Z-Image Turbo · Text to Image with Wildcards

ABOUT THE WORKFLOW

Generate Random Characters

Create character artwork from a text prompt while automatically randomizing selected details with wildcard syntax. Define your base character once, then let the workflow generate new hairstyles, accessories, objects, or other prompt variations every time you run it.

Model

• Z-Image Turbo by Tongyi-MAI (Alibaba). A fast open-weight text-to-image model designed for high-quality image generation.

• LoRA support. Optionally applies a LoRA to influence the generated style while keeping wildcard variations.

HOW IT WORKS

Step 1. Describe your character

Write the main prompt describing your subject.

Works great with: anime characters · portraits · concept art

Step 2. Add wildcard variations

Use wildcard syntax to randomize parts of the prompt. For example:

"{long hair|bob hair|ponytail}, samples/flower"

Each run selects different combinations automatically.

Step 3. Run the workflow

The wildcard processor expands your prompt before image generation.

Step 4. Download your image

Each generation produces a unique variation while keeping your base concept consistent.

Ready for: social media · concept art · character design

First time? Leave every setting as-is. The defaults are the right starting point for almost everyone.

RECOMMENDED SETTINGS

Quick-start guide. Find the goal that matches yours and copy the settings.

• General character generation – Keep the default resolution (1920×1280), 9 sampling steps, CFG 1, and random seed.

• Want more variety? Leave the seed randomized and add more wildcard options to your prompt.

• Need consistent results? Lock the seed while changing only specific wildcard entries.

• Want larger batches? Increase the batch size to generate several variations at once.

• Images look too similar? Expand your wildcard lists with more hairstyles, clothing, colors, or accessories.

• Prompt feels repetitive? Move fixed details into the base prompt and randomize only the elements you want to vary.

Prompt: Keep the core description stable and randomize only selected details. "A woman sitting in a café, {long hair|bob hair|ponytail}, wearing a samples/flower" produces more useful variations than "random woman."

USE CASES

🎨 Character Design

Generate multiple character concepts while keeping the same overall idea.

📚 Dataset Creation

Produce varied images for LoRA training or dataset collection.

🖌️ Concept Art

Explore different hairstyles, accessories, outfits, and visual themes quickly.

📱 Social Media Content

Generate unique character illustrations for posts, avatars, and creative projects.

WHAT WORKS BEST / WHAT TO AVOID

✅ Works great

• Clear base character descriptions

• Focused wildcard categories

• Portraits and character illustrations

• Multiple hairstyle or accessory combinations

⚠️ May produce softer results

• Extremely long prompts

• Too many wildcard categories in one prompt

• Contradictory prompt descriptions

• Randomizing every part of the prompt simultaneously

LEARN

📹 Videos

• Intro to Floyo

• ComfyUI 101 Free Course ft. Sebastian Kamph

• Floyo 101 for Team Collaboration

✨ Quick links

• Help docs

• LoRA Training guide

• Discover more workflows

FAQ

What are wildcard prompts?

Wildcard prompts automatically replace parts of your prompt with random values from a list or wildcard file. This lets you generate many variations without rewriting the prompt.

How does this workflow create different images every run?

The wildcard processor expands the prompt before image generation. Combined with a random seed, each run produces a different prompt variation.

Can I use my own wildcard files?

Yes. You can create or import your own wildcard collections for hairstyles, clothing, colors, expressions, locations, or any other reusable prompt category.

Can I generate consistent characters?

Yes. Keep the base prompt the same, use a fixed seed, and only randomize the attributes you want to change.

Is Z-Image Turbo good for character generation?

Yes. Z-Image Turbo is optimized for fast, high-quality image generation and works well for character concepts, illustrations, and creative exploration.

How to run Z-Image Turbo with wildcards online?

You can run Z-Image Turbo with wildcard prompts online through Floyo. Open the workflow, write your base prompt, add wildcard syntax, and hit Run. Free to try.

Ready to try it? Type your first prompt and run it. The settings are already set.

→ Launch Workflow, Free

Questions? Watch the free course or check the FAQ above.