I like generating images with speech bubbles, but the text always comes out as gibberish and I'm way too lazy to fix it in an editor every time. So I made a small node pack that does it for me: the model draws the bubble, the node erases its letters and writes my actual text (emojis too).
Works with Anima, Illustrious and NoobAI. I use LoRA Manager, so it works with it too (recipes save the fixed image), but it's optional.
Heads up: it's vibe-coded, and the node labels are in Spanish because that's my first language. The README is in English though.
https://github.com/arturor1990/ComfyUI-BubbleText
