Crop Image Into Even Pieces (FS)

Does exactly what it says: give it an image, a row count, and a column count, and it slices the image into that many even pieces. The README description is one line - "this node will cut an image into multiple pieces of rows and columns" - because there isn't much more to it than that.

Why you'd reach for it

Anywhere you need one image split into a grid of equal sub-images: prepping tiles for a per-region touch-up pass, breaking a contact sheet or sprite sheet apart into individual frames, or feeding a grid of crops into something downstream that processes each piece separately. It's the crop-and-split counterpart to "Image Region To Mask" in this same pack - that node marks one grid cell as a mask without cutting anything; this one actually cuts all the cells out as separate images.

It's also a reasonable manual stand-in for tiled upscaling, if you don't want to reach for a full tiling pack: split a large image into even pieces, run each piece through a detail pass, and reassemble afterward. That's a heavier lift than a purpose-built tiled-upscale node handles automatically, but for a one-off job it's a lot less setup than installing another pack just to tile one image.

The inputs and outputs that matter

• image (IMAGE) - the source you're slicing.
• rows (INT, default 3, range 1–32) - how many horizontal slices.
• columns (INT, default 1, range 1–32) - how many vertical slices.

Output: IMAGE - the pieces, as a batch.

Worth noting up front: the defaults are rows: 3, columns: 1, which cuts the image into three horizontal strips, not a symmetric grid. If you actually want a grid - say 3×3 - you need to set both values yourself; the defaults won't get you there.

Installing it

ComfyUI Manager: search "Comfy Fit Size" or "fitsize." Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/bronkula/comfyui-fitsize.git

Restart ComfyUI. No models, no heavy dependencies - small pack, quick install. Lives under Fitsize → Image.

Common issues

The rows/columns defaults are the thing most likely to catch you off guard - dropping this node expecting a grid and getting three horizontal strips instead, because columns defaults to 1. Set both explicitly for whatever split you actually want.

Beyond that, the usual output-batch expectation applies: with rows × columns pieces coming out as one IMAGE batch, downstream nodes that expect a single image (like a straightforward Save Image without batch handling) will need a batch-aware node, or a Pick Image From Batch node from this same pack, to grab or iterate over individual pieces. There's no community-reported gotcha specific to this node beyond that - it's a small, low-traffic utility with a straightforward job.