Many people complaining about the fact that there isn’t a slider on how much we want it to affect faces (mainly on Iris) . But beyond that, my main problem with Iris i that it doesn’t get when a face is unfocused and the main person of a scene is focused. So, it can’t understand the concept that a face is focused and the background face unfocused. So it tries to recover the background faces too without a need for it. Do we have long time before it can understand the concept of focus foreground faces - unfocused background faces
Several people including me have already suggested adjustable face recovery, but Topaz did not react. Maybe adjustable face rec. with a slider is not possible because the model was trained as it is, but then it would be possible to train/triangulate an additional IRIS model without face recovery.
The background thing, sharp or not, faces are recognized or not depends on the source material is contrasty or not. I often import the source into my video editor first, then increase contrast/colors above normal levels and play with gamma. Then I feed TVAI with this higher-contrast video. This works quite good TVAI delivers sharper results.
