⚡| Image Labeler

Image Labeler 2 is the current version of the pack's image-annotating utility: it takes an image and a string, renders the string as a label onto the image, and hands the result back. That's the whole job. If you're assembling comparison grids, building datasets, or watermarking test renders with their prompt/seed, this is the node that puts the text on the pixels before you save them.

It's the successor to the pack's original ImageLabeler (now deprecated). The difference is a single enable_label master switch, which is more useful than it sounds: you can keep the label configuration wired up and just toggle whether it applies, without deleting text or resetting 10 settings.

How it works

The label is drawn as a colored text box positioned on the image, with a background so it stays readable over busy renders. The geometry is all controllable: position, padding, font, size, and minimum label dimensions. There's also a font_scale option (none / by_image_width / by_image_height) that auto-scales the text relative to the image size - the right choice if your batch has mixed resolutions, since a fixed pixel size looks enormous on a 512px render and tiny on a 2K one.

The inputs that matter

• image (IMAGE) - what to label.
• enable_label (BOOL) - the master switch. Off returns the image untouched.
• text (STRING) - the label content. Hex colors #RRGGBB for text_color and background_color.
• font_name (COMBO), font_size (8–256) - the type. Font size is auto-adjusted relative to image height, per the tooltip.
• label_position (COMBO, default bottom_right) - corner/edge placement.
• label_padding (INT, default −1) - −1 means automatic padding based on font size.
• min_label_width / min_label_height - floor sizes so a short label or small font still gets a decent box.
• font_scale - auto-scaling mode as above.

Output: the labeled IMAGE.

Install

Part of the Z-Image Power Nodes pack:

cd ComfyUI/custom_nodes
git clone https://github.com/martin-rizzo/ComfyUI-ZImagePowerNodes

Restart ComfyUI (or ComfyUI Manager → "Z-Image Power Nodes"). No pip dependencies.

When you'll actually use it

The killer combo in this pack is Labeler → Grid Builder: label each render with its style or seed, then assemble them into a comparison sheet, and suddenly your seed/strength/style tests are readable at a glance instead of a mystery. Where people get burned: forgetting that font_size is capped at 256 and auto-scaled to image height - a "256" on a 1024px image is not as huge as you'd think, so if text comes out smaller than expected, check whether font_scale is fighting you. Set it to none if you want literal control. And the label_padding = −1 default looks right in most cases; only touch it when the box feels cramped.