Termux-Diffusion Logo

Termux-Diffusion

v1.1.1 (Dual Engine)
PyPI (Python) npm (Node.js) 💖 Sponsor GitHub Repository

Version Archive & Changelog

Historical release logs and upgrade migration guides.

v1.1.1 - Production High-Precision & Stability Release

2026-08-20
  • TOP 7 High-Precision Advanced Parameters: 10 Samplers (euler_a, dpm++2m, lcm, etc.), 6 Schedulers (karras, exponential, ays, etc.), VAE Tiling (reduces peak RAM by ~70%), Img2Img (init_img, strength), LoRA adapter weights, CLIP-Skip, ControlNet, and TAESD support.
  • Fail-Fast Missing File Validation: Instant validation on missing files (init_img, control_net, etc.) before triggering 1.5GB model downloads.
  • Async Cancellation & Process Reaping: Full asynchronous cancellation support (async_generate() in Python, AbortSignal in Node.js) with POSIX process group tree termination (_safe_kill_process).
  • Strict Hardware Acceleration: GPU/Vulkan backend fails fast if drivers are missing, preventing deceptive fallbacks.
  • Samsung RAM Plus & Low-Memory Guidance: Removed library-level blocking OOM aborts; added comprehensive virtual swap / RAM Plus configuration guide in README.

v1.0.0 - Genesis Release

2026-08-20
  • Dual-Engine Architecture: Full feature parity across native Python (PyPI) and Node.js / TypeScript (npm).
  • Zero PRoot Execution: Native ARM64 Bionic libc compilation with -O3 NEON SIMD vectorization.
  • Smart Model Hub: 5 built-in presets (realistic, speed, sdxs, turbo, anime) with streaming auto-download.
  • Samsung MediaStore & WakeLock: Automatic Samsung Gallery synchronization and Android CPU WakeLock management.