Mr DeepFakes Forums
  • New and improved dark forum theme!
  • Guests can now comment on videos on the tube.
   
Total Likes Received: 4 (0.04 per day | 0.56 percent of total 715)
(Find All Threads Liked ForFind All Posts Liked For) Total Likes Given: 16 (0.18 per day | 2.24 percent of total 715)
(Find All Liked ThreadsFind All Liked Posts)
Reputation:  0
Details

Akanee
(DeepFaker)
DeepFaker

Registration Date: 12-24-2018
Date of Birth: Not Specified
Local Time: 03-25-2019 at 02:37 PM
Status:

Akanee's Most Liked Post
Post Subject Numbers of Likes
RE: DeepFaceLab Explained and Usage Tutorial 2
Thread Subject Forum Name
DeepFaceLab Explained and Usage Tutorial Guides
Post Message
(01-21-2019, 12:21 AM)SPT Wrote: You are not allowed to view links. Register or Login to view.Hello, new to this, tried deepfaking before but it was too hard before DeepFaceLab, so thank you. Now I have obtained a satisfying result on my second try using LIAEF128 with 50000 Epoch (tho it was already good enough at 25000).

I have 2 questions :

1\Is it possible to have any way to work at 256 or even 512 resolution now with DFL ?

2\If I want to use my trained model on another dst video, how do I proceed ? What do I keep from the model folder ? (I'm thinking using another video from the same dst girl would be compatible with my model) Or do I keep my model folder as it is and just add the new .png files from my new dst video to the training ? How will it work ? Will it keep what it learned from the first dst video and add new infos from second dst video to the model ? Or will it "forget" infos from first dst video and only use infos from 2nd dst video ?

I'm a bit confused on this matter, hope you can understand what I mean.

Well 512 x 512 not possible right now. 256 stay the best in 2019.

For trained model keep it with the model folder BUT move the DST folder and dst. Mp4 on another location on your computer or if you don't need it anymore just delete it. Place the new video DST.mp4 on workspace folder and then do the extract DST png fullfps and then extract face. And train it. It will keep the SRC folder but will train the DST aligned face. And voila ! That's all !

Akanee's Forum Info
Joined:
12-24-2018
Last Visit:
03-04-2019, 03:57 PM
Total Posts:
61 (0.67 posts per day | 1.45 percent of total posts) Find All Posts
Total Threads:
3 (0.03 threads per day | 0.28 percent of total threads) Find All Threads
Time Spent Online:
18 Hours, 21 Minutes, 35 Seconds
Thanks/Likes:
Given: 16 | Recieved: 4
Members Referred:
0
Akanee's Contact Details
Additional Info About Akanee
Sex:
Undisclosed