I'm playing with upscaling some of my DVDs (the ones I want in higher quality than normal) from 480p or 480i to a couple of formats: 480p60, 720p60, 1080p60. There's a couple of reasons and *theories* that I have about doing it:

Update / spoiler alert: increasing picture size does not equate higher quality, it just slows down your encode a lot. Use higher CRF for good quality. :)

I *have* to encode my source video at some point, because I can't *consistently* get a DVD → VOB or DVD → MKV lossless copy somehow that is clean, and doesn't have playback issues somewhere, at some point (sadness). I could write more code to do it myself, but since there's so many tools out there to encode it, and I don't want to debug my own, I'm passing on it. So, encode we must.

Another reason I *have* to encode is that source material is interlaced or telecined, which needs to be fixed as well.

Playing back video that's a 480p source on an HDTV (my Sony rear projection TV is 1080p) is going to get displayed at 1080p resolution anyway, so why not go ahead and do it myself instead of having their software zoom for me? This is the argument that makes the most sense to me.

I like the idea of having a constant framerate (60fps) over a variable one. In my mind, it just makes more sense, it gives my warm fuzzies, and I've *always* had problems from day one trying to figure out VFR and the film → video transitions when encoding stuff, especially with TV shows on DVD. I'd rather just have one constant framerate. Going to 60 makes more sense than going from 23.97 or 29.97 to 30 anyway. Might as well double it so there's no minor weirdness there (again, makes sense in my head).

The filesizes of 480p60 are so far averaging 5% smaller than 480p VFR. Interesting.

The storage dimensions are going to be larger. There's almost always some kind of cropping going on from the source material, anywhere from 2 to 20 pixels on each side is a good average. That means what was originally 720×480 is now going to be around 710×476 or so on, so there's just a slightly smaller size. I can “reclaim” the space, and set it to the actual 480p (720×480) picture size at the same time.

My reasoning behind that one, is that with a larger amount of space to store the image in, the quality of the picture is going to increase as well. That's an actual reason that is obvious and doesn't need any defending to begin with. With upscaling, the same applies, but on a larger scale. Instead of 720×480 storage space, I've got 1920×1080. With more space, that means better quality since there has to be less differences between the frames, right?

The final reason though is that I just want to standardize all my content – all the same size and framerate, in addition to CRF, H.264 level and profile and so on.

480i Source

For my testing, I'm using Teen Titans. It's telecined video, so I have to run it through video filters anyway. Here's a shot that is a progressive frame from the opening sequence, encoded at CRF 14 with variable framerate. The original was cropped to 718×480, so this one is bumped back to 720×480. Not big enough change to notice.

And here's the same frame at 720p:

Finally, the 1080p version:

After doing my testing, it looks like the differences to the quality are nearly impossible to tell, and not worth the 300% increase in filesize from 480 to 1080.

Here's the original at a CRF value of 1 … I can't really tell the difference there, either, for this shot. So, maybe I picked a bad source to begin with anyway.