Mr DeepFakes Forums

Uploads are currently buggy. If recent uploads are not viewable, it will be fixed within 24-48 hours. Older videos should be working. Please CTRL+F5 before clicking play. (June 27)

Guests - Sign up to view links and videos on the forums. Please check your spam mail. (February 26, 2018)

  • New and improved dark forum theme!
  • Guests can now comment on videos on the tube.
   
  •  Previous
  • 1
  • ...
  • 134
  • 135
  • 136(current)
  • 137
  • 138
  • ...
  • 244
  • Next 

Shoutbox archive

Go to last read | Mark all read

avatar
@Cerciusx Sorry, wrong tag.
avatar
@Cerciusx There are many members of this site that know how to make DeepFakes. Any member whose name appears in an orange color is a DeepFake creator who has posted their work on the tube portion of the site. There are also many other members who know how to create DeepFakes, but have not publicly posted their work. Their names still appear red in color.
avatar
@sliverboi You can ask for a private fake in the requests section of these forums. If you find a creator that will accept your project, then you will need to communicate with them privately. MrDeepFakes does not support or condone DeepFakes of non-celebrities.
avatar
anyone know how to make your own deepfakes
avatar
@Cerciusx One of the things I'm trying out, as I "speak", is this: I stopped training, backed up the model folder, and started training again, but this time A to A. I'm going to let that cook for a couple of hours. Then I'll repeat, but train B to B for a couple of hours. Then I'll go back training A to B. I don't know if that makes any sort of logical sense to do or not, but it's a weird idea I had. I'm thinking it's similar to training over a base model. It might drop the loss values.
avatar
@Cerciusx I absolutely hate to see you throwing money into something that may, or may not, be working. I've been having issues similar to yours though. Not getting to the results that I really would like.
avatar
@Cerciusx It's like I was saying here yesterday. So much information is conflicting, and there isn't a whole lot of really useful practical advice sometimes.
avatar
@Cerciusx Well, with Faceswap Original model, you can run a high batch size. I can run 128 on an 8 Gb card. But there is conflicting advice on the batch size scenario. Many users say to start the batch size low, then increase it later. But at least 2 devs on the discord say to start high, then decrease it later. So honestly, I don't know which method is better. I've only been trying the method suggested by the devs, so far.
avatar
hello is it possible to request a deep fake for non Celebrity
avatar
@Endalus anyway if not, its cool. I appreciate your input
avatar
@Endalus I'm a bit short on cash but in exchange for training a model for me (original) I can kick you back like a 50 dollar amazon gift card I've never used. I mean if you can spare the time.
avatar
@Cerciusx true enough
avatar
@Endalus I mean Ihave an android phone but I imagine trying to set up faceswap on a phone would be a nightmare
avatar
@Cerciusx there's an android and iOS launcher too, if you've got a device or one of the linux android emulators
avatar
@Endalus I'd love to use shadowpc but its just not compaible. totally different library versions and package manager
avatar
@Cerciusx ah, bummer. I know that pain. I was a Slackware guy for years
avatar
@Endalus I've asked around if anyone has any free time to train this stubborn model for me but nobody seems to have any time lol
avatar
@Endalus Yeah I just checked it out. not available for any non-ubuntu based linux distro. I'm on arch. bleh
avatar
@Cerciusx proprietary launcher. you just put in your uname/pw, log in, hit start, and it opens a window that streams your VM
avatar
@Endalus huh. How do you connect to it. does it run in-browser or you you have to mess with rdp

  •  Previous
  • 1
  • ...
  • 134
  • 135
  • 136(current)
  • 137
  • 138
  • ...
  • 244
  • Next