Klein 9B lcs consistency lora 20260415
RunningHub:
Lora:
https://www.runninghub.ai/model/public/2044762492517228546?inviteCode=rh-v1279
Online Workflow：
https://www.runninghub.ai/post/2016109710943981570/?inviteCode=rh-v1279
Klein 9B consist lcs preivew 20260328
preview version of lcs training method.
Further version is under development
Klein 4B Consistency Lora
This lora inspired by ByteDance Heilos Easy Anti-Drifting method.
I implemented Frame-Aware Corrupt (Reference Latent Corrupt) latent and Color Statistics Loss.
In additionally, I use frequence split dataset which created by converting an image into frequence domain and split it into high freq image and low freq image.
Reference: low freq image, high freq image (0.8 dropout rate on high freq image)
Target: clean image
lr: 2e-4
scheduler: cosine
optimizer: adamw8bit
steps: 6000
Suggested Prompt:
{edit prompt}. Transform the image to realistic photograph. add realistic details to the corrupted image. restore high frequence details from the corrupted image.
Suggested Strength:
0.5-0.7
Trained on 4B based. This lora aims to preserve image consistence across multiple input images. Control the lora via lora strength.
More strength means more consistence but it would also hurt edit abilities.
Different image might has different optimal lora strength.
Running Hub
Flux2 Klein 4B一致性lora对比 runninghub.ai/post/2032812180667633666/?inviteCode=rh-v1279
Flux2 Klein 4B一致性lora 不偏色 不偏移 runninghub.ai/post/2032817113190113281/?inviteCode=rh-v1279
Flux2 Klein 9B Consist Lora
HuggingFace:
https://huggingface.co/lrzjason/Consistance_Edit_Lora
Klein 9B Consis Lora
Suggested Prompt:
restore image details
Flux2 Klein 9B Consis Lora
Trained on 9B based. This lora aims to preserve image consistence across multiple input images. Control the lora via lora strength.
Suggested strength: 0.2~1.5
Use with any prompts.
More strength means more consistence but it would also hurt edit abilities.
Different image might has different optimal lora strength.
RH:
https://www.runninghub.ai/post/2016109710943981570/?inviteCode=rh-v1279
======================QE2511================================
RH:
https://www.runninghub.ai/model/public/2010085010870640641?inviteCode=rh-v1279
Custom_node:
https://github.com/lrzjason/Comfyui-QwenEditUtils
QE2511 Alpha is trained on QwenEdit 2511.
As same as previous lora, this lora helps to maintain image consistence.
The lora is trained using nf4 qlora.
V2 lora is trained for QwenEditPlus aka QwenImage Edit (2509)
It is recommanded to use with strength between 0.5 ~1.0 for multiple images
0~0.5 for single image
It needs to use with workflow v4.0 https://civarchive.com/models/1939540
RH workflow:
QwenEdit Consistence Edit workflow v4.0
Experience link: https://www.runninghub.ai/post/1965033238311915522/?inviteCode=rh-v1279
QwenEdit Consistence Edit workflow v2.0
Solved pixels shift issue and keep qwenvl encode image ability.
RH workflow:
QwenEdit Consistence Edit workflow v2.0
Experience link: https://www.runninghub.ai/post/1965033238311915522/?inviteCode=rh-v1279
Trained with no ref image into qwenvl.
This lora typical to test the consistency with kontext like workflow.
It is suggested to use 0.45~0.5 lora stregth
Workflow:
https://civarchive.com/models/1939540/qwenedit-consistance-edit-workflow
RH workflow:
It's extremely fun and I highly recommend it to you!
Workflow: QwenEdit Consistence Edit workflow
Experience link: https://www.runninghub.ai/post/1965033238311915522/?inviteCode=rh-v1279
Description
FAQ
Comments (23)
I tried the lora with wan2gp , but gives error
I tried it with the provided F2K WF and low res img, it seems it upscales internally while maintaining same pixel numbers. sometimes lora doesn't properly transfer clothes while the default WF delivers all the clothes.
I might due to the ref longest edge setting. Could you post your input and reference in the gallery for me to check?
I can't wait to use the F2K version!
Blessed, I was waiting for this one. Let's see if it can do anime to photo-real decently now, true if big.
Update: Yeah, klein still not good for it. Sad.
Will this work with img2img distilled version?
Doesnt work for me. Sigh
For Klein the lora work fine on runninghub but with same WF has no effect locally on comfyui .. Plz Help ( ComfyUI 0.9.2ComfyUI_frontend v1.36.14DiscordComfyOrgEasyUse v1.3.4ComfyUI-OllamaLoRA Manager v0.9.11-63b087fComfyUI-Manager V3.37rgthree-comfy v1.0.2509092031 )
I dont know. Are you on latest update? In my local test, it is working
same
Same also ! Is the lora is Klein 9B or Klein 9B-base ?
Auto-answering for those who look for an answer :
It's a Klein 9b BASE lora, that creator use with turbo lora r128 (so also on 4 steps and cfg 1.0), and with some particular nodes on wf to achieve the provide result. Workflow is in the x3 old man compare picture.
As there it's a particular wf, with some blocked layers and some conditioning config, seems like it have to be used with all this settings to achieve same quality of results (not just putting the lora in your classic simple wf :p)
for some reasons .. its not working with some images, i dont understand why.. for some images it's perfect ! and itls not working on klein 4B
@mahmoudelshenavy958 It is a 9B lora. It doesn't work with 4B
I still don't get it. I use for outfit try one, prompt "Start with base Image 1. Put outfit from image 2 onto person on image 1. Preserve body shape, face, pose and everything else from Image 1, except outfit."
now when lora is "1" - it just copy image 2 on top of image 1(ignoring consistency - just crop and copy paste). If lora is "-1.0" it preserves image 1 and copy less and get disent results with -0.5 etc.
My thought was the stronger the lora the more it preserve image 1, but it seems opposite it shift to image 2 the stronger lora is.
Is it how it was trained? Is negative weights intended side effect?
PS
I dont want to use custom nodes etc, just trying to understand intended use of lora in normal pipeline
please my dear genius friend, we need it for Flux2 Klein 4B
这个一致性lora配合flein的高清修复放大，太完美了。感谢大佬
有工作流吗？还有提示词，一直不知道你们用FK高清放大具体怎么操作的，跟它说了什么
@jevencheung Running Hub Flux2 Klein 4B一致性lora对比 runninghub.ai/post/2032812180667633666/?inviteCode=rh-v1279 Flux2 Klein 4B一致性lora 不偏色 不偏移 runninghub.ai/post/2032817113190113281/?inviteCode=rh-v1279 夸克 https://pan.quark.cn/s/8dd2f61cbadc
@xiaozhijason 谢谢大佬
This is fantastic for qwen!
Details
Files
f2k_consis.safetensors
Mirrors
f2k_consis.safetensors
f2k_consis.safetensors
f2k_consis.safetensors
f2k_consis.safetensors
consis-klein.safetensors
f2k_consis.safetensors
consistence-edit.safetensors
f2k_consis.safetensors
f2k_consis.safetensors
f2k_consis.safetensors
f2k_consis.safetensors
consistency_F2k9B.safetensors
f2k_consis.safetensors
f2k_consis.safetensors
f2k_consis-mid_1939453-vid_2634354.safetensors
f2k_consis-mid_1939453-vid_2634354.safetensors