Simswap Image Dublicator (List)

Simswap Image Dublicator is about as simple as a node gets: you give it an image, it gives you back a list of N copies of that image. The whole implementation is a list comprehension. That's it.

So why does it exist? Because the Simswap swap node wants an IMAGE batch on its input_image, and sometimes what you have is a single image you need to line up against a batch of targets. The classic case: you've got a video's worth of frames and one swap source - but you want the target frames to drive the batch, and the source needs to be repeated to match. Drop a Dublicator on the source side, set count to match the target batch size, and the swap processes them pairwise instead of stalling on a shape mismatch.

There's also a second, lazier use: debugging batch flows. When your swap node is silently doing nothing because a single image slipped into a list slot, duplicating it out to the right length is a fast way to find out whether the problem is the batch count or the swap itself.

Inputs and output

• image - any IMAGE.
• count - how many copies to make. Default 1, minimum 0 (a zero-count output is an empty list, which is mostly useful for proving to yourself that you wired something wrong).
• IMAGES - the output, an IMAGE list (is_list: true). Note it's a list, not a batch tensor - wire it into inputs that expect a list of images.

One subtlety worth knowing: the swap code caches face analysis by MD5 hash of the image. Feed it five identical copies and it analyzes the face once and reuses the result, so duplicating isn't a five-times performance hit - it's just plumbing to make the shapes line up. That's the intended use: identical copies, not five different things you pasted together.

Installing it

Same as the rest of the pack:

cd ComfyUI/custom_nodes
git clone https://github.com/TaiTair/comfyui-simswap
# or: ComfyUI Manager → search "simswap" → install
# then restart ComfyUI

No models, no extra dependencies - the pack's install.py handles the shared requirements on first start (insightface==0.7.3, onnxruntime, onnx, opencv-python, numpy, segment_anything).

Gotchas

• Don't use it to "pre-batch" a bunch of different images. That's not what it does; for different images you'd batch them normally. This node is strictly for repeating one image.
• The typo is real - "Dublicator" with a b. It's in the class name, the display name, and the comfy.icu URL. Search it with the typo or you won't find it.
• It's a copy of the same hacky, unmaintained port as every node in this pack. Trivial enough that "unmaintained" barely matters here - a list comprehension isn't going to rot.