Got questions or tips about the Conversion process? This is the place to discuss them.
Read the FAQs and search the forum before posting a new topic.
This forum is for discussing tips and understanding the process involved with Converting faces from your trained model.
If you are having issues with the Convert process not working as you would expect, then you should post in the Convert Support forum.
Please mark any answers that fixed your problems so others can find the solutions.
- Posts: 10
- Joined: Sat Jan 04, 2020 6:30 pm
- Has thanked: 6 times
- Been thanked: 1 time
Hoping for some help here. I've managed to reach the stage "Manually Fixing for Converting"
Now having some difficulty. Are there any videos to show how to do the part with the DOS screens. It seems as though you need to move those lines some how depending on which way the face is facing.
I've already deleted 'multi faces' (ie more than one face in the frame) but having trouble with "Manually Fixing for Converting / Fixing alignments " section. I'm not to sure what fixing alignments means or how to change those green lines or how they are supposed to show on the face.
ANy help would be appreciated (also, any other tips to avoid a first time flop) . All the scenes in the source face clip have the same lighting and background. The face does move in multiple angles including side shots; i have a feeling this fixing alignment thing is important.
- Site Admin
- Posts: 622
- Joined: Fri Jul 12, 2019 12:49 am
- Location: San Francisco
- Has thanked: 3 times
- Been thanked: 159 times
Generally you just want to make sure that the "lines" as you say (really, the alignment landmarks) are in roughly the right place. To edit you can hit the "m" key to enter edit mode, then click on the face/move the face box". You cannot directly move the alignments themselves as they're generated by a neural network.
- Posts: 43
- Joined: Mon Dec 09, 2019 7:00 pm
- Has thanked: 4 times
- Been thanked: 2 times
My two cents, after playing about for a few months, Start with your experiments with "face on" videos and then learn the limitations of what the alignment alogrithms can do. Really 90 degree side on's are unlikely to ever be properly aligned and not even possible to manually "click to align". Is it a face or a coffee machine, the aligner will not know ??
Similarly, to do acute viewing angles requires significantly more training, regardless of alignment landmarks.
Start simple (face on), then push the boundaries. Don't expect to faceswap a full action video at the click of a mouse button !
Message from moderator
EDITTED TO REMOVE INAPPROPRIATE MATERIAL