FL Prompt Multi

You've got a list of prompts and you want to render all of them in one queue, each with its own positive and negative. FL_PromptMulti is the node that sets that up. Type your positive prompts one per line, your negatives one per line, and it turns each into a parallel list that a batch-processing workflow can iterate over. One line, one generation.

It's a small quality-of-life node for anyone doing batch prompt runs - testing a set of ideas, generating a themed series, building a dataset. Instead of duplicating your sampler chain or re-typing prompts one at a time, you write them all in a text box and let the list machinery fan them out.

How it works

It reads your positive and negative text boxes line by line and produces three lists: positive prompts, negative prompts, and generated names. Crucially, it balances the counts - if you have five positives and only two negatives, it repeats the last negative (or pads with empties) so every positive has a matching negative and nothing falls off the end. It also mints an indexed name per prompt (prompt_0, prompt_1, …) using your prefix, which is handy for naming saved files.

The output being lists is the whole point: ComfyUI runs the downstream graph once per list item, so wiring these into your CLIP-encode → sampler chain generates one image per line automatically.

The inputs and outputs

• positive_text (multiline) - one positive prompt per line.
• negative_text (multiline) - one negative prompt per line.
• name_prefix (optional, default prompt_) - the prefix for the generated per-item names.

Outputs, all lists: positive, negative, and name. Feed positive/negative into your text-encode nodes and use name for filename templating if you like.

How to install it

Part of filliptm's Fill-Nodes pack. ComfyUI Manager → search ComfyUI_Fill-Nodes → Install → restart, or:

cd ComfyUI/custom_nodes
git clone https://github.com/filliptm/ComfyUI_Fill-Nodes
pip install -r ComfyUI_Fill-Nodes/requirements.txt

No model or special dependencies.

Common issues & troubleshooting

Only one image generated, not one per line. The outputs are lists, and the rest of your graph has to actually iterate over them. If a downstream node isn't list-aware or you've collapsed the list somewhere, you'll get a single run. Keep the list flowing through the CLIP encode and into the sampler - ComfyUI's list execution is what fans it out.

My negatives got repeated unexpectedly. That's the auto-balancing. When the positive and negative line counts don't match, the last negative repeats (or empties pad) to keep the lists the same length. It's intentional, but it surprises people - if you want a specific negative per positive, give them equal line counts.

Blank lines create blank prompts. An empty line in the box is still a line. Trailing blank lines can produce an extra empty-prompt generation; trim them if you're getting one more output than you expected.

It doesn't encode anything. This node only produces the text lists - it's not a CLIP encoder. You still wire positive/negative into your normal text-encode nodes. Think of it as a prompt fan-out, not a replacement for encoding.