Quality question, 264 vs 265 with same disk space

[keybounce]

I'd like to know what, if anything, I am doing wrong.

What I expected: x.265 should compress better (less space) for the same quality, or better quality for the same space.

What I found: After doing quite a bit of testing for a TV show on DVD (with de-telecine), an x.264 quality of 21.5 is about the same quality as x.265 of 17, and about the same disk space (92.2 mb vs 91.2 mb). (Show is 636 x 480 after cropping).

Both of these are on "faster". Both are about the same disk space. Both are about the same quality.

What is the reason for wanting to use x.265 instead of x.264?

[rollin_eng]

Could you please post your logs, instructions can be found here:

viewtopic.php?f=6&t=31236

[Deleted User 13735]

Same quality in less space (generally) is correct.
Better quality in the same space is not a reasonable expectation with marginal source, such as teevee recorded on DVD.

[keybounce]

With the TV, as soon as I cranked the quality up to where I was not seeing edging artifacts on clothing, and still able to see lines on ships in the CGI, the size was the same. Both at "fast", one at Q21, one at Q16, and the size was also the same.

I'd be happy with "same quality in less space for teevee recorded on DVD". I know that "better quality" doesn't exist on the source.

(grumble grumble Crusade was filmed in widescreen, and the intro/credits are on the disk in widescreen, but the show itself is in "clipped edges".)

NB: I took "264 @ Q21" as the general standard for ripping a DVD. I was surprised to find that matching quality in 265 also matched size. Equally, I was surprised to find that Q numbers for 264 are not the same as Q numbers for 265.

[yoyomonkey2000]

how about leaving the slider RF20 for DVD source on both... x265 will yield a smaller file size.

[keybounce]

But the 265 file, although smaller, is significantly worse quality on playback.

To get the same quality, I have to go to a lower number, which makes the size almost identical.

[Deleted User 13735]

Without the requested and required logs for BOTH encodes, I don't see how ANYONE could hazard a guess as to your rather unusual problem. Otherwise, it's just another story...

Look up!

[keybounce]

OK, here is a fresh set of encodes, this time with disk 1.
So testing, crusade disc 1, episode 1, chapter 3. Takes about 1 min, 50 seconds to do a subtitle scan (to determine that there are no foreign language subtitles). All encoding tests are done with detelecine and "fast decode".

Testing with 264, medium quality as my baseline. There is no visible difference between Q20 and Q21. Q20 takes just a few seconds longer to encode than the subtitle scan; Q21 is the same speed as the subtitle scan. So I am working at the same speed as the disk reading.

Logfile: https://puu.sh/vSpVf/2c1c5cd6a8.txt

Size of chapter 3, 264, Q21: 92.4 MB

Testing with 265. "Faster" gives me almost the same speed as 264's "medium", at Q19. But quality is definitely worse. Lowering Q until the quality is a match brings me to 265, Q16.5. Same visual quality, size: 92.4 MB.

Same quality, same size. Longer encode time.

Logfile: https://puu.sh/vSq2F/5996427879.txt

Now, speeding up the encoding, to make the comparison as equivalent as possible:
Very fast ... still significantly slower, same size. Logfile: https://puu.sh/vSsvW/bc0e742d20.txt

Super fast: ... interesting. Now it's at full disk speed, and *SMALLER* - 75.8 MB instead of 92.4 MB. Visuals: ... actually better???. Logfile: https://puu.sh/vSswU/0616324eaf.txt

265, superfast, Q16.5, is what it takes to look better than 264, medium, Q21, in less space, and "less" time (actually limited by DVD i/o time). Attempting to use a "faster" preset makes it bigger. This is based on a 480p source, 636 pixels wide (and go figure: the CGI scenes are 636 (crop 2 on each side), the live action scenes are 632 (crop 4 on each side). Encoding done at 636).

(NB: My next pair of disks will include Kubo and the two strings, which should be a really "full quality" DVD; other than that, my encodings would be video editor outputs, starting with OBS Q19 recordings).

[mduell]

Sounds like your quality settings are so high you can't actually tell the difference.

[keybounce]

Hmm? I'm not sure I understood that.

For the DVD, in 264, Q 21 is the point at which the quality is there. Anything worse looks bad.

For 265, I'm just looking to see where that same quality is maintained. Oddly, I can actually get a bit better at 16.5 -- the source has "color banding in low light levels" (sorry, I don't know a better way to describe it), and that banding is actually seen at that point (264 / Q 21 does what I'd describe as a smoothing in that banding area).

I suspect that an HD recording would be different; after I rip these (disks 1 and 3), I'll get disk 4 and Kubo next (already got disk 2 ...)