🔳 Tile (jov)

Tiling is how you turn one small image into a wall of copies. A texture repeated into a backdrop, a logo stamped into a grid, a pattern you need to preview - Tile (jov) was Jovimetrix's dedicated node for exactly that, and its description says it in one line: "Tile an Image with optional crop to original image size." This is the older lowercase build; it's been superseded, but the idea lives on in the current pack's TRANSFORM node.

Usage is almost aggressively simple. pixels is the required input - any image or mask. tileX and tileY (both INTs, default 1, minimum 0) say how many copies to lay out horizontally and vertically. Set tileX=3 and tileY=2 and you get a 3×2 grid of the source image, edge to edge. The "optional crop to original image size" in the description is the safety hatch: when you don't want the tiled result ballooning past the source dimensions, the node can bring it back down - which is what makes the output usable as a seamless background instead of a runaway mosaic.

Outputs are image and mask, both straight from the tiled result, so a tiled mask is as easy as a tiled image. That's the quiet use case: tile a small mask into a full-canvas repeating selection without hand-drawing anything.

Where this actually showed up in workflows: making checkered or patterned backgrounds (tile a single square), turning a small texture into a fill, and generating a repeating alpha for effects. And because everything in this pack is batch-friendly, tiling applied to a whole batch of images in one pass.

Now the honest part. This is a legacy node from an older Jovimetrix build - the lowercase "(jov)" naming gives it away - and it's not in the current pack source. Its job got absorbed: the maintained TRANSFORM (JOV) 🏝️ node now includes a TILE control that wraps and repeats an image within the canvas (with mirror modes and edge wrapping on top), so you can do everything Tile did and more in one place. If an old workflow reports "Tile (jov)" missing, swap in TRANSFORM and set its tile axis. The dedicated Tile node also walked away from the pack's name churn: Jovimetrix v2.0 split a bunch of node families into separate Jovi_* packages, so legacy names surface in old workflow JSONs regularly.

If you want to run it, you install the whole pack:

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

or ComfyUI Manager → "Jovimetrix", then restart. The dependencies are the shared set - numpy, opencv-contrib-python, Pillow, matplotlib, the cozy-comfyui helpers - no model downloads. Keep ComfyUI at 0.1.3+ with a current frontend, and set expectations: this one's a museum piece, but the tiling trick it taught the pack is still very much alive.