You can click on the link below to try it out directly. If the effect is good, you can deploy it locally
https://www.runninghub.ai/post/2016180945006628865/?inviteCode=sdhs0trb
Fan benefits,register to get 1000 points,daily login 100 points,play 4090!Experience the super power of 48G.
Hugging Face: WarmBloodAban/Flux2_Klein_Anything_to_Real_Characters · Hugging Face

Model Description
This is an Anime-to-Realism (Live-Action) LoRA, now updated to the Flux2 Klein version. The training dataset consists of 145 curated image pairs (totaling 290 images).
🛠 Optimization List:
Constraint Refinement: Optimized the model's specific architectural constraints.
Facial Expressions: Enhanced the fidelity and restoration of character expressions.
Demographic Diversity: Improved accuracy for various groups, including children, males, and females.
Realism Boost: Significant improvements in skin texture and overall photorealism.
🔑 Main Prompts:
realistic style of a young Asian girlrealistic style of a young Asian womanrealistic style of a young Asian female childrealistic style of a young Asian manrealistic style of a young Asian male
💡 Additional Prompting Tips:
Flexibility: Adjust prompts based on gender, ethnicity, and age groups.
Enhancement: Add descriptive tags after the main prompt to strengthen character traits, background details, and specific expressions.
⚡ Klein Features:
The Flux2 Klein model excels primarily in Text-to-Image (T2I), followed by Image Editing. Supported by this LoRA, it offers exceptional editing capabilities. To maximize the model's performance, provide detailed descriptions following the main prompt.
⚙️ Recommended Settings:
Base Model:
Flux2 Klein 9BCFG Scale:
1Steps:
8
Description
FAQ
Comments (15)
Worth a try. It works way better than default Klein at drawn to realism using various non-anime styles, and even loose sketches.
Sadly, doesn't work as good as the qwen version. The QIE version is my favorite, but the Klein version still can't fix Klein's main problem. It doesn't substitute the character with a real person, it just re-skins the drawing, so bad proportions are retained, huge eyes, big head, etc. Unless someone discover a magic prompt or process to fix this, not very good. Basically it's only better for semi-real art or drawins with already perfect proportions.
Yes, I have trained many Klein's Lora models, and I found it very difficult to change its image editing function. I can only train styles based on it, which is completely different from the training of Qwen Edit.
@AIGC_Singularity may I ask how you trained this lora on klein and the same on qwen edit? I want to try using my own dataset to make one, but I have never tried to make a style transfer lora before. How many pictures have you used in your dataset, how are they captioned, and what are the settings etc. Thank you!
毕竟是新模型,未来可期
This lora is just amazing.
I created a workflow that compared the lora effects of multiple anime-to-real human conversions, including this lora(Flux2 Klein_Anything to Real Characters).You can click the link to give it a try.
Thanks to the author for providing such a powerful tool for our community!
Fantastic Lora, the consistency is amazing. The accurate transfer of the character's eye color is particularly impressive. Working with fabric and costumes is also excellent. It creates the best images among all those tested. Are there any updates/side loras or life hacks expected for creating characters with European facial features?
It's the best conversion Lora for Klein bar none. I know nothing about if stuff like this can be updated, but I can't wait to see what's next!
remember to drink water guys this lora is kinda addicting
Man im not used to comment but you deserve it. Awesome LORA+workflow. Achivied in 20 seonds something i was trying to do for hours. GJ
This lora is incredible but it seems it has a tendency to shift the colors to be somewhat dark and muted. I wonder if this could be fixed by prompting?
Might want to double check your sampler
This is the best thing I've seen in the last year. Previously, I had to write a giant Google Gemini prompt to create a similar effect. Now it's done in one click! Thanks so much to the creator of this Lora!
I really like this lora thank you !!!
How to do train from multi image ? I believe it's not khoya or ostris
Details
Files
Flux2 Klein动漫转写实真人 AnythingtoRealCharacters.safetensors
Mirrors
flux220kleinAnything-to-Real.safetensors
Flux2 Klein动漫转写实真人 AnythingtoRealCharacters.safetensors
Flux2 Klein动漫转写实真人 AnythingtoRealCharacters.safetensors
flux220kleinE58AA8E6BCABE8BDACE5.O9j8.safetensors
RealCharacters.safetensors
Anything.safetensors
any2real.safetensors
Flux2 Klein动漫转写实真人 AnythingtoRealCharacters.safetensors
flux220kleinE58AA8E6BCABE8BDACE5.O9j8.safetensors
Flux2 Klein_Anything to Real Characters--Flux2 Klein动漫转写实真人 AnythingtoRealCharacters-mid_2343188-vid_2635669.safetensors
Flux2 Klein动漫转写实真人 AnythingtoRealCharacters-mid_2343188-vid_2635669.safetensors
Anything to Real Characters - Klein9B - realistic style of...,cfg1,steps8.safetensors
Anything to Real Characters - Klein9B - realistic style of...,cfg1,steps8.safetensors
Flux2 Klein_Anything to Real Characters--Flux2 Klein动漫转写实真人 AnythingtoRealCharacters-mid_2343188-vid_2635669.safetensors
Flux2 Klein动漫转写实真人 AnythingtoRealCharacters-mid_2343188-vid_2635669.safetensors
18-flux220klein.safetensors
AnythingtoRealCharacters.safetensors
flux220kleinE58AA8E6BCABE8BDACE5.O9j8.safetensors
Flux2 Klein_Anything to Real Characters.safetensors
AnythingtoRealCharacters.safetensors












