total) into %LOCALAPPDATA%\AEP\tools\.
py -3.12 -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -e ".[dev]"
python scripts/fetch_tools.py # downloads pinned ffmpeg/mkvtoolnix/etc. (~2 GB)
aep-guiTo build the installer locally, see BUILDING.md.
Anime4KCPP models list: https://github.com/TianZerL/Anime4KCPP/wiki/Model
GPL-3.0-or-later. See LICENSE.
Bundled binaries (FFmpeg, MKVToolNix, NCNN-Vulkan tools, RIFE/CUGAN/ESRGAN models) ship
under their own licenses; see THIRD_PARTY_NOTICES.md.