Very important and a must read.

尊敬的用户，

我为之前出现的bug向您致以诚挚的歉意。经过调查发现，这些问题是由于我没有及时上传最新版本所导致的。感谢您的理解与支持，我们会尽快解决这些问题，并确保未来不会再发生类似情况。

再次感谢您的耐心与支持！

诚挚的，
baicai

Dear Users,

I sincerely apologize for the bugs you encountered previously. Upon investigation, I found that these issues were caused by my failure to upload the latest version in a timely manner. Thank you for your understanding and support. We will resolve these issues as soon as possible and ensure that similar situations do not occur in the future.

Thank you once again for your patience and support!

Sincerely,  

baicai

Q: Why do this?

A: Because the midjourney-generated characters have a kind of oil painting texture, and they need to be fixed with sd every time. So I developed a workflow to fix them directly.

Q: Why isn't finger repair good?

A: Because this process only focuses on realism, it doesn't handle making unreasonable things reasonable. In other words, it can't fix a distorted finger back to normal, it can only add blood vessels to the original hand.

- Recommended big model: [majic V7](https://www.liblib.art/modelinfo/bced6d7ec1460ac7b923fc5bc95c4540)

- Recommended Vae: [Realistic difConsistency Negative (Pack)](https://www.liblib.art/modelinfo/232ff495f7c14381910dc6f4df78a7ab)

- lora: [YFilter_PortraitEnhance recommended](https://www.liblib.art/modelinfo/ec6239d3fcdc46beb8aefbd587e291b7)

- Recommended high-definition repair node. If you want the image to be more realistic, even including pore details, please use this scaling node. [Link](https://openart.ai/workflows/seven947/1minute-8k-upscale/1IPTks1gL7v0EPmvsMcx)

My nodes use G-DinoSAM to recognize objects, not just redraw faces and such. You can freely copy and paste, freely combine them.

1. If the whole head is recognized when identifying hair, please adjust the threshold parameters in G-DinoSAM.

2. After all, it's automation. Please don't expect too much. I can only try to replicate manual operations, and post-processing in Photoshop is unavoidable.

3. If there is anything that is not user-friendly or any suggestions, please leave a comment actively.

My links:

[My Personal Website](https://www.zhengjiyuan.top)

[Douyin (China)](https://www.douyin.com/user/MS4wLjABAAAA478CnKTrQH1n87AkVVEPcwKV-MQsyUQwnxzGLDm_LA0)

[Xiaohongshu](https://www.xiaohongshu.com/user/profile/5fc2948e0000000001000b4e?xhsshare=CopyLink&appuid=5fc2948e0000000001000b4e&apptime=1716350434)

[liblib](https://www.liblib.art/userpage/c0e1c819d36c4bce9b077e04f9eaf693/publish/image)

[civitai](https://civitai.com/user/1637083533489/posts)

Update Log

V2.0.3

- add wd-tagger on face.

V2.0.2

- Optimized update parameters.

V2.0.1

- Optimized update parameters.

V2.0.0

- Change: Change Yoloworld ESAM to G-DinoSAM.

- Update: Brand new layout.

- Update: Brand new parameters, smoother and gentler.

- Add: Soften the mask.

V1.0.8

- Fix: the previous idea was wrong, cutting through the mask can not achieve better results, but will slow down the running speed. Change mode to expand mask scope.

- Fix: It seems that it can be fixed well in HD without tile, and tile also comes with the problem of darkening the skin.

- Fix: Adjusted crop size and mask blur, face fix removed mask overlay.

V1.0.7

- Added: Shortcut switch to directly shut down unwanted node groups.

- Fix: Change the parameters, the character will turn black, but the details are much more, I don't know why.

V1.0.6

- Known problem: Arm cannot be identified correctly. The thigh cannot be correctly identified.

- Repair: The previous node could not effectively detect the neck, resulting in neck fault. The scene of the neck can be met by cutting each other between masks, and all 30 images in the test can be used normally.

- Fix: the previous node is not valid for clothes, may detect pants and the like, through the mask cutting to meet the detection of clothes only scenario, the test 30 pictures can be used normally.

- Fix: Change keywords and Settings to make pants more accurate.

- Abandoned: The high and low frequency node is officially abandoned.

V1.0.5

- Fix: Seems to fix the problem that the face will turn black.

- Fix: It seems that there is no need for high and low frequency modules, so you can get a picture in one step.

- bug: I was wrong, I shouldn't use components to pass, it will lose image quality.

V1.0.4

- Known issue: Face gets darker as redraw magnitude and facial details increase.

- Fixed, the third node connection error.

- New node style adding pictures, avoiding the error in the following steps which will lead to the need to start the sample movie again.

- New multi-player judgment, false is multi-player, true is single.

- Face repair high and low frequency nodes, over sharpening and lazy to adjust the parameters, two pictures into the PS with a mask erase.

V1.0.2

- Bug fix, image delivery is not in order, but all in parallel.

V1.0.1

- Modifying the parameters allowed the masquerade problem to be mitigated.