Anima Image Grid

Here's the honest pitch: AnimaImageGrid is the least interesting node in the Anima LoRA Comparison pack, and it's also the one that makes the pack usable. The AnimaXYSampler hands you a batch of images - one per LoRA, all the same prompt and seed - and if you look at that batch in the preview you get a slide deck, not a comparison. This node flattens the whole batch into a single contact sheet so you can actually see "LoRA A beat LoRA B" at a glance. Nobody brags about their grid node. Everybody needs one.

How it works

The mechanism is dead simple and, usefully, it's honest about it. The node takes the incoming IMAGE (a batch, i.e. any number of frames), builds a canvas sized to the widest and tallest image in the set, and pastes each frame in order, centered. Set a Gap and it leaves that many pixels of breathing room between frames, filled with whatever Color you chose. One frame in, one out - it leaves a single image alone and returns it unchanged rather than padding it into a weird one-cell grid.

The inputs

The four inputs are the whole story:

• Image - the IMAGE batch from the pack's AnimaXYSampler (or anything else that hands you multiple frames).
• Direction - Horizontal lays frames left to right, Vertical stacks them top to bottom.
• Gap - 0–256 pixels between frames. Default 0; a small gap (say 8–16) reads much better than touching frames.
• Color - the gap (and any letterboxing) fills in Black, White, Gray, Red, Green, or Blue.

The single output IMAGE is one picture you can preview or save like any other - which is the quiet trick here: a comparison you can drag back into ComfyUI or share on CivitAI as one file. Order follows the batch, so the first LoRA in your list is the leftmost (or topmost) cell. Keep your list order in mind when you're reading results, because the grid won't label anything for you.

Common confusion

Where people get confused: the Color is also the letterbox color. The canvas is sized to the largest frame and everything is centered, so if your batch mixes resolutions (Anima is fine from 512² to 1536², and a comparison list can easily span sizes if one LoRA pulled a different latent through), the smaller frames get colored bars on one or two sides. That's not a bug - it's the node telling you your frames don't match. If you want a clean grid, feed it a batch that's all one size. And a single image passes straight through, so don't panic if the node looks like it "did nothing" on a one-frame input.

It's also worth knowing what this node is not: it's a plain array, not an annotated comparison tool - no labels, no captions, no per-cell text. The genuinely useful thing it does is turn twenty separate generations into one image you can actually evaluate, which is the entire reason the pack's sampler outputs a batch instead of a folder. It's the finishing move, and for that job it's exactly right.

Install

Same install as the rest of the pack - ComfyUI Manager (search "Anima LoRA XY") or:

cd ComfyUI/custom_nodes
git clone https://github.com/yunqiankuangyu/comfyui-anima-lora-comparison.git

Restart ComfyUI and it sits under Anima/LoRA Comparison, ready to catch everything the sampler produces.