Why your LoRA's trigger word prints on clothes, and how to test for it

We trained a character LoRA on Z-Image with the trigger word wr3nn. The character came out well, but the test images had a surprise: "wr3nn" appeared on a woolly hat, a hoodie patch and other clothes.

It makes sense. Z-Image is good at drawing text, and a word that looks like a name or label can get drawn as one.

How to test any new LoRA:

1. Use prompts with a blank surface where text could land, such as a plain white t-shirt, a woolly hat or a hoodie.

2. Render several seeds. The problem doesn't show up in every image.

3. Check every saved checkpoint, not just the final one.

4. Zoom in. Small labels and patches are easy to miss at thumbnail size.

What we changed: on our next Z-Image character we used a plain full name as the trigger instead of an l33t-style word. In 24 test images across 4 checkpoints, nothing printed on the clothes. That's one character, so it's not proof, but it's a cheap thing to try.

If it happens with a LoRA you've already got, adding "text, letters, logo" to the negative prompt may help on models that use one, or you can reword the clothing in the prompt, for example "unbranded plain hoodie".

We're using this for the next version of our free character, Wren.

More guides: https://bravikron.com/guides.html