FLUX.2 KLEIN9B — Custom Merge
Model Overview
A custom merged model based on FLUX.2 [klein] 9B by Black Forest Labs. This model has been tested and optimized for high-quality realistic image generation with exceptional detail in textures, lighting, and atmospheric effects.
Key Features
Photorealistic skin rendering with natural subsurface scattering, visible pores, and subtle imperfections
Exceptional water and wet surface handling — droplets, reflections, refraction, and flow patterns
Advanced lighting capabilities including volumetric god rays, golden hour, rim lighting, and chiaroscuro
Superior hair detail with individual strand rendering and natural backlighting effects
Rich material textures — velvet, linen, silk, leather, wood, metal, and stone
Strong atmospheric effects — fog, mist, steam, rain, and environmental depth
Anatomical accuracy for both portrait and full-body generation
Recommended Settings
Optimal Configuration (Tested):
CFG Scale: 2.0
Steps: 15
Scheduler: Euler Ancestral (EulerA)
Quick Previews:
CFG Scale: 1.5
Steps: 8-10
Scheduler: EulerA
Complex Scenes (Landscapes, Materials):
CFG Scale: 2.0-2.5
Steps: 15-20
Scheduler: EulerA
Resolution: 1024x1536 (portrait) or 1536x1024 (landscape)
Best Use Cases
Photorealistic portraits and character art
Environmental and landscape photography
Fashion and material studies
Cinematic and atmospheric scenes
Fine art nude photography (artistic interpretation)
Macro detail shots
Water and weather effects
Known Limitations
Transparent/translucent fabrics may require LoRA or enhanced prompting for proper rendering
Text rendering is readable but not perfect — acceptable for most use cases
Complex multi-element scenes may require 2-3 generations for optimal results
Official 4-step inference is too low for detailed work — 15 steps recommended
Technical Details
Base Model: FLUX.2 [klein] 9B by Black Forest Labs
Architecture: Rectified Flow Transformer
Type: Step-distilled model (optimized for speed)
License: Refer to original FLUX.2 license terms
Performance Notes
This model excels at organic, natural scenes with complex lighting. The EulerA scheduler at 15 steps provides the best balance between speed and quality. While the official recommendation is 4 steps at CFG 1.0, extensive testing shows 15 steps at CFG 2.0 delivers significantly better results for detailed work without sacrificing the speed advantage of distilled models.
Credits
Base model by Black Forest Labs. Custom merge and testing by community contributors.
Enjoy creating! If you generate something amazing, consider sharing it in the images section. 🎨
Description
Sorry, work in progress....
FAQ
Comments (20)
I can't get this to work! a WF sample would be great
Sorry, i don't use Comfy
@FASCIUM Not even the provided workflows work. I am curious about what did you use to generate these images. ComfyUI is just an interface, you can emulate whatever environment the model requires. What is the magic requirement, let us know something about the model. You want us to use it, don't you? Looks promising...
@nerd_bgd all my images contains promt and all technical information, no magic
@FASCIUM what are you using? I use both ComfyUI and SwarmUI - which is similar to FORGE. When I try to generate an image I get an error each time. Would like to test this model if possible.
@Ghostj27f i work only in ForgeNEO, haven't any another UI. Dont forget - its early test, some troubles is normal
Correct encoder is qwen_3_8b, correct VAE flux2vae
@FASCIUM You should really switch to ComfyUI, that's what all cool kids use nowadays :)
Not working in ComfyUI
Correct encoder is qwen_3_8b, correct VAE flux2vae
@FASCIUM Yes, all correct but i got only noise
@210881175 He uses layer skip, but I can't get it to work with it neither. Provided workflow complains about missing parameters, I believe my sfz nodes is the wrong version.
I will investigate a bit more - alternative is the ZIT version, it works good, with similar results...
Not working in ComfyUI
this could potentially be a good model but no one can use it lol
i don't even know how that's possible, to train a model that can't be used in comfyui for whatever reason.
Something is wrong with the quantization. Why don't you just provide us with the FP/BF16 version? I don't understand...
Its very specific model, found many troubles, its VERY EARLY checkpoint. Use clear model
I have my own metod of merging, with this model it works very strange
@FASCIUM That's okay, I'm looking forward to the next versions.
@heinerheinervonvielen417 to be honest - its a long way.... very difficult customizing
@FASCIUM The first usable fine tunes from SDXL appeared about a year after release. Unfortunately, this process is part of it. There will be new methods and new formats. It's going to be a great year.
















