Faceswap Forum

I am filming klipps that I extract. I wonder what is most important when setting up lighting. I have 4K resolution and very good fokus. However I use a greenscreen and to avoid shadows on the background I use very soft lighting with softbox. I have done some training with data obtained this way and I think it would be better with lighting that gives higlights shadows that more clearly shows the shapes of the face rather than uniform light in the face. So, maybe use a good general lighting and add a smaller lightsource as well to give the images more dynamic? I can see that loss when training is generally poorer with my data then data obtained from a DVD that from what I can see often has more dynamic in the way the face is lit.

Loss is a poor measure, especially if you control one side of the data.

I'd say that the most important thing is Variety. Variety is key because the model can learn all the different types of lighting itself.

I'd suggest trying to move a light across the face to get different angles. Getting some filming outside and in all the different lighting conditions can really help too.

Thanx, yes I know loss is a poor way to measure but if I run several similar training sessions I still get some feel for what it means specially if I combine this with the actuall converted result. I imagine that resolution must be important if you want to get a really realistic result, when you even get a feel for the skin texture? I do have many angles in my klipps but I can se that faces from DVD often has more highlights and shadows in the faces.

I have filmed now with all possible angles of my face and it looks very mutch better, training is also much faster, thanx!