Generate sequential seeds

If you've ever wanted to batch out a handful of deterministic variations without wiring up a separate seed widget for each one, LF_SequentialSeedsGenerator does the counting for you. Give it one starting seed, and it outputs twenty sequential seed values - your seed, then seed+1, seed+2, and so on, all the way up - as twenty individually wireable outputs. It's the same "seed node that remembers what it did" idea rgthree's popular Seed node covers, just scaled out to a batch instead of a single remembered value.

The mechanism is genuinely that simple: pick a starting integer, get back twenty offsets from it. There's no randomness involved - same starting seed always produces the same twenty outputs, which is the point if you're doing controlled comparisons (say, twenty KSamplers each getting one seed, so you can eyeball a grid of variations from a single number). enable_history (on by default) logs the seeds generated at each execution into the node's built-in history widget, so you've got a record of what's already been run without writing it down yourself.

Be clear-eyed about the shape of this: the outputs are fixed at exactly twenty, all named seed, distinguished only by which output slot you drag a wire from. If you need more than twenty, you'll need a second node (or a different starting seed to avoid overlap); if you need fewer, just ignore the extra outputs - nothing forces you to wire all of them.

Installing it. ComfyUI Manager: search "LF Nodes," install, restart. Manual: cd ComfyUI/custom_nodes && git clone https://github.com/lucafoscili/lf-nodes, then restart. Use that repo specifically - the one linked from comfy.icu, comfyui-lf, is the pack's original home and has been archived since the author moved development to lf-nodes. No models, no heavy dependencies - pure integer math, nothing that touches your GPU.

Where people get tripped up. The main confusion is expecting the twenty outputs to be labeled distinctly (seed_1, seed_2, etc.) - they're not; the schema names every single one just seed, so you have to track which output index maps to which offset yourself, usually by connecting them in order and trusting the sequence. Second, don't confuse this with randomization - it's purely additive from your starting point, so if you want twenty different random seeds rather than twenty sequential ones, this isn't the node for that job. And if the history widget looks empty on a fresh install, that's expected - it only logs seeds from executions that have actually happened in this session.