ℹ️ Exif check

You've got an image and you want to know one specific thing baked into it - the width it was actually shot at, the software that last saved it, the compression it used. Exif check is the small utility that digs a single named field out of an image's EXIF metadata and hands you the value back as plain text. That's the entire job, and it's a genuinely useful one when you're auditing where a picture came from or wiring a value out of a file into the rest of a graph.

The mechanism is worth getting straight because it's the thing people trip on. EXIF is the metadata block that cameras and photo editors write into JPEG and TIFF files - camera model, exposure, dimensions, orientation, timestamps, all of it under a couple hundred standardized tag names. This node doesn't crack open a file on disk itself. It takes an exif string that's already been pulled out - typically fed from one of the pack's other Info_check nodes, the ones that read the raw metadata off an image - and an info_type picker that lists 104 standard EXIF tags (ImageWidth, ImageLength, Compression, ProcessingSoftware, BitsPerSample, and so on). You choose the tag, it returns that tag's value on the info output.

So the two inputs that matter are exactly those two: exif, the metadata blob you're inspecting, and info_type, the tag you want out of it. The single info output is a STRING, which means you can drop it into a text-display node to just read it, or route it somewhere downstream - into a prompt, a filename, a switch - if you're building something automated around a photo's real properties.

One distinction to keep in your head: EXIF is camera and photo metadata, not the generation metadata AI tools embed. If what you actually want is the ComfyUI workflow JSON or an A1111 parameters string tucked into a PNG, that lives in a different place and you'd reach for the pack's Metadata check node instead. Exif check is for the traditional photographic tags, and asking it for a generation prompt will just come back empty.

Installing it comes with the whole pack, not on its own. In ComfyUI Manager, search the pack title (SDVN_Comfy_node) and install, or do it by hand: cd ComfyUI/custom_nodes && git clone https://github.com/StableDiffusionVN/SDVN_Comfy_node, then from your ComfyUI root run pip install -r custom_nodes/SDVN_Comfy_node/requirements.txt and restart. There's no model to download for this one - it's a pure metadata reader, no weights involved.

Troubleshooting is mostly about empty output, and the cause is almost always the image. A blank info means that tag simply isn't present. PNGs exported from AI tools generally carry no camera EXIF at all - they were never photographs - so pointing this at your last Flux render will get you nothing. Screenshots and images scrubbed by social platforms usually have their EXIF stripped on the way through, too. Feed it an actual camera JPEG and the tags fill in. And if the node won't even appear after install, that's the pack-wide failure mode rather than anything specific to this node: it's the requirements.txt step that didn't complete, so check your ComfyUI startup log for an import error and re-run the pip install. Beyond that, there's not much to break here. It reads a tag and gives you the value, and that reliability is the whole appeal of a node this small.