It's really hard to pick a good release on the first try since they usually don't have samples or they're just still images that don't show you if they have bad interlacing.

I pick some lower quality compromise usually, for older shows usually 480p for newer 720, mostly to save disk space. For shows I really like I'll get some BD release. If there's a fansub I'll usually pick that over dual audio releases also.

I often go back to download and archive BD releases of shows I enjoyed.

Care a lot for personal favorites. Preferably 1080p x264 FLAC BDrip with scans of the packaging and audio CDs. Also the BDMV.
For shows I enjoyed, HEVC Hi10P, preferrably OPUS.
In both cases I look for encoders and groups I trust/have downloaded from previously and just watch some parts to check for inconsistencies. Sometimes I download an episode or 2 to compare.

For shows I didn't like much, I go for a smaller 720p release, HEVC if available, and just check that it wasn't a 16yo hackerman encode.

Usually raws.
Quality and completion over size.

Because I like to reminisce about shows I've watched; Admire the cover art and character design of the characters I like; Do not trust companies to keep them; Like to think that some day, may be able help someone who is looking for the files of a show they'd like to watch/archive in the future.

These shows and characters are my little treasures.

And to be honest, disk space, transfer speeds, .. do not really constitute a concern for me anymore. Amazing how far we've come in that regard, compared to a decade ago.

It's the same for manga too. I also own a few, physical, manga/doujis/BDs, but still like to keep a digital archive.

For older anime I don't mind if it's a DVD rip (it just has to be a good one). I find most of my stuff in high-quality anyway. But for recent anime I try to get the best release available since I've a lot of space with my 2x3TB HDD.

I like a decent quality ~ 360/480 for older releases.720 minimum for new releases.

I was curious enough to look up exactly what those encoders do, obviously one part is the tweaking the actual hyperparameters for the encoder (e.g. compression level, crf, etc.) but that part is boring and I'm skeptical that you really get that much extra compression beyond a certain point, especially when disk space is relatively cheap so an extra 50MB is nothing.

Then the other part is the pre-processing of the input source (probably either transport cap, webrip, BD source, or someone else's encode) before you encode it with filters. This part is theoretically easy to understand, basically just applying standard image-processing filters (or if you want to be fancy temporal filters as well), but wow is the actual software they use a mess. You'd think that there's be some nice standard library of image-processing filters, and then community packages or something, complete with documentation as to what exactly the filter is doing from a signal processing perspective. But no, instead it seems to be a mess of poorly documented transforms you just pipeline together.

Like say you want to do some antialiasing on an image. In a sane world you'd assume there's some central repository of filter people have contributed, and it would provide an overview of what exactly it does in terms of image processing, with some example before/after shots. This is what you'd expect in the programming world whenever you have 3rd party libraries, and even domain-specific languages like R and LaTeX have high-quality documentation for packages. But that's not the case at all here apparently, the standard for VaporSynth (or AviSynth) functions is to just chuck them onto a forum. If you're lucky, they might even explain in pseudo-code how they work, but almost none of them will bother to even include a visual before/after comparison. The only examples I could find are for outdated AviSynth [1, 2, 3].


[1] https://www.animemusicvideos.org/guides/avtech31/post-qual.html
[2] https://www.aquilinestudios.org/avsfilters/
[3] https://guideencodemoe-mkdocs.readthedocs.io/encoding/video-artifacts/

I actually set the video size to 3/4 scale, instead of having it cover the entire field of view on my monitor. I think I originally started doing this because I was "used" to how things felt on a smaller monitor, but I think there's also good reason to force most of the frame within your visual center. True it's less "immersive", but I think for anime this tradeoff is worth it. It also makes it easier to scan subs, if you use them.