Nodes/LF Nodes/Update usage statistics
ComfyUI Node

Update usage statistics

Track which checkpoints and LoRAs you actually burn time on

By lucafoscili·Created 2 years ago·Updated 2 years ago· 50
Update usage statistics
  • dataset
  • ui_widget
  • dir
  • dataset
datasets_dirWorkflow_name

Here's a question nobody can answer off the top of their head: which model have you actually generated the most images with this month? UpdateUsageStatistics is the node that finds out. It's a logging node - an output node whose whole job is to keep a running count of which resources you use, save it to disk, and let a sibling node draw charts from it later.

It's built around the analytics side of LF Nodes. The mechanism: every time the workflow runs, this node reads the resource dataset you give it, bumps a counter for each entry, and writes the updated dataset to a JSON file in your ComfyUI user directory under LF_Nodes. The datasets_dir input (default "Workflow_name") is just a subfolder name so different workflows don't stomp each other's stats - the tooltip spells it out: files live in the ComfyUI user directory under LF_Nodes, and this field adds extra folders on top.

The inputs that matter:

  • dataset - the JSON containing the resources used, produced by CivitAIMetadataSetup. That's the crucial wiring: this node isn't for free-typing; it expects the resource metadata that LF's CivitAI setup node emits, so it knows which checkpoint/LoRA/VAE you actually loaded.
  • datasets_dir - the subfolder name for the stats files.

The outputs are dir (the path it saved to) and dataset (the updated dataset, passed through so you can chain it). Once you've accumulated runs, you read the stats with the companion UsageStatistics node, which plots them as area/scatter charts in a widget - the "did I actually get my money's worth out of this model" view.

Where you'd use it: if you have a stable personal workflow and genuinely want usage telemetry on your own machines - which LoRA you keep reaching for, whether that new checkpoint is earning its download size, what your A/B tests actually landed on. It's a niche tool; if you never wonder about your own usage patterns, you don't need it. But for people who curate a local model collection, it's the only node that answers the question with numbers instead of vibes.

Installing LF Nodes - same pack, same steps:

  • ComfyUI Manager → search "LF Nodes" → install → restart, or
  • cd ComfyUI/custom_nodes && git clone https://github.com/lucafoscili/comfyui-lf then restart.

Two honest caveats. The repo is in legacy mode - frozen at its last stable release while development moved to lucafoscili/lf-nodes - and this node's name and behavior carry over, but check the new repo's docs before depending on it in a fresh setup. And the data only accumulates if this node actually runs, so put it on the main path of your workflow, not on a dead-end branch - otherwise your stats will look like you only ever generated once. No models to download; the only dependency of note in the pack is piexif, used for reading metadata off images.

Category✨ LF Nodes/Analytics

Inputs (3)

NameTypeDefaultDescription
datasets_dirSTRINGWorkflow_nameThe files are saved in the user directory of ComfyUI under LF_Nodes. This field can be used to add additional folders.
datasetJSONDataset including the resources (produced by CivitAIMetadataSetup).
ui_widgetoptKUL_CODE[object Object]

Outputs (2)

NameTypeDescription
dirSTRING
datasetJSON