The clearer the original videos, the more detailed are the resulting ones. The quality of the original videos is very relevant too. I could see that if a 'nograin' version is used as a start, the results are better. Unfortunatelly, mpeg2 is very old and totally outdated, and that limited the best quality possible for the videos (you can have much more quality using modern codecs like h.264 or h.265, and use highter resolutions -UHD, 8K- and highter fps rates) so it's possible to see some pixelations in some situations too, but better than with the other method.

In the first AI remasters i used this method (the Dragons Lair 1 & 2 16:9 versions) but later (and after buying an extra new 8TB hard drive) i decided to use the third method, to process the videos to independient pngs and then process them with a video editor like Vegas Pro with the hightest quality allowed by the mpeg2 codec. Video Enhance AI, for now, uses h.263 and 15-20 MB of bitrate, so the resulting video has in some situations, visible compression errors or pixelations.

As i commented when i was talking about the three methods to do the work, the second one (using Video Enhance AI to process to mp4 video directly) has the problem of not offering options about codec or bitrate used by Video Enhace AI for the final video.

Download brain dead 13 3do