1080p Handbrake settings for new Apple TV?

[shortcut3d]

Wirelessly posted

My question is would the Apple TV 2 preset 720p encode be better PQ than a High Profile preset 1080p with the CQ slider down to 22. The source is Blu-Ray. I have movies like Avatar that are 11GB at 1080p. I'd like to cut down the size a little or just go back to the Apple TV 2 preset.

I did a comparison test between the High Profile preset at CQ RF:20 and CQ RF:22 using chapters 2 - 6 of Avatar Blu-Ray. I ran both m4v files synchronous and over lapped the images. I was not able to distinguish the two images on my 2010 27" iMac (I'm impressed I was able to watch two H.264 1080p movies simultaneously). I focused on dark areas, motion, and sharpness. I could not find any areas in the CQ RF:22 sample that were worse than the CQ RF:20 sample. There was some minimal banding in both samples (probably why those who want Blu-Ray PQ go to RF:18), but very acceptable. The picture quality is also far better than premium cable HD movies. Based on the CQ RF:22 sample size the full file size will 73.5% of the CQ RF:20, which is very close the 720p Apple TV 2 Preset for the same movie.

I'm re-encoding my large file size 1080p offenders done over the weekend with CQ RF:22. Handbrake is right when they recommend CQ RF:20-22 for HD sources (Blu-Ray).

 

[mic j]

I bet I have the exact same model (PX-120U) and it works very well.

----------


I've converted the first 2 Lord of the Rings movies (each just shy of 3 hours) and they are 8.24 and 9.45GB respectively. I cant remember what I chose fo the first one, but the second was done as High Profile and took 22 hours to convert on my MBP (with little else running).

Don't the LOTR series use DTS? How did you deal with that. I finally got done transcoding Avatar (22hrs) only to find out that the aTV does not pass the dts audio through.

 

[mic j]

I did a comparison test between the High Profile preset at CQ RF:20 and CQ RF:22 using chapters 2 - 6 of Avatar Blu-Ray. I ran both m4v files synchronous and over lapped the images. I was not able to distinguish the two images on my 2010 27" iMac (I'm impressed I was able to watch two H.264 1080p movies simultaneously). I focused on dark areas, motion, and sharpness. I could not find any areas in the CQ RF:22 sample that were worse than the CQ RF:20 sample. There was some minimal banding in both samples (probably why those who want Blu-Ray PQ go to RF:18), but very acceptable. The picture quality is also far better than premium cable HD movies. Based on the CQ RF:22 sample size the full file size will 73.5% of the CQ RF:20, which is very close the 720p Apple TV 2 Preset for the same movie.

I'm re-encoding my large file size 1080p offenders done over the weekend with CQ RF:22. Handbrake is right when they recommend CQ RF:20-22 for HD sources (Blu-Ray).

Thank you for that analysis. And my question to you as well, is how are you dealing with the DTS audio track?

 

[dynaflash]

Thank you for that analysis. And my question to you as well, is how are you dealing with the DTS audio track?

If you haven't changed anything on audio settings, hb should encode the first output track to aac and the second to AC3 .... unless the source track is AC3 then it gets passed through.

 

[mic j]

If you haven't changed anything on audio settings, hb should encode the first output track to aac and the second to AC3 .... unless the source track is AC3 then it gets passed through.

God, I don't know why I didn't catch that. I just opened Avatar in HB and noticed the DTS-MA track sets to AC3-Discreet 6.1. I think I just got so focused on doing pass-through I wasn't noticing all my options (or the default). That cost me...big time.

So now the question is...I am able to demux the dts track out of the original mkv using Subler. The exported file is an MKA. Using a Mac, what is the best way to convert the dts to AC-3 6.1? Then I assume I would use Subler to remix, correct?

 

[dynaflash]

God, I don't know why I didn't catch that. I just opened Avatar in HB and noticed the DTS-MA track sets to AC3-Discreet 6.1. I think I just got so focused on doing pass-through I wasn't noticing all my options (or the default). That cost me...big time.

yep AC3 Discrete is the AC3 encoder.

So now the question is...I am able to demux the dts track out of the original mkv using Subler. The exported file is an MKA. Using a Mac, what is the best way to convert the dts to AC-3 6.1? Then I assume I would use Subler to remix, correct?

TBH I don't know of anything easy ... ffmeg should be able to do it. ( its the ffmpeg ac3 encoder we use to go from dts to ac3). Not sure whats easiest and fastest. maybe others can chime in.

Worse comes to worse ... re encode it overnight with the correct audio settings

 

[mic j]

yep AC3 Discrete is the AC3 encoder.



TBH I don't know of anything easy ... ffmeg should be able to do it. ( its the ffmpeg ac3 encoder we use to go from dts to ac3). Not sure whats easiest and fastest. maybe others can chime in.

Worse comes to worse ... re encode it overnight with the correct audio settings

Thanks for your help. I was just looking at the ffmeg route and it was looking to be the most practical. I'll probably give it a try and if it's not quick and straightforward, I'll just have to bite the bullet.

 

[osohardy]

God, I don't know why I didn't catch that. I just opened Avatar in HB and noticed the DTS-MA track sets to AC3-Discreet 6.1. I think I just got so focused on doing pass-through I wasn't noticing all my options (or the default). That cost me...big time.

So now the question is...I am able to demux the dts track out of the original mkv using Subler. The exported file is an MKA. Using a Mac, what is the best way to convert the dts to AC-3 6.1? Then I assume I would use Subler to remix, correct?

iFlicks will do this. It's a metadata tagger but will also remux h264 or transcode if you choose to make it "itunes compatible." For audio, as necessary it will create an AAC and secondary AC3 640kb track from DTS in your remuxed track. It is paid though. I think you could probably also do another handbrake run with really low settings for speed, transcode the audio to AC3, then demux/remux just the new audio with the original video.

 

[rayward]

Could you explain your work flow for the forced subs? I've been ignoring them because most of the workflows I've seen include like 3 extra programs needed.

I posted a link to my Mac workflow further up this thread.

 

[dynaflash]

I think you could probably also do another handbrake run with really low settings for speed, transcode the audio to AC3, then demux/remux just the new audio with the original video.

yes if you go the hb route just to get the ac3 encoded track try the Legacy > Classic preset but add the AC3 encoded track to the Audio Panel. the video (which your not going to use anyway) will be done with ffmpeg and is *much* faster than x264. Shouldn't take too long at all. Then as said demux the audio track out and mux it back into your chosen file.

 

[mic j]

yes if you go the hb route just to get the ac3 encoded track try the Legacy > Classic preset but add the AC3 encoded track to the Audio Panel. the video (which your not going to use anyway) will be done with ffmpeg and is *much* faster than x264. Shouldn't take too long at all. Then as said demux the audio track out and mux it back into your chosen file.

Thought the HB way made sense. Was in the process when I seen your post about using the Legacy preset. I see it has constant bit rate at 1000kbps encode as the default. Should I stay with that or do a 1-pass with RF set to low as quality as possible?

 

[shortcut3d]

Don't the LOTR series use DTS? How did you deal with that. I finally got done transcoding Avatar (22hrs) only to find out that the aTV does not pass the dts audio through.

As others have said, the default Audio settings for High Profile preset change the DTS track to AC3 6.1 (6-channel).

I'm encoding 5 movies all over again using CQ RF:22. It going to take 18-20 hours to finish all 5. Someone posted earlier that we over bit our encodes and I am leaning towards his point of view. I checked Salt and Rango this morning; and both movies show no visible loss in quality, but significantly smaller file size (30-40% reduction). Note: This is for Blu-Ray encodes. I may attempt a few WTV TV Shows.

 

[mic j]

As others have said, the default Audio settings for High Profile preset change the DTS track to AC3 6.1 (6-channel).

I'm encoding 5 movies all over again using CQ RF:22. It going to take 18-20 hours to finish all 5. Someone posted earlier that we over bit our encodes and I am leaning towards his point of view. I checked Salt and Rango this morning; and both movies show no visible loss in quality, but significantly smaller file size (30-40% reduction). Note: This is for Blu-Ray encodes. I may attempt a few WTV TV Shows.

Thanks for your info. In the meantime I did discover my error with the DTS/HB settings. It took me 22 hours to transcode the 2h45m Avatar.

So the thought of transcoding the 6 LOTR extended version BD discs (all 2-3hr long) is not the most pleasant thought. But, I generally operate from the point of view that you cannot add lost data back in...and it's only processing time after all. So, I will probably bite the bullet and use the 20RF and just let the MBP crank for a week. If it was something I cared less about I would probably compromise a bit, but I want those to be as good as I can get them.

 

[Los]

Why don't you encode the audio like someone suggested and then use Subler to mux the video and audio together?

 

[mic j]

Why don't you encode the audio like someone suggested and then use Subler to mux the video and audio together?

Doing it now. Have a HQ video mp4. Producing a HQ audio mp4 as I type. Will use Subtler to combine the best of both files.

 

[staccato83]

I really wish Apple would support mkv files. This encoding crap takes way too long and subtitles are a pain.

 

[mic j]

I really wish Apple would support mkv files. This encoding crap takes way too long and subtitles are a pain.

Yep, if they were to only add support for one additional format, I would vote for mkv. However, I am not at all sure a high bit rate mkv would wifi stream well. That's been a problem area for lots of other units. Compressing the file/bitrate has made streaming over wifi a lot smoother.

But now that I think of it, being able to attach a USB HDD would be a good problem solver.

Of course, they would rather you not transcode and buy/rent their movies. So they really have no motivation to make watching local content easier.

 

[kagharaht]

Bunraku

On screens from 60" on up, try this movie. All the large solid colors will reveal encoding limitations.

 

[khollister]

I posted a link to my Mac workflow further up this thread.

YOu and I are doing basically the same thing except you are using MKVtools v2 instead of MKVtoolnix. The problem I have had with MKVtools is it sometimes chokes on MKV's that every other tool I use accepts without complaint. MKVmerge & MKVextract have never choked for me.

BDsup2sub choked on the Avatar subs, but I did locate a forced sub SRT track that I dropped in with HB.

 

[dynaflash]

Thought the HB way made sense. Was in the process when I seen your post about using the Legacy preset. I see it has constant bit rate at 1000kbps encode as the default. Should I stay with that or do a 1-pass with RF set to low as quality as possible?

the ffmpeg encoder is *much* faster than the x264 encoder (with much worse quality) so for this purpose ( encoding just to get the AC3 track) it would be faster ... that said ... if you already on your way .... then let it go.

Otherwise. Legacy > Classic is very fast. We really only keep it for testing as the quality is relative crap .. but if you need a fast encode. its the way to go.

 

[shortcut3d]

Thanks for your info. In the meantime I did discover my error with the DTS/HB settings. It took me 22 hours to transcode the 2h45m Avatar.

So the thought of transcoding the 6 LOTR extended version BD discs (all 2-3hr long) is not the most pleasant thought. But, I generally operate from the point of view that you cannot add lost data back in...and it's only processing time after all. So, I will probably bite the bullet and use the 20RF and just let the MBP crank for a week. If it was something I cared less about I would probably compromise a bit, but I want those to be as good as I can get them.

I suggest encoding a couple chapters in Handbrake at RF20 and RF22 for the same movie. Then open one in iTunes and the other in Quicktime to do a side by side comparison. The floating iTunes window and Quicktime window will not have a title so you could even have someone mix them up to prevent bias. My point is there is no additional data loss between RF20 and RF22 when encoding a Blu-Ray at 1080p. RF20 is just over bitted for the movies I have tested so far. Maybe something like the upcoming Disney title Brave will have original PQ that makes RF20 a better choice. Conversely, some movies like Sucker Punch that are intentionally grainy result in increase file size because of the grainy picture.

 

[Mike Valmike]

Just for accuracy, the High Profile or Normal preset was recommended for the aTV 3 by Handbrake on the Handbrake forum. That should be a pretty good source for a recommendation.

All my settings are used in the High Profile preset. Sorry I did not specify that.

Also you can go a little lower than 2800-3000 and still get virtually the same quality... I made an encode just recently at 2500k 2-pass without turbo first pass to experiment and it came out absolutely brilliant. (TRON Legacy). It's probably 99% of the quality of the 38GB blu-ray it came from and it's 90% smaller file size. On consumer-grade equipment, I'll take that. If you want better, you'd watch the bd anyway.

----------

It's all subjective of course but I think you are in the minority here with a 1.9GB 1080p file being acceptable to watch, especially on 55" TV unless you like banding, blocking, and softness. What is your "magic" option sting?

The "magic" comes from the encoder being able to do higher quality with fewer bits when you use 2-pass. I've been encoding video since 2002 using various tools available from year to year and consistently, multipass produces better results. The software creates a "map" of the entire video and is able to allocate higher bitrate better for scenes that really need it, rather than parsing through the file front to back on an RF quality setting and adjusting bitrate on the fly, as it were.

I know the HB guy and his team don't recommend 2-pass. I can't imagine why not, as it produces better results in fewer bits. It just does, and always has. It always will because of the way it works and the way video is encoded. Though over time I concede that RF will get better and better and storage will get cheaper and we all won't really care as much down the line.

And again, you start with high profile, as the rest of the settings in high profile are really good at default and work well, especially the entire way that the HB team has the audio set up. You just change the average bitrate based on some experiential knowledge and some trial and error, and turn on 2-pass and if it's not critical reference material you can turbo the first pass.

 

[crazee928]

is it just me, or do i not see much of a difference between a ~1280px width and a ~1800px width movie resolution (i forget the height)? i had ratatouille encoded for the ipad2 native resolution, then got one at 1080p encoded via handbrake high profile and i could barely tell the difference between the two, if at all, yet the file size doubled. please correct me if i'm wrong, but as of now, i don't think any visible difference is worth cutting my new ipad's storage in half if i were to update all my movies.

 

[chiefpavvy]

I know the HB guy and his team don't recommend 2-pass. I can't imagine why not, as it produces better results in fewer bits. It just does, and always has. It always will because of the way it works and the way video is encoded. Though over time I concede that RF will get better and better and storage will get cheaper and we all won't really care as much down the line.

Law of diminishing returns. Yes 2-pass produces a slightly better encode but at a massive encoding time penalty. CQ produces superb quality encodes in half the time. The general frame of mind here is that it is better to trust the encoder to make bitrate decisions as opposed to setting some finite rate which may not be appropriate for all titles, and I agree completely with that. (For those who set bitrates.) I used to do 2-pass encodes @ 2500Kbps for all my DVD material. I now do single pass CQ, the bitrate is often half that, and the encodes look just as good - if not better - than the ones I did years past. x264 is constantly evolving and if you use the latest Handbrake nightly builds you will get those improvements as well.

The great thing about it is you can tinker with the settings to your heart's desire. I gave up tinkering a long time ago. I pretty much use the default profiles and let her rip. Yet to be disappointed, even with fairly low bitrate encodes.

 

[dynaflash]

I know the HB guy and his team don't recommend 2-pass. I can't imagine why not, as it produces better results in fewer bits. It just does, and always has.

Lol. you might want to check with the x264 devs on that one They write the x264 encoder hb uses ... with all due respect ... they have forgotten more than you or I will ever know about video encoding.

None of them use average bitrate 2 pass anymore (unless for some strange reason they have to hit a specific bitrate/file size). Period. As one of the (if not main) developers of x264 said ... "do you really think you can arbitrarily guess the correct bitrate to choose better than the encoder .. ?" At any rate of course in hb as in the x264 exe you can set your own average bitrate and two pass ... which of course is fine. Thing is ... since about four years ago when the crf method reached the maturity it now has ... no on really serious uses it anymore if they want to maintain a given quality. HandBrake maintains a very tight relationship to the x264 project given our video encoding lives and dies with x264. ABR 2 pass vs. CQ is not something we dreamed up. Its from much advice and testing from the x264 devs.

That said ... just my .02

Just sayin' .