This LoRA is suitable for the Qwen Edit model
NEW!_Workflow:：https://www.runninghub.ai/post/1966557857586065410/?inviteCode=rh-v1172
Register for free use
保持一致性强度推荐：0.5-0.8 万物迁移强度推荐：1.2-1.6
Recommended consistency intensity: 0.5-0.8; recommended migration intensity of all things: 1.2-1.6
Maintain consistency in image modifications, remove grease, and keep the quality consistent with the original image. Increase the intensity of all functions!
Thank you for the feedback images and testing from friends. Acknowledging participants: wallen, zero, jiong, yunor
Trigger words: Put it here, maintain the light and shadow of the painting（It works better to use Chinese)：Put it here,保持画风光影不变
If you need commercial customization or any suggestions or help, please contact me
V:budabiaoa Q:123326400
Description
This LoRA is suitable for the Qwen Edit model
Workflow: Qwen_edit single image experience address:https://www.runninghub.ai/post/1966557857586065410?inviteCode=rh-v1172/?inviteCode=rh-v1172
Register for free use
Maintain consistency. Recommended intensity: 0.6-0.8
Migration, recommended intensity: 1-1.6
Maintain consistency in image modifications, remove grease, and keep the quality consistent with the original image. Increase the intensity of all functions!
Thank you for the feedback images and testing from friends. Acknowledging participants: wallen, zero, 囧, yunor
Trigger words: Put it here, maintain the light and shadow of the painting（It works better to use Chinese)：Put it here,保持画风光影不变
If you need commercial customization or any suggestions or help, please contact me
V:budabiaoa Q:123326400
FAQ
Comments (39)
Your lora is the best ,It open so many trick to use in my game,cut me a ton of time consume and great consistent keep is up brother ,what is annoying me the most when making game art is background ,i really want to train lora of fixing background to match foreground perceptive just like your put in lora but too bad i had only 16gbvram.
Haha, 16G video memory can already do a lot of things. More than most people, if it doesn't work, there is still cloud use. Thank you for your support of me. Thank you for your attention
this is simply the best lora!
Ha ha. Thank you for your support. Having the opportunity to rebate the chart is also the biggest help for me. Or if you have any new ideas, you can also communicate them
请问一下作者能否放几个示范提示词吗？这样可以更好的理解这个lora的提示词运用方法，是“放在这里”作为开头，然后保持画风光影不变，还是英文的"put it here"开头，保持画风光影不变，又或者是提示词放在最后面？
提示词在KONTEXT V4版本有范例。或者你去lib看下。长图C站无法放出来，给我屏蔽了
It's working with version V2.0! what did you add there, but it works, the first version didn't work properly, thanks, we'll test it.
More reasonable and more fused data, what can't work? I didn't understand it too much
@Futurlunatic with version two, everything works thanks! the first version didn't work fine for me, so everything is fine. Thank you for your efforts, I really liked LORA.
@soyv4 Hahaha, you scared me to death!." I thought there was something wrong somewhere. I'm very glad to meet your usage requirements. Thank you for your attention. Thank you
If it's convenient, you can give me a picture to make it more perfect
@Futurlunatic I added a photo, my results
@soyv4 Well done!
用这个2.0版本换脸，会让角色莫名其妙的微笑，可我只想让角色保持原表情
通过调整强度来试试
能做个支持输入两张图的工作流吗？懒得P一起了
一直都有啊只不过双图不支持线上跑而已。你去看KONTEXT的。然后把输入图片节点换成老狗的就行了
remove grease?
Reducing the intensity is to remove the grease and maintain consistency. 0.6-0.8
this can for mouth like teeth or lips changer expression? or change gesture the hand..?
I edit an image by pasting an object with a white background and prompt it to edit, but it just gives me the same thing with my input image with bad quality..... it doesn't even remove the white background from the image like in your examples....
Try to increase the strength of the lora. You can see the introduction of lora and the recommended strength of the example diagram. Try 1.5 strength and it should have different effects. Qwen Edit is just that. Different adjustment lora weights are required to adjust the effect
If you have any questions, you can send them out and study them together. Thank you for your attention
Same happened with me, it just returns the same image in bad quality. If i increase the strength to 1.5 it just degrades the quality even more.
@felipe781 I have never encountered this problem. I feel like it has something to do with your prompt. Or maybe Lora didn't work.
@felipe781 Same thing happened to me. Try reducing the Lora strength. I went down to .9 and it was working much better
The background ends up with terrible quality if use this…
For example? Let's study it together
Does this works with qwen edit 2509 model?
生成较大尺寸的图时会在中心切出小尺寸的区域生成完整的图，总之很混乱
不支持qwen edit 2509吗？ 尝试在2509上换脸效果不好
Same question, works?
@Milor123 It works with Nunchaku, but needs clarification: Lora doesn't work with svdq-***-qwen-image-edit-2509-lightningv2.0-4steps and svdq-***-qwen-image-edit-2509-lightningv2.0-8steps models. You need to download the svdq-***-qwen-image-edit-2509 model and separately connect the Qwen-Image-Edit-2509-Lightning-4steps-V1.0-bf16 Lora.
有一个比较尖锐的问题，如果是换衣服，模型如何知道新衣服下原本的体貌特征比如胸部大小应该是多大呢？
太强悍了，发现自己一直在用V0.1，刚刚切过来画面质量提升的效果吓到我了，哈哈
Love this model, use it so much, thanks Futurlunatic. I've made an easy to use workflow with background remover and compositor cavas if anyone needs it - workflow link.
The lora works great, but it keeps introducing a fairly noticeable 'checkerboard' type noise/artifacts to the resulting image. Anyone seeing similar results or know of a fix? Not sure what I'm doing wrong :(
what worked for me: change the sampler and scheduler to the: er_sde and beta. also, I increased the steps, but still not sure that it actually matters in that workflow. note: i don't use the upscaler provided there. hope it helps
Details
Files
Put it her_Qwen edit_V2.0.safetensors
Mirrors
Put it here_Qwen edit_V2.0.safetensors
Put it here_Qwen edit_V2.0.safetensors
Put it her_Qwen edit_V2.0.safetensors
PuV2.0.safetensors
Put it here_Qwen edit_V2.0.safetensors
put_it_here_Qwen_edit_V2.0.safetensors
Put it here_QwenEdit_V2.0_V2.0.safetensors
Putithere_Qwen edit_V2.0.safetensors
PUTITHERE_QWEN.safetensors
Put it here_Qwen edit_V2.0.safetensors
Mirrors
Put it here_Qwen edit_V2.0.safetensors
Put it here_Qwen edit_V2.0.safetensors
Put it her_Qwen edit_V2.0.safetensors
PuV2.0.safetensors
Put it here_Qwen edit_V2.0.safetensors
put_it_here_Qwen_edit_V2.0.safetensors
Put it here_QwenEdit_V2.0_V2.0.safetensors
Putithere_Qwen edit_V2.0.safetensors
PUTITHERE_QWEN.safetensors