[Release] MagicNodes - clean, stable renders in ComfyUI (free & open)

Hey folks 👋

I’ve spent almost a year for research and code, the past few months refining a ComfyUI pipeline so you can get clean, detailed renders out of the box on SDXL like models - no node spaghetti, no endless parameter tweaking.

It’s finally here: MagicNodes - open, free, and ready to play with.

At its core, MagicNodes is a set of custom nodes and presets that cut off unnecessary noise (the kind that causes weird artifacts), stabilize detail without that over-processed look, and upscale intelligently so things stay crisp where they should and smooth where it matters.

You don’t need to be a pipeline wizard to use it, just drop the folder into ComfyUI/custom_nodes/, load a preset, and hit run.

Setup steps and dependencies are explained in the README if you need them.

It’s built for everyone who wants great visuals fast: artists, devs, marketers, or anyone who’s tired of manually untangling graphs.

What you get is straightforward: clean results, reproducible outputs, and a few presets for portraits, product shots, and full scenes.

The best part? It’s free - because good visual quality shouldn’t depend on how technical you are.

I’ll keep adding tuned style profiles (cinematic, glossy, game-art) and refining performance.

If you give it a try, I’d love to see your results - drop them below or star the repo to support the next update.

✨ Grab it, test it, break it, improve it - and tell me what you think.

p.s.: To work, you definitely need to install SageAttention v.2.2.0, version v.1.0.6 is not suitable for pipeline. Please read the README.

• The pipeline is designed for good hardware (tested on RTX5090 (32Gb) and RAM 128Gb), try to keep the starting latency very small, because there is an upscale at the steps and you risk getting errors if you push up the starting values.

• start latent ~ 672x944 -> final ~ 3688x5192 across 4 steps.

• Notes

  • Lowering the starting latent (e.g., 512x768) or lower, reduces both VRAM and RAM.

  • Disabling hi-res depth/edges (ControlFusion) reduces peaks. (not recommended!)

  • Depth weights add a bit of RAM on load; models live under depth-anything/.
    
    DOWNLOAD HERE:
    https://github.com/1dZb1/MagicNodes
    DD32/MagicNodes · Hugging Face
    

Workflow.
MagicNodes\workflows\mg_SuperSimple-Workflow.json

MagicNodes\workflows\mg_Easy-Workflow.json