Upscaling SD Video

Video AI General
1

I am a novice user of Topaz Video AI and not especially an experienced video editor. I make videos of places we visit and upload them to YouTube so we, and family/friends, can watch them at leisure.

I have some Standard Definition video which was shot about 25 years ago. Some of it is in 4:3 format and some of it is in 16:9 format. Some of the video is from a tour of Syria visiting places that were destroyed by ISIS and I would like to improve that video if I can. I realise there will be limitations but any improvement will hopefully be better. So my questions are:

Should I just try to upscale to HD? I assume that trying to go further (e.g. to 4K) would not be worth it.
Which AI model should I use? The default Proteus?
Should I change any of the settings or just use the auto ones?
Should I stick to the Proteus 422 HQ encoder (which produces .mov files, not the most useful) or use one of the others (e.g. H264 High NVIDIA) to produce a ,.mp4 file?

Thanks in advance for any help.

2

Hi, I work on old interlaced SD sequences (720 x 576) that date from 1996 to 2002 I go through hybrid to deinterlace with QTGCM + some filters and VAI with Artemis Medium + Themis + Appolo Sharpen to finish with Vegas Pro

3

So sounds like Hi8 or MiniDV video?

I’ve gotten some impressive results with Artemis. I deiinterlace the video first and make it all square pixels. Many times I use an external sharpen program at the end though, as it makes the video look too clean in certain settings.

5

Artemis Low loses a lot of details in any case. And upscaling an SD video to Full HD also loses details depending on the model, and simply using it without artificial intelligence does not bring more details either.