Text Image with FX

Burning text onto an image in ComfyUI is normally an exercise in pain - you fight font loading, you fight positioning, you give up on glow. Text Image with FX is the node that does it properly: it renders a text string with optional outer glow and drop shadow, then anchors it to any corner or edge of your image for watermark and copyright placement. You get both a composited image and a mask output, so the text layer stays a separate element you can use however you like.

The positioning model is the part to understand before anything else. position is an anchor - bottom_right, top_left, and the rest - and margin_x/margin_y are the offsets from that anchor's edges. So a copyright mark is: pick bottom_right, set margins, done. No manual coordinate math, and it stays glued to that corner regardless of image size. That anchor-based flow is genuinely nicer than most generic text renderers, and it's exactly what a watermark workflow wants.

How it works

The essentials:

• text - what to render (defaults to "© Eclipse", which tells you what the author was thinking).
• font_file - picked from ComfyUI/models/fonts/ (the pack copies bundled fonts there on first run - Allura-Regular, Roboto, Oswald, and friends). Drop your own .ttf in that folder and it appears in the dropdown.
• font_size, text_scale, char_per_line, leading - sizing and wrapping. text_scale is a percentage of font_size, so it's a quick nudge without retyping a point size.
• text_color, stroke_width/stroke_color - fill and outline, hex values.
• position, margin_x/margin_y - the anchoring.
• opacity - overall transparency of the composited text.
• enable_glow + glow_intensity/glow_range/glow_blur/glow_inner_color/glow_outer_color - an outer glow with inner and outer colors, so you get that neon gradient edge. The intensity is "buildup iterations," which is just a way of saying more iterations = denser glow.
• enable_shadow + shadow_offset_x/shadow_offset_y/shadow_grow/shadow_blur/shadow_color/shadow_opacity - the drop shadow.

Outputs are image (the composited result) and mask (the text layer as a mask), so you can composite manually, blur the background, or re-position downstream.

Where it fits

Watermarks and copyright marks are the headline use - the anchor system is built for them. But it's also fine for titling, date-stamping, or any "put readable text on this image" need that would otherwise drag in a full graphics stack. The mask output is the sleeper feature: it means the text region is first-class data you can mask with.

Installing it

Part of ComfyUI_Eclipse:

git clone https://github.com/r-vage/ComfyUI_Eclipse custom_nodes/ComfyUI_Eclipse

Or ComfyUI Manager → search "ComfyUI_Eclipse" → install and restart. It needs the bundled fonts copied into models/fonts/ (automatic on first run) and Pillow, which ComfyUI already ships.

Gotchas

Two things catch people. First, fonts: if the dropdown is empty, the models/fonts/ directory didn't get populated - restart once and check the folder; any .ttf you add there shows up next time. Second, glow and shadow are off by default - a "why is my text flat" moment is usually a missed enable_glow toggle, not a bug. And a hex color that looks wrong is worth checking against the color picker the frontend adds for these fields. Old RvTools-era workflows predating v4.0.0 need the migration tool.