As I have been in the process of documenting an outline of my approach to setting up video enhancement using VEAI 3-7b, I’ve come across several spots that need attention to make VEAI 3 into a better enhancement tool.
The old saying regarding computer accuracy is/was “Garbage in = Garbage Out.” Originally, that meant that the best computer programs could not do “computer perfect” work if the input data was flawed. Back at that time, the problem was mainly due to having information keyed in by hand; which was a very error-prone process.
VEAI is meant as a utility that uses AI techniques for cleaning up problems in video. The enhancements work fairly well, but these enhancements work best when you feed them clean video.
This is a discussion of what is needing in a module for taking source video and doing a very thorough format conversion, *deinterlacing, (if necessary,) and decompress/de-artifact into a lossless video stream that can be run through VEAI to obtain optimal results.
It should live somewhere in VEAI up near where the Stabilize and Deinterlace filters do. - In fact, it would probably be a superior replacement for the current Deinterlace.
Currently, in order to get the input clean enough, it requires processing by 3rd-party apps to get it cleaned up sufficiently to be effective.
It would also require adopting a lossless video codec that could be used for repeated passes through VEAI without degradation. - That same codec should also be one that can be read and written by other video editing applications.
Please Reply with Comments, and vote it up if you agree.
Thank you.
