Latent Batcher

A small, single-purpose node: take a latent and produce a batch of batch_size copies of it. Conceptually it's RES4LYF's version of what ComfyUI's stock "Repeat Latent Batch" node does - you use it when you want several sampling passes to start from the same base but diverge from there, most often because you're about to feed the batch into something that gives each item different treatment (a noise generator with different seeds per item, a set of different conditioning zones, or simply running several seed variations off one setup in a single queue instead of one at a time).

It's a building block more than a destination - on its own it just duplicates, it doesn't introduce any variation itself. The divergence between batch items comes from whatever you wire in after it.

Inputs and output

• latent (LATENT) - required, the source latent to duplicate.
• batch_size (INT, default 0) - how many copies to produce.
• Output: latent_batch (LATENT) - the resulting batch.

Given the widget's range technically allows negative values, treat that as an artifact of the parameter's slider bounds rather than a real option - set batch_size to whatever positive count of copies you actually want.

Installing it

Comes with the full RES4LYF pack:

• ComfyUI Manager - search "RES4LYF," install, restart.
• Manual - cd ComfyUI/custom_nodes && git clone https://github.com/ClownsharkBatwing/RES4LYF, then cd RES4LYF && pip install -r requirements.txt, restart.

Nothing else to install for this node.

Common issues

batch_size left at the default. It defaults to 0, which won't give you what you want - set it explicitly before running, rather than assuming a sane default batch is already in place.

Every item in the batch looks identical. That's expected - this node only duplicates the latent itself. If you wanted variation across the batch (different seeds, different noise, different conditioning per item), that has to come from a node downstream that treats each batch item differently; this node's job ends at producing identical copies.

VRAM spikes or slow runs after batching. A larger batch means more simultaneous work for the sampler and the VAE decode afterward. If you bump batch_size up and things slow down or run out of memory, that's the batch size doing exactly what it says - scale it back, or process in smaller batches if your hardware can't hold the whole set at once.