Flux.2 Klein in ComfyUI: Simple Guide for Text-to-Image and Editing

About This Workflow

Flux Klein – Text to Image (How It Really Works)

Flux Klein makes text-to-image generation simple and flexible. You write what you want to see, and the model turns it into an image. It works well for both clean photos and creative art.

1. Clear prompts give better results

Flux Klein responds best to simple and direct prompts. Short descriptions with clear subjects work better than long, messy sentences. A common mistake is adding too many ideas in one prompt, which often confuses the model.

Tip: Describe one main subject first, then add style or mood.

2. It understands lighting and mood well

In real use, many users notice that Flux Klein handles lighting very naturally. Words like soft daylight, studio lighting, or cinematic light actually change the image in a visible way. This makes it useful for realistic photos and stylish scenes.

3. Style words matter more than people expect

Small style words like modern, clean, cinematic, or painterly can change the whole look. A common issue is skipping style hints and wondering why the image feels plain.

Tip: Add one or two style keywords, not a long list.

4. Simple backgrounds improve subject quality

When beginners use very complex backgrounds, the main subject often looks weaker. Flux Klein usually performs better when the background is described in a simple way. You can always add more detail later using image edit.

5. Text-to-image is best for creating a strong base image

Many users use Flux Klein text-to-image to create a clean starting image. After that, they switch to image edit for changes like poses, expressions, or style shifts. This workflow gives more control and saves time.

Required Models

Model links

diffusion_models

• flux-2-klein-9b-fp8.safetensors
• flux-2-klein-base-4b-fp8.safetensors - {use this model on the 4b workflow}

text_encoders

• qwen_3_8b_fp8mixed.safetensors

• qwen_3_4b.safetensors - {use this model on the 4b workflow} vae

• flux2-vae.safetensors

Model Storage Location

📂 ComfyUI/
├── 📂 models/
│   ├── 📂 diffusion_models/
│   │      ├── flux-2-klein-9b-fp8.safetensors
│   │      └── flux-2-klein-base-4b-fp8.safetensors {for th 4b workflow}
│   ├── 📂 text_encoders/
│   │      └── qwen_3_8b_fp8mixed.safetensors
│   │      └── qwen_3_4b.safetensors {for th 4b workflow}
│   └── 📂 vae/
│          └── flux2-vae.safetensors

Workflow Graph