Image Batch Combiner

Image Batch Combiner joins up to eight image batches into a single batch, along the frame axis. The mental model is concatenation, not compositing: if batch1 has 10 frames and batch2 has 5, you get one batch of 15 frames, in order. It's the "make one long animation from several shorter pieces" node.

Where this earns its place is in workflow assembly. Video and animation work rarely produces one continuous batch - you'll have a batch from one sampler, a batch from another, a couple of frames you generated separately. Rather than regenerate as one big batch, you produce the pieces and staple them together. The combiner is also the natural counterpart to the pack's own Image Batch Splitter: split a long sequence, do something to the interesting chunk, combine everything back. Round-trip media surgery.

The inputs

batch1 is required - you always need at least one batch. batch2 through batch8 are optional; the node just skips whatever's left disconnected. All connected batches must have identical height, width, and channels - the node checks and raises a clear error if you mix a 512×512 batch with a 768×768 one. That constraint is real and you'll hit it. If the sizes differ, fix the mismatch upstream (resize or crop) before combining.

Outputs

• combined_batch - the IMAGE output: all frames in order.
• info - a STRING of JSON describing what happened: total_batches, total_frames, and a batch_info array giving each source batch its start_frame and end_frame in the combined result. If you ever need to know where batch2 landed inside the 15-frame result, this is the map.

The two outputs work together nicely: wire combined_batch into a GIF maker or video save, and wire info into a text display if you want proof of the layout.

Installing it

ComfyBros installs like any custom node pack. ComfyUI Manager - search "ComfyBros" - or:

cd ComfyUI/custom_nodes
git clone https://github.com/turnbros/ComfyBros

Then restart ComfyUI. The pack's README is stale (no requirements.txt actually exists; the deps are in pyproject.toml). The node lives under ComfyBros/Image Utils.

Gotchas

Dimension mismatch is the failure you'll actually see, and it's not a bug - the error message tells you which batch doesn't match the reference. One quirk worth knowing: if only batch1 is connected, the node skips the whole combine path and just returns batch1 unchanged (with its info). So if you're debugging a "combiner did nothing" moment, check that you actually plugged in a second batch. Order matters too - frames are appended in input order, batch1 first, so wire them in the order you want the animation to play.