2🐕Text watermark addition

The README calls this section "Watermark Master," and it's not overselling it much. This is the text half of it: type a string, style it, and it gets stamped onto an image (or rendered on its own blank canvas) with real control over font, color, position, rotation and opacity - closer to a proper text tool than the usual single-purpose "add watermark" node you find in most packs. Pair it with the pack's batch image loader and the README's own pitch is that you can stamp a whole folder in one pass.

How it works

You're rendering styled text and compositing it onto an image. If you plug something into the optional image input, the text gets stamped directly onto it; leave it empty and the node draws the watermark onto a blank canvas sized by width/height instead - handy if you want a standalone watermark layer to composite elsewhere. Position, rotation and spacing are all independent controls, so you can park a signature bottom-right, tile a repeating mark, or rotate a diagonal "SAMPLE" banner across the frame.

One thing to know before you touch it: the text field doesn't ship blank. It defaults to the author's own usage note (a wall of Chinese and English text about font licensing and where fonts live on disk) - that's not a bug, it's just the placeholder. Clear it and type your actual watermark text first.

The inputs and outputs that matter

The core styling knobs, the ones you'll actually touch:

• text (multiline STRING) - what gets stamped. Overwrite the default placeholder.
• Font (enum) - pulled from the pack's bundled fonts folder.
• Font_size, Font_color, Font_transparency - size, color, and alpha of the text.
• Vertical_position / Horizontal_position (enums) plus Text_margin, Row_spacing, X_offset, Y_offset - placement, for both quick presets (centered/left/right, up/down) and fine pixel nudges.
• Rotate (degrees, float) and Rotation_center (rotate around the text block, or around the whole image) - for angled or tiled watermarks.
• Background_color - a fill behind the text, separate from the text color itself.

Font_color and Background_color are both enums with about 30 named swatches, and one of those choices is literally custom - pick it to unlock the matching optional hex fields, Font_colour_Hexadecimal and Background_colour_Hexadecimal, if none of the presets is the exact shade you need. width and height set the canvas size when there's no source image. The only other optional input is image - supply it to stamp onto a real photo instead of a blank canvas.

Output is a single Output_Image (IMAGE).

How to install it

Search Comfyui-ergouzi-Nodes in ComfyUI Manager, or clone it manually:

cd ComfyUI/custom_nodes
git clone https://github.com/11dogzi/Comfyui-ergouzi-Nodes.git

Restart ComfyUI. The fonts this node ships with are, in the author's own words, public web resources meant for learning and communication - fine for personal work, but if you need this on commercial output, swap in a font you actually have rights to rather than assuming the bundled ones clear that bar. Also worth knowing: this English repo is marked as no longer updated in the README, with a more complete Chinese-named sibling pack getting the ongoing work instead - if you hit a watermark feature you'd expect but don't see here, that's likely where it lives.

Common issues & troubleshooting

Typed a hex code and nothing changed. Check that Font_color or Background_color is actually set to custom - with a named swatch selected, the hex field you typed into is being ignored in favor of the dropdown choice.

Watermark text still says the author's install notes. That giant default string in text is a placeholder, not a bug - clear it out before you connect this node to anything real.

Text looks squashed or cut off after rotating. Rotation_center changes whether the text rotates around its own block or around the whole canvas; if a rotated watermark is landing somewhere you didn't expect, that's the setting to flip before you start fighting the offset values instead.

Batch-stamping a folder. The README frames this node as designed to pair with the pack's batch image-loading nodes - load a folder, run each image through this node, and you get consistent watermark placement across the whole set instead of repeating the setup by hand.