Sign In

Theycantalk

33

144

4

15

Updated: Jan 11, 2026

style

Verified:

SafeTensor

Type

LoRA

Stats

0

0

Reviews

No reviews yet

Published

Jan 11, 2026

Base Model

ZImageTurbo

Training

Steps: 3,000

Usage Tips

Strength: 1

Hash

AutoV2
2BA0957D2B
default creator card background decoration
First Birthday Badge
tigerart's Avatar

tigerart

A comic strip style reminiscent of the stylistics of the famous “They can talk” comic book series

Z-Image Turbo

LoRA for Z-Image Turbo is very sensitive to prompts because it is trained on style rather than specific objects. Short prompts will give unsatisfactory results: LoRA will try to draw frames for comics, dialogue texts in a strange mixture of Cyrillic and Latin characters, and add the author's logo.

Write long, detailed prompts describing objects.

* If you are using AI to generate prompts, try adding the following to the initial conditions: “Do not describe style, lighting, or mood; describe only the objects in the scene.”

Let LoRA deal with style.