Seed Logger (Inspire)
Keep a running history of the seeds you actually used
Here's a small pain everyone with a randomizing seed has felt: you generate a batch, one image is perfect, and by the time you go looking for its seed... it's already been randomized away. The number that made the magic is gone. Seed Logger exists to stop that. It quietly records the seeds passing through it and keeps the last several visible right on the node, so when you strike gold you can scroll back and find the seed that did it.
It's not glamorous and it doesn't touch your image at all. It's a notebook. But if you run with control_after_generate set to randomize - which most people do while exploring - it's the difference between "I'll just find that seed" and "well, that's lost forever."
How it works
You wire your seed into it, and each run it appends that seed to an on-node log, trimming the list to the most recent N. Because it's an output node with a text display, the log shows up in the node itself - a live history you can read without digging through metadata. There's no output wire; the whole value is the record it keeps.
The inputs and outputs that matter
seed(INT) - the seed to log. Connect this to the same seed feeding your sampler (convert the sampler's seed widget to an input and split it, or drive both from one seed source) so the logger records exactly what was used.seeds(multiline text) - the log itself: the running list of recent seeds displayed on the node. This is what you read to recover a good seed.limit(default 5) - how many recent seeds to keep. Bump it up if you're doing long exploration sessions and want more history; keep it small to stay tidy.
There are no outputs - it's a terminal display node (an output node), so it sits on a branch off your seed and just logs. Nothing downstream depends on it.
How to install it
ComfyUI Manager: search ComfyUI Inspire Pack, install, restart. Manual:
cd ComfyUI/custom_nodes
git clone https://github.com/ltdrdata/ComfyUI-Inspire-Pack
then restart. Dr.Lt.Data's pack (ComfyUI-Manager, Impact Pack).
Common issues & troubleshooting
The log is empty or not updating. The seed input has to actually receive the seed each run. If it's not connected to a live seed source - or it's wired to a fixed value that never changes - there's nothing new to record. Route the same seed that drives your sampler into it.
It logs a different seed than the one that made the image. Classic gotcha with control_after_generate. The seed shown on the sampler after generation may already be the next seed, not the one just used. Drive both the sampler and the logger from one shared seed source so they always agree, rather than reading the sampler's post-run widget.
The good seed already scrolled off the log. Your limit is too low. Raise it so more history is retained during a session - the default of 5 is fine for quick work but short for a long exploration run.
I want the seed as an actual output to reuse. This node doesn't provide one - it's purely a display. To use a recovered seed, read it off the log and type it into a fixed seed input. For programmatic seed control, look at Inspire's Global Seed or Seed Explorer nodes instead.
Inputs (3)
| Name | Type | Default | Description |
|---|---|---|---|
| seed | INT | 00–18446744073709550000 | — |
| seeds | STRING | — | |
| limit | INT | 50–18446744073709550000 | — |
Outputs (0)
No outputs