Z-Image Workflow Builder FAQs

Question 1

How does it handle LoRA integration?

Accepted Answer

The skill provides specific paths and strength recommendations for ZImageTurbo and ZImageBase LoRA categories, including automated sequential stacking logic for styles, concepts, and characters.

Question 2

What is the difference between Z-Image Base and Turbo?

Accepted Answer

Z-Image Base is a full model supporting negative prompts and 10-30 steps, while Turbo is DMD-distilled for high-speed 8-10 step generation without effective negative prompt support.

Question 3

Does this skill support the RedCraft DX1 checkpoint?

Accepted Answer

Yes, it includes specific configurations for the 17GB RedCraft DX1 combined checkpoint, offering optimized settings for both fast iteration and full-quality outputs.

Question 4

Can I use standard CLIP prompts with Z-Image?

Accepted Answer

While technically possible, this skill recommends natural language descriptions because Z-Image uses the Qwen LLM tokenizer, which performs better with descriptive sentences than traditional tag-based prompts.

Question 5

What are the VRAM requirements for Z-Image?

Accepted Answer

VRAM requirements range from ~8GB for Turbo separate components to ~17GB for the full RedCraft DX1 checkpoint, making it suitable for modern consumer GPUs like the RTX 4090.

Z-Image Workflow Builder

主要功能

使用场景

Z-Image Workflow Builder

主要功能

使用场景