- Nov 2, 2023
- 10
- 13
Hey, someone have all this videos, the download not working.
You must be registered for see links
Comment
Fakes Kpop Deepfakes (Com2star, MondoKpop, OnceFake, Etc)
- Thread starter Mina Simp
- Start date
- Watchers 1,920
- Dec 6, 2023
- 4
- 67
hello, just hoping someone saved this yeji gem and could kindly reupload, it was initially in a bunkr but it had since been deleted. i need a hero 
and for the trouble i have a chaewon vid, not too sure where i got it, the original code is the title, enjoy
and for the trouble i have a chaewon vid, not too sure where i got it, the original code is the title, enjoy
3
Comments
- Feb 28, 2024
- 8
- 26
Once again, good news has arrived.
So anyone who have access to the Baidu link, please contact me.
So anyone who have access to the Baidu link, please contact me.
2
Comments
Q
- Mar 4, 2023
- 7
- 368
Found some next-gen tools
You must be registered for see links
19
Comments
View previous comments…
All should work, as they are also all more on the paler skin tone side.
TKTube has all the vids you need~
I
they only have 720p unless I make an account(?), sukebei is better tho and I use 123av and missav to scout good source.
- Sep 29, 2022
- 2
- 2
- Dec 3, 2024
- 3
- 1
You must be registered for see links
안녕하세요.
수지의 풀 비디오 파일을 가지고 있는 사람 있나요? 공유해 주시면 감사하겠습니다.
Comment
Do anyone know the code for this? Thank you in advance
You must be registered for see links
1
Comment
- Mar 4, 2023
- 7
- 368
Since she got new hair color... and the nj you requested
I mean its always easy for individuals. For multiples I need recommendations on clips where they can stop kissing so the app won't recognize two as one...
You must be registered for see links
ssss
I mean its always easy for individuals. For multiples I need recommendations on clips where they can stop kissing so the app won't recognize two as one...
13
Comments
View previous comments…
- Nov 26, 2024
- 11
- 140
Ive sex interview
I hope someone uploads the full version of these videos. I beg of you
You must be registered for see links
I hope someone uploads the full version of these videos. I beg of you
You must be registered for see links
You must be registered for see links
5
Comments
View previous comments…
- Mar 4, 2023
- 7
- 368
how are the new ones
4
Comments
I
Hmmm the only thing that bothers me are those bangs/strands of hair messing the consistency maybe you could tweak something on the blending there, but overall model looks good even in multiple models
Just curious is the new tool you uses support VR- Nov 5, 2023
- 9
- 12
You must be registered for see links
HMN-566
Is there anyone here who can do magenta AI for me?
13
Comments
View previous comments…
8
how many iterations did it take to get this and how do you source proper facesets. i had alot of trouble with it, including the merging part
I
1m iterations
sourcing facesets is difficult, there are a lot of ways but here's mine:
1. I started selecting/downloading videos from her tiktok, twitter, youtube (@hapycb) and their QWER YT channel (with good closeups)
2. Compile and Cut the videos with closeups and good yaw distribution, then export. (if you have AI Video enhancer use it and skip No.4)
On DFL:
3. '2) extract images from video data_src' then '4) data_src faceset extract' (in here you will set the size depending on your pc's computing capabilities, for me I go with whole face/ 512x512 cause my pc is trash) - Go to /workspace/data_src/aligned folder and delete blurred/extra face images.
4. This is optional but this will increase better facial artifact in training:
- select '4.2) data_src util faceset metadata save' and locate 'meta.dat' inside aligned folder, make a new folder on this path /workspace/data_src name it enhanced_aligned and paste the meta.dat file inside.
- If you have Gigapixel 7 or 8 enhance the whole aligned folder and output with same input setting (512x512) - you could export the enhanced imgs to 'aligned_enhanced' folder
- once finished, delete ng aligned folder and rename "enhanced_aligned" folder to "aligned" then '4.2) data_src util faceset metadata restore'
now you have enhanced faceset!
but in the merging part you mention, I guess there are misaligned on labeled dst? would you like me to explain how to prepare data_dst?
There's also Training, Xseg (optional) and Merging.
sourcing facesets is difficult, there are a lot of ways but here's mine:
1. I started selecting/downloading videos from her tiktok, twitter, youtube (@hapycb) and their QWER YT channel (with good closeups)
2. Compile and Cut the videos with closeups and good yaw distribution, then export. (if you have AI Video enhancer use it and skip No.4)
On DFL:
3. '2) extract images from video data_src' then '4) data_src faceset extract' (in here you will set the size depending on your pc's computing capabilities, for me I go with whole face/ 512x512 cause my pc is trash) - Go to /workspace/data_src/aligned folder and delete blurred/extra face images.
4. This is optional but this will increase better facial artifact in training:
- select '4.2) data_src util faceset metadata save' and locate 'meta.dat' inside aligned folder, make a new folder on this path /workspace/data_src name it enhanced_aligned and paste the meta.dat file inside.
- If you have Gigapixel 7 or 8 enhance the whole aligned folder and output with same input setting (512x512) - you could export the enhanced imgs to 'aligned_enhanced' folder
- once finished, delete ng aligned folder and rename "enhanced_aligned" folder to "aligned" then '4.2) data_src util faceset metadata restore'
now you have enhanced faceset!
but in the merging part you mention, I guess there are misaligned on labeled dst? would you like me to explain how to prepare data_dst?
There's also Training, Xseg (optional) and Merging.
I
I don't usually take requests but the Jav is good and has a lot of closeup. unfortunately, my model cannot handle those and the face may become unrecognizable after merging :< (I could try tho)
8
for me even after training to 1m+ iterations the faces seem to be blurry, dont align well with the model and doesnt exactly look like the face i want. so i just gave up with dfl
8
i did do the same thing for sourcing facesets, using youtube videos and others, but ill try with a the topaz ai enhancer this time
I
It's one of the misconception in merging is the 'Super Resolution' below 25 you could see the model is too blurry, adjust it to 40 up to 70 you could see the difference there.
The misalignment has to do with preparing your data_dst.
The misalignment has to do with preparing your data_dst.
- Sep 11, 2024
- 4
- 6
I don't get satisfactory results, so is there anyone who can help me
I use deepfakelab and 4080super 7800x3d
The source video was compiled from TikTok, and the dst video was just about anything on the face
If I have to use xseg, it's hard, but I'll try. Or can you recommend a program other than deepfakeelab
I use deepfakelab and 4080super 7800x3d
The source video was compiled from TikTok, and the dst video was just about anything on the face
If I have to use xseg, it's hard, but I'll try. Or can you recommend a program other than deepfakeelab
23
Comments
View previous comments…
I
So here, you are mentioning Xseg DST aligned images? can you screenshot your specific problem?
If it's unavoidable to include the whole face yeah you could only label important parts of the face, just try to include the whole face as possible but don't delete aligned images just leave others unlabelled. In here, if you have 500-1k aligned dst images to label just make at least 1 label per 10 unlabelled images don't just clump them.
Now you can proceed to '5.XSeg) train' and train for at least 3-5k iters, then apply. By applying the Xseg mask you can see the effect when merging.
Here's an example/guide, and you can apply my instructions too:
If it's unavoidable to include the whole face yeah you could only label important parts of the face, just try to include the whole face as possible but don't delete aligned images just leave others unlabelled. In here, if you have 500-1k aligned dst images to label just make at least 1 label per 10 unlabelled images don't just clump them.
Now you can proceed to '5.XSeg) train' and train for at least 3-5k iters, then apply. By applying the Xseg mask you can see the effect when merging.
Here's an example/guide, and you can apply my instructions too:
A
I've been too busy playing games lately, so I haven't been able to train it. I've repeated it about 120,000 times, and it feels like it's working but also not working at the same time. Even if it fails, I plan to repeat it over a million times and make adjustments afterward.
- Dec 27, 2021
- 2
- -1
You must be registered for see links
Does anyone have this vid?
Comment
- Nov 3, 2023
- 11
- 739
Alright guys, i havent ever uploaded before. I am thinking of posting some of my collection here, but Gofile isnt uploading my files. and bunkrrr isnt allowing new registration. any help how to upload files for all of you?
2
Comments
- Nov 3, 2023
- 11
- 739
Merry Early Christmas! my first time uploading. haven't seen these mr.hdwork being uploaded here. I have tons of Newjeans stuff if anyone is interested, i can upload those too. I even have a rare Eunchae vid, but im not sure its time to upload that one yet. Do let me know. In the meantime Enjoy
These are 11 files of Karina, Dahyun, Yujin, Wonyoung, Rose & a cool Haerin Bj
Size: 7.63 GB
These are 11 files of Karina, Dahyun, Yujin, Wonyoung, Rose & a cool Haerin Bj
Size: 7.63 GB
You must be registered for see links
14
Comments
View previous comments…
N
all videos you shared is actually get it from a freesite on https://celebsrule35.com/ , there are more videos on the site actually
Yeji BBC.mp4 | Bunkr