Best route to go from SD interlaced to 4K progressive?

I can think of two ways to de-interlaced SD media and upres it to 4K. The first is to de-interlace and up-res to HD first, then upres to 4K. The other would be to go straight from SD to 4K. Is either of these routes prefered by TopazAI? Does it not matter which route you take?



most people claim that they get better results performing this type of upscale in two step solution.
but I would say try both with a 5-10 sec’ preview and see what works best for you.

P.S. forgot, most people claim that going from SD → 4K no matter what method you select, would be disappointment.


I find I get the best result by starting with a de-interlacing run with no upsampling, using one of the Dione models. V1 seems to work best on videos with a lot of small textures, like gravel, for others I use V3.

Once I have the video de-interlaced at original resolution, I use that as the base for upsampling. In addition to saving the time by not having to de-interlace every attempt, it gives me the option of using two other models together if I want to.

Whether to go from SD to 4k or to make an interim upsample to HD seems to vary with different sources, but I do find that the two approaches produce different results.