RC LR/ACR Preset

If you've ever spent money on Lightroom presets - or, more likely, hoarded free ones - you know they're just recipes: a list of exposure, contrast, tone-curve, HSL and split-toning values that a developer applies to a photo. RC_LRPreset is the node that reads those recipes and applies them inside ComfyUI. Point it at a folder of .lrtemplate (Lightroom) or .xmp (Camera Raw) preset files, preview them on your actual image, pick one, and grade. No Lightroom install required.

It's part of the RC Image Compositor pack, under RC/Adjustments, and it's the kind of "I didn't know a node could do that" feature that this pack keeps surprising people with.

How it works

You give it an image and a preset_directory - a root folder that it scans recursively for .xmp and .lrtemplate files. A frontend browser widget lists the presets and renders real-time previews on your image, so you can flip through looks without running the workflow over and over. When you pick one, it applies the full grading pipeline the presets describe: exposure, contrast, highlights/shadows, tone curves, HSL adjustments, split toning, and vibrance.

The inputs you'll set:

• preset_directory - the folder to scan. It handles subfolders, and the path resolver is forgiving: it strips quotes, expands ~, and even maps Windows-style C:\... paths to /mnt/c/... if you're running ComfyUI under WSL. Nice touch.
• strength (0–100) - how hard the preset hits. 0 = no effect, 100 = full recipe. The preview uses the same value, so what you see is what you get.
• pass_forward - the interaction gate. The tooltip says it: run once with it blocked to pick a preset, then enable it to continue the workflow. Same two-phase pattern as the pack's interactive canvas.

Output is a single IMAGE, same size as the input.

Installing it

ComfyUI Manager → search "RC Image Compositor" → install, restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/kj863257/ComfyUI_RC_Image_Compositor
cd ComfyUI_RC_Image_Compositor
pip install -r requirements.txt

Dependencies are just pillow, numpy, opencv-python. The preset files are data, not models - nothing to download, you just need a directory of presets you already own.

Where people get burned

• The pass_forward dance trips everyone first time. The node blocks execution until you've picked a preset with pass_forward off, then you flip it on and run again. If the node "does nothing" or errors, that's the design talking.
• Presets are only as good as their files. .lrtemplate is the older Lightroom format; .xmp is the modern Camera Raw one. A preset built for a specific camera's raw processing may do odd things to a generated PNG - presets assume photographic tonal ranges, not diffusion output. Expect to dial strength down for AI images.
• Directory is a real filesystem path. It's not a ComfyUI dropbox picker; paste an absolute path. The resolver helps with quotes and WSL, but it can't read your mind about which folder holds the presets.

If you already own a preset library, this node is a genuinely fun bridge between the Lightroom world and the ComfyUI graph. Just don't skip the preview step - generated images break preset assumptions, and the preview is exactly what saves you from a full workflow run of a bad grade.