Faceswap Forum

Not worth it together in one system, no. Mixing generations is not recommended. As a second system, it may be worth it.

If in the GUI, please look around your options. If using the command line try

python faceswap.py (category) -h

Don't use on-the-fly extraction. Instead run the extract step and make sure you select the alignment file when converting.

No difference. The software will autodetect if NVLink is available and use it. There is no difference in operation, just a small speedup when using multiple GPU.

Unfortunately, 1gb is not enough vram to use Faceswap. You'll have to use CPU or get a GPU with more vram.

We unfortunately don't have a universal scaling option since it leads to very poor results in our opinion. Instead we recommend you try out the preview tool and use the blur options. These can help cover the artifacts.

For Faceswap i recommend the multiple smaller gpus and a mid/high end cpu. Ryzen 5/7 and 1-4 2060 supers is a great setup.

It's important to note that DFL does not support multiple gpus, only Faceswap can use more than one GPU.

Cpu is still heavily used for things like augmentation, but it's far less important than the gpu.

Yes you can rename files. If shift isn't working for you, try holding ctrl. Both work slightly differently but let you select now than one file. Worst case scenario is select one file at a time, copy the box into a new window, as you add each file then copy the whole collection back.

Mainly pci-e lanes. Most setups could actually go slower with 4 cards instead of 3 since it'd drop the number if pci-e lanes available for each card.

Faceswap doesn't use Json anymore, and saves alignments in .fsa format. Please read the guide which mentions fsa files and how to use them.

See the mask section at viewtopic.php?f=5&t=27 for info on masks.

The only time you might not use a mask is if you're manually compositing the video, in all other situations you'll want a mask of some sort.

Unfortunately you're going to bed to understand that technical stuff to get pix2pix working. It's not like Faceswap where you're able to get running without technical knowledge.

Unfortunately that's just the nature of academic projects.

No, this is really not how the alignments are used. Even when copying from a previous frame, we recommend just using that as the starting point of detecting so that you get valid alignments for the face, otherwise you are just adding bad data that will confuse the model.

Renaming them is the correct way. You can simply add a number to the beginning of each set or you can use the alignment tool to rename them back to their original file names. Either way, only clean the alignments file AFTER you have rejoined the files into a single folder.

This really isn't what Faceswap does. I don't know of any toolkit made specifically for this but any VFX/compositing tool like AfterEffects or Hitfilm could do this easily. You may want to check one of those out.

Yes, and our guide covers both the alignment tool and the sort tool. The sort tool is definitely going to be the faster and easier option for doing this.