PDA

View Full Version : Am I wasting encode time with this Handbrake preset?




obsidian1200
Aug 18, 2011, 06:21 PM
Hello forum,

So, I've decided to take the plunge and encode all my DVDs (roughly 500 of them) to .m4v using Handbrake. My friend gave me a preset a h264 string (copy and pasted below), and I've tried it out on a couple of movies. They take about an hour and a half or more to encode (depending on whether there's a deinterlace or other correctional filter on), but when I try them using the :apple:tv2 setting, the encode time drops to about 1/3 of the other preset.

My question to the forum, in particular those who are more video savy than I (I'm guilty of just letting my DVD player do all the thinking for me :P), is whether this string is worth the extra time, or if I'm better off using the :apple:tv2 preset like everyone else seems to default to.

I'll be viewing the movies on my iMac, my friends' ps3s from time to time, and an iphone 4. If I need view the videos on other devices, then I'll re-encode using an appropriate preset.

Here's the string: ref=5:mixed-refs=1:bframes=16:bime=1:weightb=1:direct=spatial:b-pyramid=strict:me=umh:subq=9:analyse=all:8x8dct=0:trellis=2:nr=150:no-fast-pskip=1:merange=32:no-dct-decimate=1:b-adapt=2:deblock=-3,3

As far as I can tell visually on the movie and TV episodes I encoded last night, there's not really a difference on the iMac when I have them stretched to fullscreen. I'm more or less worried about blocking that might result, since I had issues with that in the older presets of HB.

Any advice/input is appreciated. Thanks!



roidy
Aug 19, 2011, 04:33 AM
The reference frame value is too high and the b-frame value is set to a stupid value, 16 is way to high. I would just use the ATV2 preset, thats what it's there for. Also the deblocking values are a bit too low, I never go below -1,-1.

slothrob
Aug 19, 2011, 06:17 AM
I'm more or less worried about blocking that might result, since I had issues with that in the older presets of HB.
The aTV2 preset is pretty darn good. A lot of issues you run into will depend a lot more on how well the source video itself was encoded. I've never run into any real deficiencies in the aTV2 setting, short of the expected Detelecine and Decombing issues inherent in many video sources, which are fixable by checking the appropriate boxes in HB.

If your TV is big and good, you may be able to see some slight artifacts, such as blotchiness or crawl in large areas of solid color, like walls, during still shots. On the other hand, depending on what your TV's noise reduction settings are, these minor artifacts may not even be detectable.

For SD video, you'll probably see a little benefit from knocking Constant Quality down to 19, making it virtually indistinguishable from the source in most cases. However, if you don't spend much time looking for minor differences, you may not even pick up on the improvements.

obsidian1200
Aug 19, 2011, 09:34 AM
The reference frame value is too high and the b-frame value is set to a stupid value, 16 is way to high. I would just use the ATV2 preset, thats what it's there for. Also the deblocking values are a bit too low, I never go below -1,-1.

Yeah, that's what I thought too. I actually talked with the guy that gave me the preset last night, and he told me those values were like that to compensate for being unable to use the 8x8 transform with his devices. It's probably not terrible for devices that don't support high profile h264, but since all my devices do support it, I thought that the values were over kill.

Thanks!

The aTV2 preset is pretty darn good. A lot of issues you run into will depend a lot more on how well the source video itself was encoded. I've never run into any real deficiencies in the aTV2 setting, short of the expected Detelecine and Decombing issues inherent in many video sources, which are fixable by checking the appropriate boxes in HB.

I figured as much; if it wasn't that good, then people wouldn't use it as much as they do :) I'm glad that HB will tell you whether or not it sees interlacing or telecining in the source via the activity log (but it can't tell which one it is, yet). A few of my DVDs have some heavy interlacing/telecining, so I'll need to play with those settings. I've actually gotten good results by combing the decomb filter with the detelecine filter.

If your TV is big and good, you may be able to see some slight artifacts, such as blotchiness or crawl in large areas of solid color, like walls, during still shots. On the other hand, depending on what your TV's noise reduction settings are, these minor artifacts may not even be detectable.

For SD video, you'll probably see a little benefit from knocking Constant Quality down to 19, making it virtually indistinguishable from the source in most cases. However, if you don't spend much time looking for minor differences, you may not even pick up on the improvements.

My big concerns, as far as artifacting goes, are interlace/telecine/comb lines and blockiness. On an encode I did on an earlier version of handbrake (pre :apple:tv) that the video was really blocky when there was fire or an explosion of some kind. Since I know the interlace/telecine/comb lines can be solved with the correct filter settings, I'm not too concerned with that. It's just the blocky artifacts that I'd rather not see.

Anyway, after playing with the :apple:tv2 preset, I've determined that it's just as good as I need it to be. I did go ahead and knock down the Constant Quality to 19, simply because the current setting produced files that were larger than I knew they should be (about 15mb a piece on some tv shows). When I made that change, the image looked just as good and I saved some disk space :) a small amount for now, but as I encode more movies, I'll appreciate how much space I gained.

Thanks again guys :)

dynaflash
Aug 19, 2011, 09:41 AM
I did go ahead and knock down the Constant Quality to 19, simply because the current setting produced files that were larger than I knew they should be (about 15mb a piece on some tv shows). When I made that change, the image looked just as good and I saved some disk space :) a small amount for now, but as I encode more movies, I'll appreciate how much space I gained.
Everything else being equal setting the rf for constant quality to 19 instead of the atv presets rf of 20 will make your files *larger* but higher quality (remember rf values are like golf, lower numbers are higher quality / higher output file size). So if you set it to 19 and noticed smaller files ... something else must have changed. Maybe it was a different source you tested on ?

obsidian1200
Aug 19, 2011, 09:47 AM
Everything else being equal setting the rf for constant quality to 19 instead of the atv presets rf of 20 will make your files *larger* but higher quality (remember rf values are like golf, lower numbers are higher quality / higher output file size). So if you set it to 19 and noticed smaller files ... something else must have changed. Maybe it was a different source you tested on ?

Nope, I only have three MKVs on my HDD right now, and sent all three of them through the gauntlet. One thing I did notice, but it didn't bother me since I lack a home theatre, was that the AC3 Passthru tracks didn't get encoded one round 2 with rf 19. Thinking back, I'd bet that that's what made the difference :( my mistake.

dynaflash
Aug 19, 2011, 10:19 AM
just remember that constant quality will produce higher/lower bitrates for the same visual quality depending on the content of the video, not just the duration and resolution. So grainy sources or complex sources will use a higher bitrate during those sequences and a still dark scene will use very little. This is why a comparison for file size requires the exact same source be used for both encodes. Just making sure you are getting the results you expect.