Image Info Count Debug
Is your metadata list longer than the images you fed it?
- image_info
- image_info
Image Info Count Debug is the metadata twin of the pack's Image Batch Count Debug: it takes an image_info payload, logs how many entries it contains to the ComfyUI console, and relays the payload unchanged. Same idea, one level up the stack. Where the batch debug node probes image tensors, this one probes the image_info bundles - and in this pack, mismatched counts between the two are exactly the kind of bug that makes a workflow silently skip half your images.
Why the count matters
This pack's image_info travels as a list when you're processing multiple images - the Image Directory Reader, for instance, outputs image, image_info, and caption as aligned lists. "Aligned" is the load-bearing word. If the image list has 40 entries and the image_info list has 39, everything downstream that zips the two together produces garbage, wrong captions, or an error that points somewhere unhelpful. A node that reports the metadata count at a specific point in the graph tells you which list is lying and where the alignment broke.
It also reveals the subtler mismatch: one image, but a list of image_info. Batch-of-one on the image side, metadata-for-many on the other. You can't see that from the canvas; you can see it in one console line.
Inputs and output
image_info- optional, the payload to inspect and pass through.enabled- defaulttrue. Disable it to silence the logging while keeping the relay, so your probe can stay parked in the workflow.label- optional prefix for the console line. Name your probes -"dir_reader","after_context"- and the log reads like a trace instead of a wall of numbers.
Output is the same image_info, untouched.
Using it well
Same discipline as any pass-through probe: place one before and one after the node you suspect, run once, compare the two logged counts, and the culprit announces itself. Disable rather than delete when you're done. If the pack's Image Info Context is in the chain, probe both sides of it specifically - that's the node that rebuilds image_info from whichever inputs are connected, and its output count is a function of what went in, not what you intended.
The honest limit, same as its image sibling: it reports the count, not the cause. But knowing the count diverged between these two points narrows a search that can otherwise eat an evening.
Installing it
Part of the ComfyUI-Info-Prompt-Toolkit pack. ComfyUI Manager - search "Info Prompt Toolkit" - or:
cd ComfyUI/custom_nodes
git clone https://github.com/kinorax/comfyui-info-prompt-toolkit.git
cd comfyui-info-prompt-toolkit
pip install -r requirements.txt
Restart ComfyUI. No model downloads. Small single-author pack (GPL-3.0) with a modest footprint - a ten-second console read that saves you a thirty-minute bisect is a good trade.
Inputs (3)
| Name | Type | Default | Description |
|---|---|---|---|
| enabled | BOOLEAN | true | If false, this node logs nothing and only relays image_info |
| image_infoopt | IPT-IMAGEINFO | Pass-through image_info payload | |
| labelopt | STRING | Optional prefix shown in console logs |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| image_info | IPT-IMAGEINFO | — |