Skip to content

Home

Jump to bottom
The Duskfall Portal Crew edited this page Jun 12, 2026 · 11 revisions

Ktiseos Nyx Trainer

Train, tag, and generate — a full LoRA and checkpoint training suite with ComfyUI built in.

A web interface built on Kohya SS. Prepare datasets, auto-tag, train across many model types (SDXL, Flux, SD3, Lumina, Anima...), then generate and test results in a bundled ComfyUI workspace. Runs locally on Windows/Linux or on cloud GPUs (VastAI, RunPod).

Version Python License
v0.2.1-BETA 3.10 or 3.11 MIT

Confirmed working: SDXL, Illustrious, Pony, NoobAI, Flux, Anima

Quick Start

git clone https://github.com/Ktiseos-Nyx/Ktiseos-Nyx-Trainer.git
cd Ktiseos-Nyx-Trainer
install.bat          # Windows
start_services_local.bat

Then open http://localhost:3000 — full walkthrough on the Getting-Started page.

Documentation

For Users

Page Description
Getting-Started Quick installation, setup, first training run
Installation Full platform-by-platform install (Windows, Linux, VastAI, RunPod)
Configuration Training parameters, presets, TOML system
Training-Guide End-to-end workflow: dataset prep, tagging, captioning, training, model management
Troubleshooting Common errors, diagnostic tools, how to report bugs

For Contributors

Page Description
Architecture System design: FastAPI, Next.js, service layer, Kohya SS integration
Contributing Development process, code standards, testing requirements
Changelog Release history and known issues
Upstream-Reference Summary of vendored sd-scripts and LyCORIS documentation

Features

Dataset preparation:

  • Drag-and-drop upload (individual files, zip archives, URLs)
  • WD14, BLIP, and GIT auto-tagging
  • Tag editor with bulk add/remove/replace, frequency viewer, image gallery

Training:

  • Supported models: SDXL, SD1.5, Flux, SD3/SD3.5, Lumina, Anima, HunyuanImage
  • LoRA types: Standard, LoCon, LoHa, LoKr, DoRA
  • 40+ community presets
  • Live training monitor: step count, loss, learning rate, ETA

Models:

  • Civitai browser: search, filter, download directly
  • HuggingFace upload after training

Image generation (ComfyUI):

  • Bundled ComfyUI workspace installed by default
  • LoRA Manager for browsing and applying LoRAs

Utilities:

  • LoRA merge, resize, metadata editing
  • Checkpoint merge

External Resources

Getting Help

  1. Check Troubleshooting first
  2. Search existing issues
  3. Open a new issue with:
    • Your OS and deployment mode (local/cloud)
    • Error output / logs
    • Output from diagnose.bat (Windows) or ./diagnose.sh (Linux)

Documentation is a work in progress. Something unclear or missing? Open an issue or PR.

Clone this wiki locally