Blu-ray -> New AppleTV: Ripping HW/SW advice needed

[Scott R]

I've been messing around with the Acer Revo w/XBMC for a while now and I finally think I've got it working (thanks to the XBMC Dharma build under Windows 7) to the point where I can play a full 1080p Blu-ray rip (non-HD sound) with forced subtitles (Avatar was my test of that) using MakeMKV. But the Netflix PQ (and lack of seamless integration in XBMC) has been disappointing, the box (with Windows 7 license) costs over $300 (which becomes an issue for me since I would ideally like to have four rooms with their own device - so that can become $1200), and there's still some glitchiness to the usability/reliability of it all, so I'm thinking about going with the new AppleTV. I have no problem with having compressed, 720p, and no-HD sound since most of my sets are 720p and I don't currently have an HD audio receiver, so not having HD audio in my content actually saves me some money since I won't feel the need to have to upgrade. My "theater room" has a 720p projector so, again, not having 1080p content means I can be happy with what I've got.

Anyways, here's my dilemma...my current Blu-ray ripping machine is an old HP Slimline PC which I upgraded with a Blu-ray ROM drive. It works fine for ripping standard-def DVDs to ISO files and Blu-ray discs to MKV files, as all I do is get rid of the audio tracks, extras, etc. but don't compress anything further. Standard-def DVDs take about 20 mins to rip and Blu-rays about an hour. Any time I've tried to recompress/downconvert things, though (using DVDFab mostly), it seems to run for over 12 hours and I usually kill it. So I think I need to upgrade my ripping machine, and I don't mind doing that, but I'd still like to stay as cheap as possible (if possible), so I'm hoping for some advice as to what features are important and what isn't. I'm thinking of going with an HP refurb desktop since I've gotten two HP refurbs in the past (my current machine is one) and have had very good luck with the long-term reliability of them.

FWIW, I'm not currently planning on using this box for anything else, so I'd like to just get what I *need* and keep the price down as low as possible while getting the biggest bang for the buck. So, again, if a feature upgrade adds $50 but only saves me 20 minutes of ripping time, I'd probably prefer to save the $50. But if we're talking about ripping time going from 8 hours to 4 hours, that's certainly worth a $50-100 upgrade for me.

So on Tigerdirect or eCost, for example, I can find several different HP refurbs to choose from in the $350-500 range which all have Windows 7 and Quad Core CPUs. Please advise as to which of these features are important and how important they are...

- CPU: I'm thinking that getting a Quad Core is a must and that CPU speed is somewhat less important, but please correct me on that (e.g., is a 2.7 processor going to rip my movies a lot faster than a 2.6 processor?). What I have no clue on is whether an AMD chip is significantly better/worse than an Intel, and if there are certain types of chips that are significantly better than others (due to cache RAM, etc.). Or, again, is getting any Quad Core chip going to be A-OK and provide me with a fast ripping experience, with those other factors only providing marginal improvements?

- RAM: Based on other threads I've found when searching here, it sounds like having 8GB of RAM isn't going to matter vs 4GB. If that's not correct, let me know.

- Hard drive: 500GB should be plenty since I'm not planning on serving my media from the drive (after ripping, I'll copy it over to one of my external drives which will probably be "served" by my low-power Acer Revo). But what about RPM? It seems like most/all of these refurbs just have 5400RPM drives. Is that going to be a problem for ripping? I suspect not, since the bottleneck would seem to me to be the time it takes for the CPU/RAM to convert the video and not the time it takes to write the resulting compressed movie to disk.

- OS: I haven't looked into it, but I could probably save a few dollars if I found a refurb with Vista vs Windows 7. Will Handbrake (and any other ripping tools I'll need) run much better/faster on one vs the other?

Thanks in advance! More questions will surely follow.

[tlindgren]

- CPU: I'm thinking that getting a Quad Core is a must and that CPU speed is somewhat less important, but please correct me on that (e.g., is a 2.7 processor going to rip my movies a lot faster than a 2.6 processor?). What I have no clue on is whether an AMD chip is significantly better/worse than an Intel, and if there are certain types of chips that are significantly better than others (due to cache RAM, etc.). Or, again, is getting any Quad Core chip going to be A-OK and provide me with a fast ripping experience, with those other factors only providing marginal improvements?

If the goal is solely x264 (Handbrake) video encoding I think you want an Intel cpu, as many cores as you can get and if you can a Nehalem (i7 or even i5). A 2.7 GHz quad-core is 3.8% faster (in clock) than a 2.6GHz quad-core, so expect perhaps 3-3.5% faster encode. Basically, I would suggest finding a price point you're comfortable with and then find the biggest/widest Intel box you can find for that money.

I'm told that dual-core Intel i5's are actually faster than similar priced quad-core AMD, probably due to the very high "Turbo" frequencies the Intel dual-core i5's feature, if that's true Intel is basically always the best choice for a dedicated video encoder box.

if you use special (non-threaded) filters the performance on a single core may matter in which case an getting an Intel box will be even more important but under normal circumstances x264 will scale up to about ~5 (real) cores and the performance will be close to "number of cores*frequency" for a specific types of CPU. AMD's will be cheaper for multi-core machines but the Intel CPUs are usually faster enough at video coding that they're likely a SLIGHTLY better choice at any given price range. But AMD isn't a BAD choice, it's just that you're likely to be able to do somewhat better with Intel CPU for any Video encoding.

- RAM: Based on other threads I've found when searching here, it sounds like having 8GB of RAM isn't going to matter vs 4GB. If that's not correct, let me know.

4 GB should be plenty for just running Handbrake. 5400 RPM disks shouldn't be a problem either.

[Scott R]

Thanks for the reply! So here are a few examples...hopefully you can give me some ballpark estimates as to how big of an improvement I might see with various upgrades...

1) My present box. HP Slimline s3020n. Windows Vista. 2GB RAM. AMD Athlon 64 X2 Dual Core Processor 3800+ @ 2.00 GHz.

Side-question: This box has an on-board NVIDIA GeForce 6150 LE. I later bought a GeForce 8500 add-on card (exact specs: MSI NX8500GT-MTD256EH GeForce 8500 GT 256MB 128-bit GDDR2 PCI Express x16 HDCP Ready Silent Heatsink Video Card), but more recently removed that card since I was no longer using this desktop to actually watch anything, but only for ripping, and I needed the room in the case to add a gigabit ethernet card (which I figured would be more useful now that I was using the box as a media file server). But if Handbrake can make use of the add-on card to significantly improve conversion time, I'm wondering if it might be beneficial for me to put that card back in and save myself some money. So if Handbrake can do this, perhaps you can give me some ballpark guesses for this box with the built-in 6150 LE vs the add-on 8500 GT.

In case Handbrake can make use of video cards to improve conversion performance, I'll also include the video chipset info in my examples below...

2) HP refurb desktop w/AMD Athlon II X4 635 Quad-Core 2.9GHz. NVIDIA GeForce 6150SE. Windows 7 64-bit. $360.

3) Asus refurb desktop w/Intel Pentium E5300 2.6GHz (Dual core). Intel Graphics Media Accelerator X4500. Windows 7 64-bit. $430.

4) HP refurb desktop w/Intel Core 2 Quad Q8300 2.5GHz. Intel GMA X4500 HD graphics (I think). Windows 7 64-bit. $440.

5) HP refurb desktop w/Intel Core i5 750 2.66GHz. ATI Radeon HD4350. Windows 7 64-bit. $498.

6) HP refurb desktop w/AMD Phenom II X4-925 Quad-Core Processor 2.8GHz. ATI Radeon HD4350. Windows 7 64-bit. $520.

[Scott R]

Update: Well I did some more searching here and it sounds like Handbrake doesn't support GPU encoding (with the reason why being that apparently the current state of GPU encoding sucks), so if I've understood that correctly, no need to comment on the various video chipsets/cards in my last post.

I guess then, the main question is whether my current hardware should be good enough (perhaps with a 2GB RAM upgrade?), or if I'll see a significant performance improvement in encoding for the AppleTV with one of those other refurb desktops I've found online. And, which one has the most bang for the buck (e.g., if the $360 Athlon II X4 635 will make my encodes run in half the time compared to my current rig, and the $500 Intel Core i5 750 will only save me an extra 15 minutes of encoding time over that, then I'd probably just go with the Athlon II X4 635).

[randomreuben]

@ Scott R

Here's a thought for a set up. A netbook (approx. $300) for ripping purposes only. A copy of DVDFab (or whatever ripper you like) and an external optical drive will let you rip most things. A desktop running Ubuntu for encoding purposes with a high-core number/high RAM combination would let you encode movies very quickly. You can also run VLC on that Ubuntu desktop and use that to watch your movies.

Handbrake on Ubuntu is like a rock. If you look at that huge list of fixes on trac.handbrake.fr, most of them are for Windows. Even if you dual-boot with Ubuntu, you can run a great stable system for encoding.

Handbrake and VLC on Ubuntu (on a speedy computer). DVDFab on Windows (on some low end computer). A fantastic combination.

Take care,

randomreuben

[tlindgren]

1) My present box. HP Slimline s3020n. Windows Vista. 2GB RAM. AMD Athlon 64 X2 Dual Core Processor 3800+ @ 2.00 GHz.
2) HP refurb desktop w/AMD Athlon II X4 635 Quad-Core 2.9GHz. NVIDIA GeForce 6150SE. Windows 7 64-bit. $360.
3) Asus refurb desktop w/Intel Pentium E5300 2.6GHz (Dual core). Intel Graphics Media Accelerator X4500. Windows 7 64-bit. $430.
4) HP refurb desktop w/Intel Core 2 Quad Q8300 2.5GHz. Intel GMA X4500 HD graphics (I think). Windows 7 64-bit. $440.
5) HP refurb desktop w/Intel Core i5 750 2.66GHz. ATI Radeon HD4350. Windows 7 64-bit. $498.
6) HP refurb desktop w/AMD Phenom II X4-925 Quad-Core Processor 2.8GHz. ATI Radeon HD4350. Windows 7 64-bit. $520.

Anandtech uses x264 as one of their benchmarks, the video encoding core that also Handbrake uses. The list only has the CPU's they've run full reviews on it's not how I would run x264 but I suspect the fps value for the second pass gives a reasonable guess for Handbrake performance. http://www.anandtech.com/bench/CPU/28

Most of the CPU's mentioned isn't in that list but we can extrapolate from very similar CPUs that is listed and guess that your s3020 would get around 5 fps on their test. Machine 3 would be the next slowest at perhaps around 8.5, with machines 2, 4 and 6 is likely around 17 fps and machine 5 scoring 21.0.

All these are guesses and with a somewhat different setup to boot, but I think it's fairly likely around $360-$430 will more than triple your encode performance unless you're doing something special. $500 ought to more than quadruple it, but compared to machine 2 it's "only" 25% faster for $140 extra. It may or may not be worth it depending on how many DVDs you plan to encode, though the i5 machine should pull further ahead if you have any single-threaded filters. I did say I expected Intel and AMD to be fairly balanced even if I didn't expect it to be this close, though there were some outliers (machine 3, the older non-Nehalem dual-cores can't compete with quad cores, and machine 6 which appears to be slightly slower than the much cheaper machine 2 and 4).

All this assumes the machines have sufficient cooling, without that they can be much slower than they ought to be. Some laptops/nettops has this problem, it can happen with slimline machine like your machine too but it's somewhat less common.

[Scott R]

Excellent input. Thanks! As a matter of fact, I came across that Anandtech page earlier and, as you indicated, was thinking that #2 seemed to offered great bang-for-the-buck. But if I'm going to spend $360 anyway, spending an extra $140 might still be worth it to get the additional performance. As for cooling, all of the machines are full-size/larger cases, so hopefully that shouldn't be an issue.

#5 (the Core i5) is a bit of a mystery to me right now, as I've found this online from several sites, only one of them offered any level of specs about the box (RAM, HD size, etc.).

[Scott R]

Well, I've found a refurbed HP Core i7 920 online as well, priced at about $620, so I may add that to my list of possibilities. So, if I'm looking to convert a 2 hour Blu-ray to a 720p AppleTV (new model) capable format, and want the best possible PQ (ideally same level of quality as the iTunes purchased HD movies, if not better), it is my understanding that I would use the nightly build and either a modified iPad setting or High Quality setting. With an Intel Core i5 or i7, what type of encoding time should I expect to see (I'm just talking about Handbrake here - not counting the initial time to rip the movie using something like MakeMKV)?

FWIW, one of my rooms has a 720p LCD projector paired with a 92" diagonal 16:9 screen so, again, I really want to keep PQ as high as possible, and I'm not concerned about filesize, but I also don't want to waste processing time to eek out a few extra pixels worth of quality if I'm not likely to notice it and it adds significantly to the encoding time.

[Rodeo]

and want the best possible PQ (ideally same level of quality as the iTunes purchased HD movies, if not better)

iTunes has access to higher-quality sources than Blu-Ray, so that may not be possible for all sources.

With an Intel Core i5 or i7, what type of encoding time should I expect to see (I'm just talking about Handbrake here - not counting the initial time to rip the movie using something like MakeMKV)?

http://forum.handbrake.fr/viewtopic.php?f=9&t=17457

For an idea of 16:9 1080p performance, divide encoding speeds by 8: (1920x1088)/(720x368). For 16:9 720p, divide encoding speeds by 3.5/4.

[Scott R]

Thanks. So if I'm looking at your High Quality 102fps example, and divide that by 3.5 (for 720p), that would be about 30fps.

Is this single-pass encoding? I realize that I have much to learn still, but I seem to see a lot of recommendations for two pass encoding, and typically with the first pass set to turbo. Can you provide your input on that?

Also, I see that you had hyperthreading off, but I've also come across posts recommending to turn that on (I believe a Core i7 920 was being used by that person). Can you also comment on that?

Lastly, I believe you have a Mac...is there any significant difference in encoding times if I'm using Windows 7 or Linux with the same CPU?

Oh, one more thing...you commented about Apple having access to higher quality sources, but if I'm starting with an uncompressed 1080p Blu-ray rip, I would think that should be a pretty good source. What sort of video PQ issues might I see when downconverting to a (new) Apple TV capable 720p format (other than the obvious resolution loss)?

[jbrjake]

Oh, one more thing...you commented about Apple having access to higher quality sources, but if I'm starting with an uncompressed 1080p Blu-ray rip, I would think that should be a pretty good source.

....but still far worse of a source than Apple has access to. Blu-Ray is 420--color planes are half width and half height (one pixel of color information is applied to 4 pixels in the frame)--and uses average bitrates of around 40 megabits. Apple's sources are ProResHQ which uses 422--color planes are half width but full height (one pixel of color info is applied to 2 pixels in the frame)--and has an average bitrate of 220 megabits.

[Scott R]

OK, let me ask this a different way...if I use an iPad or High Quality profile and a straight 1080p Blu-ray rip as the source, other than the resolution difference, how close in picture quality can I expect the Handbrake-converted file to be to the original Blu-ray?

[mduell]

It will look about as good as the original BR would look if you could play it on the iPad.

OTOH, if you watched the HB encode on a HDTV, you'd probably be able to see the difference.

[alehel]

Oh, one more thing...you commented about Apple having access to higher quality sources, but if I'm starting with an uncompressed 1080p Blu-ray rip, I would think that should be a pretty good source.

....but still far worse of a source than Apple has access to. Blu-Ray is 420--color planes are half width and half height (one pixel of color information is applied to 4 pixels in the frame)--and uses average bitrates of around 40 megabits. Apple's sources are ProResHQ which uses 422--color planes are half width but full height (one pixel of color info is applied to 2 pixels in the frame)--and has an average bitrate of 220 megabits.

Makes Blu-ray sound rather puny .

[JootecFromMars]

If you want an idea of speed. I have a 2.8Ghz Core 2 Duo (E8250) iMac with 4GB and it takes about 2.5x the length of the movie to rip from a bluray 1080p file down to Apple TV preset at 720p.

[Scott R]

Thanks for the info.

On the subject of PQ, let me go at this from a different angle...is there anyone here who is currently converting Blu-ray discs to an Apple TV-capable 720p format and watching them on a big screen (ideally a front projector)? Regardless of how it compares to the original Blu-ray, would you say that it's still a significant improvement over standard-def DVD in all areas, or does the process of using Handbrake, combined with the end result of a fairly small filesize (3-4GB), result in new PQ issues that don't exist on the uncompressed standard-def DVD (e.g., tearing, stuttering, macroblocking, etc.)?

[mduell]

On the subject of PQ, let me go at this from a different angle...is there anyone here who is currently converting Blu-ray discs to an Apple TV-capable 720p format and watching them on a big screen (ideally a front projector)? Regardless of how it compares to the original Blu-ray, would you say that it's still a significant improvement over standard-def DVD in all areas, or does the process of using Handbrake, combined with the end result of a fairly small filesize (3-4GB), result in new PQ issues that don't exist on the uncompressed standard-def DVD (e.g., tearing, stuttering, macroblocking, etc.)?

With reasonable settings and intelligent filtering (as applicable) it's going to look way better than DVD (which are far from uncompressed!) and I see no reason to expect new artifacts.

The size of the screen isn't very significant because viewing distance generally increases with screen size. It's going to look the same on a 10' screen at 15' as it does on a 24" monitor at 3'.

[Scott R]

With reasonable settings and intelligent filtering (as applicable) it's going to look way better than DVD (which are far from uncompressed!) and I see no reason to expect new artifacts.

Thanks for your input!

The size of the screen isn't very significant because viewing distance generally increases with screen size. It's going to look the same on a 10' screen at 15' as it does on a 24" monitor at 3'.

That's true generally speaking, but I suspect it varies greatly from person to person and it also becomes less true when you're dealing with front projection. For example, I'm sure there are many younger geeks who buy a very big flat-screen and sit quite close to it, whereas I heard a couple with "the older mindset" talking recently at a store where they thought that a particular 27" LCD might be "too big" for a particular room. I don't know about you, but the only rooms I would ever consider to be too small for a 27" TV might be a kitchen or a small bathroom. And the type of person (like me) who is into front-projection typically believes the "bigger the better" and probably sits a factor (or more) closer than the average person with a flat-screen, because we want most of our field of view to be filled with the image.

At this point, some of you are probably thinking...oh, well if you're going to watch it on a big screen like that, you won't be happy with anything less than 1080p, but that's not true either. My first front projector (a used Sony CRT - quite the behemoth) back in the day when laserdisc was considered the best there was, and that was analog and about 400 lines of resolution. And even around then (and obviously before), there were people happily feeding their front projectors low-res VHS content (240 lines of res). I'll take a good movie at lower res and stereo over a bad movie in 1080p with DTS-HD sound any day.

[Scott R]

OK, so I'm testing out the latest Windows nightly build with my ancient AMD Athlon 64 X2 3800+ with what is likely an insufficient 2GB of RAM. Just want to get a gauge for how bad/good the conversion time will be and (hopefully) how good the end result PQ-wise can be. Since I only really care about the upcoming Apple TV and, perhaps, the iPhone 4 (which I also own, but likely will never watch video on), I'm following the advice from another thread which suggested using the "High Profile" setting adjusted to RF19 (which, if I got this right, is the "Constant Quality" slider under the "Video" tab). I also adjusted the Width to 1280 (under the "Picture" tab), but didn't change anything else. Note: my source is a Blu-ray rip in .ts format.

So, while I let this cook, some questions:
1) Did I get these settings right or did I forget to change something important?
2) I'm a bit confused about audio. Apple's spec page for the upcoming Apple TV states that it supports 5.1 Dolby Digital pass-through. For my rip, I've got the Audio tab left with the defaults, which shows two tracks. The 1st one shows Source: AC3 5.1 audio, Audio Codec: AAC (faac), and Mixdown: Dolby Pro Logic II. The 2nd one shows Source: <same as previous>, Audio Codec: AC3 Passthru, and Mixdown: Automatic. Are these settings correct? Should this allow me to play this file on an both an HTPC or Apple TV connected to a stereo LCD TV via HDMI (with, presumably, the HTPC/aTV either sending a mixed-down stereo signal to the TV or sending something more than stereo which the TV should be able to decode)?

[mduell]

1) 19 is a little low for HD, I'd say 21-22 minimum.
2) Yes, but without either connected to a AC3 or DPL2 receiver all you'll get is stereo.

[Scott R]

1) 19 is a little low for HD, I'd say 21-22 minimum.

Doh. I guess that was probably a waste of 8 hours. I've got my computer turned off for a bit and want to let it cool down before I try again, for fear it might catch fire. Ugh.

So when you say 21-22 *minimum*, can you (or others) give me a hard number that you'd suggest when converting a 1080p Blu-ray rip down to 1280 width, with optimal PQ as the goal (I have a 720p front projector, so I'd like it to look as close to the source as possible)?

Will boosting these numbers add significantly to the encoding time? As I mentioned, I believe that last night's encode took about 8 hours.

[TedJ]

It won't boost the encoding time, but it will increase filesize. My personal preference for a 720p downconversion from Blu-ray is RF 22-23.

[nightstrm]

I've used 20-21 before on my HD encodes (from Bluray/HDDVD) and had no issues streaming to my AppleTV. If output from the high profile preset streams reliably to the new AppleTV, I'll likely switch to using that preset with RF 21-22.

[Scott R]

Thanks for the added comments, everyone. I'm still a bit confused, though. You're all providing ranges, rather than a specific number. How are you deciding when to use 21 vs 22? If this whole concept is going to work for me, I'm going to want it to require no thought on my part. Right now, I can rip a Blu-ray with MakeMKV and just deselect the audio tracks, subtitles, etc. that I don't want. Beyond that, I don't need to think about it. If I'm going to utilize Handbrake to convert things into an AppleTV/iPhone4-friendly format, I'd like it if the next step was as easy as firing up Handbrake, selecting a preset (which can be a modified version of an existing preset that I've saved as my go-to preset), and clicking start. I can appreciate that I might want/need two presets (one for DVD sources, and one for Blu-ray sources), but beyond that I'd prefer not to have to try out different values, re-convert movies, etc. It's bad enough that the conversion itself is going to require several hours, but at least that's something that doesn't require me to sit by the computer while it's doing its thing.

FWIW, I tried out that CF19 Blu-ray file. It doesn't look bad at all. Surprising for a 2GB file. Initially I was worried because playing it on my Acer Revo nettop with XP via MPC-HC resulted in the audio being out of sync. Switching to the 2nd audio track seemed to fix that. I'm not sure which track was which, and the test room was with the Revo connected via HDMI to my stereo TV. The source .ts file did not have this problem (and, FWIW, MPC-HC only shows one audio track when playing that file). I'll have to try playing it in my theater room and confirm that I can get Dolby Digital working with the video and audio still in sync. Hopefully video/audio sync issues won't be a problem that requires manual futzing around. It's also possible that this will be a non-issue when playing this file via the new Apple TV.

As an added test, I synced the file over to my iPhone 4 and it played back beautifully there. Holding the phone up to my face, I could make out lots of detail and the audio was perfectly in sync. The added detail is a bit of a waste on the iPhone 4, IMO, due to the screen being so small. It wouldn't feel comfortable holding it up to my face, and at a more normal viewing distance, I imagine that I'm not seeing a good deal of the added resolution. That said, having a single file which can be easily copied over from my server to an iPhone without any added transcoding, is very nice.

So with this Blu-ray rip, what will moving up from RF 19 to RF 22 bring me? Is this going to result in more detail or in less motion artifacts/macroblocking/etc. FWIW, while I could see the resolution loss between the 1080p .ts source and this 2GB 720p file when standing a few inches from my 52" 1080p LCD TV, I didn't see any motion artifacts/macroblocking, which was a pleasant surprise. But if moving up to RF 22 can result in a little more detail, I'm sure I'd appreciate that when I watch these movies with my projector.

[dorito]

Moving up to 22 would degrade the quality, the lower the number the higher the quality!

Peronsally I find 17-16 gives the Best results. 22 looks awful in comparison .