Load and edit images

ComfyUI's built-in Load Image is fine for grabbing a file. LF_LoadAndEditImages replaces it with something closer to a mini photo editor: pick an image from a file-system tree inside the node, and do real edits - inpainting, brightness, contrast, saturation - right there, before the image ever reaches the sampler. It's the pack's flagship IO node, and the README calls it out as a headline feature.

The workflow it unlocks: a source image that needs a touch-up before img2img or inpainting. Instead of the dance of "edit in Photoshop, save, reload in ComfyUI," you load once, fix what needs fixing, and hand the edited result downstream. If you work with reference images that need constant tweaking - brightness on a product shot, removing a logo before re-roll - this kills a whole class of round-trips.

How it works

The node is essentially all UI: its KUL_IMAGE_EDITOR widget is a file-tree browser plus an editing canvas. You navigate to the images, select one, and edit it with the built-in tools (the same editor shared with the pack's other image-edit nodes, like the mid-workflow breakpoint). The editing is interactive - you're not setting parameters, you're using the canvas.

It does have an internal detail worth knowing: how edits flow into the rest of the graph. The static node schema here is thin (the widget is the real interface), and in practice the node feeds the edited image into the workflow for further processing - that's the "send them downstream to other nodes" behavior the README describes. So think of it as a Load Image that happens to have Photoshop stapled to the front.

Installing it

Part of LF Nodes:

cd ComfyUI/custom_nodes
git clone https://github.com/lucafoscili/comfyui-lf

Restart ComfyUI, or search LF Nodes in ComfyUI Manager. No model downloads; editing tools run on the pack's light dependencies (numpy, Pillow, opencv-python).

Gotchas

The editor is custom frontend, which means it's the most fragile part of the pack. Canvas not rendering? Frontend problem - update via Manager and restart clean before blaming the node. And because the editing is manual, an unattended batch is out of the question for this node; it's a human-in-the-loop tool by design.

The pack note applies double here: comfyui-lf is legacy, frozen at Feb 2025, and the image editor got a substantial rebuild in the successor lucafoscili/lf-nodes. If you're new and choosing between the two, the new repo is where the editor is getting the attention.