Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Video Encoding > New and alternative video codecs

Reply
 
Thread Tools Search this Thread Display Modes
Old 1st August 2009, 19:06   #1  |  Link
JackSnap
Registered User
 
Join Date: Jan 2007
Posts: 22
Help with some Expression Encoder Options

Hi There,
I wonder if someone can help me with some options in MS Expression Encoder.

I've finally converted from Windows Media Encoder as it just cant cope with what I throw at it anymore.
Im pleased with the results of EE, but a couple of the options are puzzling me as to why i would use them.

Im converting to file from a 1080P source, and looking in the VC-1 encode settings.
The video complexity is set at 4, so the adaptive dead zone and differention quant are in use, but dont these two options slow down conversion and end up with a potentially lower quality output?

I believe if I set the video complexity to 5 it will ignore these two options, is that still the case?
if so wouldnt it be better to use level 5, and then i would get a better output quality and a faster encode?

Video encoding is not really my specialist subject, so sorry if its a dumb question. I will probably end up watching the WMV on my xbox 360 so any other tips on conversion for this platform would be good.
I aim to end up with a 8-10GB file depending on the length of the movie. (its got 5.1 surround too)

Cheers in advance for any help.
JackSnap is offline   Reply With Quote
Old 2nd August 2009, 07:56   #2  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 3,981
Quote:
Originally Posted by JackSnap View Post
Hi There,
I wonder if someone can help me with some options in MS Expression Encoder.
Sure. Which version are you running.

Quote:
I've finally converted from Windows Media Encoder as it just cant cope with what I throw at it anymore.
Im pleased with the results of EE, but a couple of the options are puzzling me as to why i would use them

Im converting to file from a 1080P source, and looking in the VC-1 encode settings.
The video complexity is set at 4, so the adaptive dead zone and differention quant are in use, but dont these two options slow down conversion and end up with a potentially lower quality output?
Complexity has a big perf impact, but DQuant and ADZ shouldn't have much of one. They're both perceptual optimizations, and so trade off reduced PSNR for what it hopes to be a psychovisual enhancement.

ADZ=Conservative is almost always helpful, as it keeps the video from getting blocky at moderate bits per pixel.

DQuant as implemented in EEv2/3 isn't normally that useful at 1080p below 15 Mbps ABR, as it tends to spend too many bits on smoother parts of image, making more complex areas blockier. For film/video content, if I use it at all it's normally just on I-frames, and then to prevent banding or slight blocking in smooth gentle gradients, like fog.

There are better VC-1 DQuant implementations which I really hope we get backported into the EE codebase next time around.

Quote:
I believe if I set the video complexity to 5 it will ignore these two options, is that still the case?
if so wouldnt it be better to use level 5, and then i would get a better output quality and a faster encode?
The specified chroma search and motion match methods were ignored in complexity 5 in WME/Format SDK, but that was fixed in the VC-1 Encoder SDK that EE uses. But ADZ and DQuant were always honored.

Complexity 5 just makes things a bunch slower for a very slight chance of improved quality. Complexity 5=Insane, Complexity 4=HQ, Complexity 3=default in my mind.

Improving Chroma search or Motion Match and particulary using Auto Motion Vector Search Range pay off much more in quality than Complexity 5>4.

Quote:
Video encoding is not really my specialist subject, so sorry if its a dumb question. I will probably end up watching the WMV on my xbox 360 so any other tips on conversion for this platform would be good.
I aim to end up with a 8-10GB file depending on the length of the movie. (its got 5.1 surround too)
2-pass VBR at complexity 3 looks better and encodes faster than 1-pass CBR at complexity 4.

Don't forget 5.1 audio can be 2-pass VBR as well.

The Denoise is pretty subtle and tuned to reduce compression-killing random noise without softening edges. If your source is noisier than you like and don't want to AVISynth, it'd

Really, you've got a lot of bits to work with there for 1080p24. A 2-hour movie in 8 GB would be 8.8 Mbps. So you can do 8 Mbps ABR and PBR at say 20 Mbps, with 440 Kbps ABR 768 PBR audio, and you should be looking pretty gorgeous without going crazy on encoding settings.

If you're using EEv3 (and you should be for HD encoding; it's quite a bit faster), The VC-1 HD 1080p VBR default is a pretty good starting point.

If it's only something you're going to watch yourself and so don't want the encoding time to be a big multiple of the content duration, I'd probably use something the "Personal" settings here.

But if it was something that was going to have a big audience and I didn't mind spending 3x longer on the encode, I'd use this professional version. Whether or not there's a quality difference probably depends on the content.
Attached Images
  
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 3rd August 2009, 17:28   #3  |  Link
JackSnap
Registered User
 
Join Date: Jan 2007
Posts: 22
Thanks for the reply,

Im using EE Version 3.

I actually dont mind too much about the encoding time, actually im normally more fussy than any audience. I dont mind waiting a week for an encode as long as its as good as its going to get.

As far as the options I had set, you would be glad to know they are almost exactly as you said in your second (better quality) settings, the only thing I had slightly difference is a keyframe at every 5 secs (not sure about the quality effect of this)
and a peak of 10000,
The reason for such a low peak is that i transfer the films wirelessly to my xbox, and if the bitrate gets much higher i start to get pauses in the video at high action moments (which is a definate no-no) Im not sure why as my wireless can handle better speeds than that, but through trial and error that seems to be the best rate it can handle with the 360 and films.

I dont tend to put the noise filters on as the sources are generally noise free.
I have deinterlacing set to force off, and the resize mode and resize quality shouldnt matter as i copy the source size.

I do try to use the dual pass audio too, I used to have a big problem with it im WME when the source was AC3 (something to do with the jitter on the second pass I think), so Ill see if EE is better at handling that.

I see it now handles AVIsynth without an issue, but I dont think I even need that now, as I only normally use it to chop the credits off the end, its looking like I can pull in an MKV directly and do this all from EE. and Im yet to test the multi audio channel selection, but I hear it can do that too.

One other question that im not sure will make a difference is that im going to try an encode of a HD 3D film where you use the coloured glasses to watch, these films tend to have fringes of colour around edges of objects, would any of the settings affect this, as i dont want to loose any 3d effect.

One last point about speed, with the options ive posted here, its taking about 3.5 days for a full encode of a 1080P 1:35h film.
would you say thats about right? Im on an AMD Athlon 64 x2 Dual Core 6000+ running at 3000MHZ

Again, thanks for your help and advice.
Attached Images
  
JackSnap is offline   Reply With Quote
Old 4th August 2009, 09:56   #4  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 3,981
Quote:
Originally Posted by JackSnap View Post
Im using EE Version 3.
Excellent!

Quote:
I actually dont mind too much about the encoding time, actually im normally more fussy than any audience. I dont mind waiting a week for an encode as long as its as good as its going to get.
Doubly excellent. Although let's not overkill; the challenge is to get as good as you can actually notice to encode as quickly as possible.

Quote:
As far as the options I had set, you would be glad to know they are almost exactly as you said in your second (better quality) settings, the only thing I had slightly difference is a keyframe at every 5 secs (not sure about the quality effect of this)
Probably not huge at these bitrates. It can be useful to get higher bitrates.

I tend to use 4 sec GOP and 4 sec buffer for HD. I recall I had a reason at one point, but alas don't recall the reason itself .

Quote:
and a peak of 10000,
The reason for such a low peak is that i transfer the films wirelessly to my xbox, and if the bitrate gets much higher i start to get pauses in the video at high action moments (which is a definate no-no) Im not sure why as my wireless can handle better speeds than that, but through trial and error that seems to be the best rate it can handle with the 360 and films.
Hmmm. Well, WiFi's not great for media streaming. Note you're probably getting a combination of buffer duration and peak bitrate that's your limit; using a smaller buffer at a higher peak could possibly work as well. Repositioning your antennae might be an easy fix.

I'll assume the 640x360 was just an artifact for the screen cap .

Given your average and peak are so close, I'm wondering if a CBR encode would make sense. Using Lookahead or Lookahead Rate control you can get quite good results. Basically LRC is our live mode from the VC-1 Encoder SDK that does lookahead based on fixed chunks. it's generally the highest quality option, but it's not something that's really been tested a lot for these bitrates and frame sizes. Lookahead is just a 16 frame lookhead for fade/flash/edit detection, to make sure we use intensity compensation, BI frames, and I-frames appropriately.

Lookahead Rate Control
  • 1-pass CBR
  • Closed GOP=On
  • Adaptive GOP=Off
  • GOP length under 5 seconds

Lookahead
  • 1-pass CBR or VBR (the frame mode decisions are always on for 2-pass encoding)
  • Adaptive GOP=On
  • Closed GOP=On (not required, but less chance of strobing)
  • GOP length under 5 seconds

Quote:
I dont tend to put the noise filters on as the sources are generally noise free.
Good. Is this animation content? I was thinking film, and hence film grain.

Quote:
I have deinterlacing set to force off, and the resize mode and resize quality shouldnt matter as i copy the source size.
As long as it shows up in the Media Content window as "24p" or another "p" frame rate, deinterlacing will be off automatically. But doing it manually doesn't hurt anything.

Quote:
I do try to use the dual pass audio too, I used to have a big problem with it im WME when the source was AC3 (something to do with the jitter on the second pass I think), so Ill see if EE is better at handling that.
The paid version includes its own AC-3 decoder with 2-pass support. But as long as whatever decoder you've got uses the same seed for the random number generator on each pass, you should be okay. WMA barfs on 2-pass encoding if the CRC of the passes don't match, which gets triggered by a random number dither in AC-3.

Quote:
I see it now handles AVIsynth without an issue, but I dont think I even need that now, as I only normally use it to chop the credits off the end, its looking like I can pull in an MKV directly and do this all from EE. and Im yet to test the multi audio channel selection, but I hear it can do that too.
If you can play it in WMP, it should decode in EEv3, due to its DirectShow support.

Plus the pipeline was engineered by one of the top 3 DirectShow experts on the planet .

Quote:
One other question that im not sure will make a difference is that im going to try an encode of a HD 3D film where you use the coloured glasses to watch, these films tend to have fringes of colour around edges of objects, would any of the settings affect this, as i dont want to loose any 3d effect.
Compressing anaglyph in 4:2:0 is kinda painful in general. Full True Chroma Search you'll absolutely need, since the chroma motion will often be all akimbo from the luma for this kind of encoding.

Other than that, keep your QPs low, I'd say. Whatever it takes to keep blocking down. Try the Denoise if you've got trouble. I've seen it drop my average QP by 1.5-2 with grain 1080p sources, which actually helped grain retention on net.

Quote:
One last point about speed, with the options ive posted here, its taking about 3.5 days for a full encode of a 1080P 1:35h film.
would you say thats about right? Im on an AMD Athlon 64 x2 Dual Core 6000+ running at 3000MHZ
Ooof. Dropping in a Phenom X4 would be a big improvement; 2-3x faster? Honestly, I think 8-core is the baseline for a decent compression system these days, since most codecs will scale at least that far.

I only recall one time in my 20 years of compression that I had an encode take more than three days and actually work .

Quote:
Again, thanks for your help and advice.
Good stuff. Keep us posted on your results.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 11th August 2009, 17:11   #5  |  Link
Jonaldinho
Registered User
 
Join Date: Aug 2008
Posts: 49
Hi Ben,
I'm encoding blu-ray to vc-1 to burn to dual layer discs. I change the resolution according to what bitrate I can use, 9000 kbit/s for 1080p, 4000kbit/s for 720p ie. 0.181 bits per pixel or above.I really want to get the best quality. I wondered if you could help me on a few settings.

I read a guide saying to use keyframe Interval @ 2 and peak buffer size @ 5. Will using 4 for each increase compressibilty & still play ok on an Xbox 360?

The guide said to leave Adaptive Dead Zone off and DQuant set to I Frames Only. I have done this for a few encodes and it looks good. No obvious compression artifacts although i can only view the encode on a 720p 32'' tv and i've only encoded very clean sourced films. I understand DQuant gives more bits to smoother parts meaning detailed parts loose bits, I know it's not recommended below 15 mbps for HD but what about animation which compresses well or a very clean source film? I've never used Adaptive Dead Zone, with it set to conservative does it degrade sharpness quality making it easier to compress therefore resulting in it being more clean and increase quality compared to it being off? Does it only do this where it really needs to? I really don't want to degrade quality but if it increases quality overall i'll use it. What settings would you suggest for Adaptive Dead Zone and DQuant for a) Grainy Film b) Moderately grainy film c) very clean film d) CGI animation e) 2D animation (probabably same settings as CGI but i'll just ask incase).

Thanks in advance if you can help me.

I'm using EE2, these are the settings i have been using with an example bitrate.

Jonaldinho is offline   Reply With Quote
Old 12th August 2009, 04:37   #6  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 3,981
Quote:
Originally Posted by Jonaldinho View Post
Hi Ben,
I'm encoding blu-ray to vc-1 to burn to dual layer discs. I change the resolution according to what bitrate I can use, 9000 kbit/s for 1080p, 4000kbit/s for 720p ie. 0.181 bits per pixel or above.I really want to get the best quality. I wondered if you could help me on a few settings.
Well, bits per pixel should generally be proportional to the power of 0.75. So if 4Mbps is good enough for 720p, you can probably get away with (1920x1080)/(1280x720)^.75*4000=7348 Kbps for 1080p.

Quote:
I read a guide saying to use keyframe Interval @ 2 and peak buffer size @ 5. Will using 4 for each increase compressibilty & still play ok on an Xbox 360?
Smaller buffers will increase compatibility, but can reduce compression efficiency a bit on the most complex scenes.

You can use a longer GOP if you're not worried about super-crisp random access. It can help efficiency, particularly if you're using I-Frame DQuant which can really increase bit utilization on I-frames.

Quote:
The guide said to leave Adaptive Dead Zone off and DQuant set to I Frames Only. I have done this for a few encodes and it looks good. No obvious compression artifacts although i can only view the encode on a 720p 32'' tv and i've only encoded very clean sourced films. I understand DQuant gives more bits to smoother parts meaning detailed parts loose bits, I know it's not recommended below 15 mbps for HD but what about animation which compresses well or a very clean source film?
What guide are you reading?

It can work for Animation at lower rates. Try it and tell me what you see . DQuant is most useful if the player you're using doesn't apply postprocessing, since it reducing blocking that a postprocessor would normally also soften.

I don't recall what resolutions the Xbox 360 does postprocessing at.

Quote:
I've never used Adaptive Dead Zone, with it set to conservative does it degrade sharpness quality making it easier to compress therefore resulting in it being more clean and increase quality compared to it being off? Does it only do this where it really needs to? I really don't want to degrade quality but if it increases quality overall i'll use it. What settings would you suggest for Adaptive Dead Zone and DQuant for a) Grainy Film b) Moderately grainy film c) very clean film d) CGI animation e) 2D animation (probabably same settings as CGI but i'll just ask incase).
I find Conservative ADZ almost always beneficial, or at least neutral. Turning it off or to Aggressive pretty much will always make it worse if there's a visible difference.

Basically it will trade off detail for reducing blocking supression, but doesn't do anything if there wouldn't otherwise be artifacts. It's like a weaker-but-safer version of DQuant in pratice, plus it applies to Simple and Main Profiles as well. They can be used together, of course.

Quote:
I'm using EE2, these are the settings i have been using with an example bitrate.
Get EEv3. It's quite a bit faster, and has some quality improvements as well.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 12th August 2009, 18:41   #7  |  Link
Jonaldinho
Registered User
 
Join Date: Aug 2008
Posts: 49
Ok thanks for your advice . I'd rather use 9000 kbit/s for 1080p and a higher bitrate for lower resolutions but i'm not quite sure how you worked out the bitrate for 1080p. (1920x1080)/(1280x720) = 2.25 .75x4000 = 7000 but i don't understand what ^ means and how it comes to 7348. I didn't know what resolution to use if the bitrate was too low so I got the idea from this site: http://www.binsearch.info/viewNFO.ph...776729&server= but if it's safe to increase resolution higher than what i have, when i have to use a lower bitrate then that's good .

So Adaptive Dead Zone set to conservative wouldn't degrade quality even with animation that compresses well, because it only uses it when there would be artifacts anyway?

The guide i read was this: http://thenexus.tk/encoding-xbox-360...wmv-hd-movies/ the first guide I read but now i want to understand and improve settings.

I'm encoding animation with DQuant set to I frames only right now with a bitrate of 12075 kbit/s. Adaptive Dead Zone is off though. I'll let you know the results.

Yeah i've read version 3 is faster. I'll get that soon. Thanks a lot for your time and help. It's much appreciated.
Jonaldinho is offline   Reply With Quote
Old 13th August 2009, 03:16   #8  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 3,981
Quote:
Originally Posted by Jonaldinho View Post
Ok thanks for your advice . I'd rather use 9000 kbit/s for 1080p and a higher bitrate for lower resolutions but i'm not quite sure how you worked out the bitrate for 1080p. (1920x1080)/(1280x720) = 2.25 .75x4000 = 7000 but i don't understand what ^ means and how it comes to 7348.
"^" means "to the power of"


Quote:
I didn't know what resolution to use if the bitrate was too low so I got the idea from this site: http://www.binsearch.info/viewNFO.ph...776729&server= but if it's safe to increase resolution higher than what i have, when i have to use a lower bitrate then that's good .
Play around and see what works for you. At those high resolutions, you might be getting more filtering from your display as from anything else.

Quote:
So Adaptive Dead Zone set to conservative wouldn't degrade quality even with animation that compresses well, because it only uses it when there would be artifacts anyway?
That's the theory, and I've not seen any counter examples. But like all perceptual optimizations, there could be pathological content that doesn't work well, but I'm comfortable making it an always on recommendation.

Quote:
The guide i read was this: http://thenexus.tk/encoding-xbox-360...wmv-hd-movies/ the first guide I read but now i want to understand and improve settings.
Yikes! I find it hard to believe that Complexity 0 wouldn't cause a quality hit.

Quote:
I'm encoding animation with DQuant set to I frames only right now with a bitrate of 12075 kbit/s. Adaptive Dead Zone is off though. I'll let you know the results.
Cool.

Quote:
Yeah i've read version 3 is faster. I'll get that soon. Thanks a lot for your time and help. It's much appreciated.
And it's a free download. Just install the "trial" - it's full featured for what you're doing. There's a bunch of nice fixes and tweaks; no reason to bother with v2 for this.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 15th August 2009, 17:15   #9  |  Link
CruNcher
Registered User
 
CruNcher's Avatar
 
Join Date: Apr 2002
Location: Germany
Posts: 4,949
Benwaggoner is there any paper how Microsofts Adaptive Deadzones HVS algorithm works internally and on what it adapts itself (Quantization, Motion, Both) ?
i saw a lot of interesting possibilities with x264 in terms of perception in different situations (smooth out Quantization for sharper (without risk amplifying blocking), or blurrier (save bits) results in different motion situations) and sources (trellis is already some kind of it though i find it destructive visually, and i had better results speed wise and visually doing something like trellis outside of x264 using mvtools in combination with the GPU). Also using it as a artificial Encoder sharpening is interesting especially with Sharp Mpeg-2 Sources i get very interesting visual results using it (trying to smooth out Quantization Blurrienes effect on lower bitrates for interesting ROI scenes) that way (though it is not easy to hold it stable in motion frenzy out edges especially when AQ and Deadzones come together)

Btw x264 can by now with very low effort compress down your "What happens in Vegas" DNLA EE sample by Half in 1.5x Realtime (1pass) on 2 cores (K8) and keeping the grain structure (and most of the VC-1 compression artifacts intact) in most of the scenes visually (without using PSY-RD or any RD @ all) (especially the hardcore 2 woman stand in front of the door background grain) (The most visual problem still are the fades but soon hopefully this will be also history)
Im about to compare it with The Island Trailer that you did back then with the Rhozet Carbon Encoder (Mpeg-2, blocky high motion scenes)

Maybe you could create a new Sample with EEV3 using higher compression efficiency and maybe unrestricted to anything @ 2 mbits 720p
__________________
all my compares are riddles so please try to decipher them yourselves :)

It is about Time

Join the Revolution NOW before it is to Late !

http://forum.doom9.org/showthread.php?t=168004

Last edited by CruNcher; 15th August 2009 at 18:08.
CruNcher is offline   Reply With Quote
Old 16th August 2009, 21:51   #10  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 3,981
Quote:
Originally Posted by CruNcher View Post
Benwaggoner is there any paper how Microsofts Adaptive Deadzones HVS algorithm works internally and on what it adapts itself (Quantization, Motion, Both) ?
Not that I know of, but I don't really track the academic/patent side of these things. Generally any kind of psychovisual optimization publication from Microsoft will have Tom Holcomb as a co-author.

Quote:
i saw a lot of interesting possibilities with x264 in terms of perception in different situations (smooth out Quantization for sharper (without risk amplifying blocking), or blurrier (save bits) results in different motion situations) and sources (trellis is already some kind of it though i find it destructive visually, and i had better results speed wise and visually doing something like trellis outside of x264 using mvtools in combination with the GPU). Also using it as a artificial Encoder sharpening is interesting especially with Sharp Mpeg-2 Sources i get very interesting visual results using it (trying to smooth out Quantization Blurrienes effect on lower bitrates for interesting ROI scenes) that way (though it is not easy to hold it stable in motion frenzy out edges especially when AQ and Deadzones come together)
Yeah, blocking/dithering and the non perceptual uniformity of gamma are an interesting and probably under-researched area of codecs. We've certainy done a fair amount around dark luma DQuant which has proved to be a really big deal when targeting consumer LCD displays. And when a display shows adjoining blocks of Y'=16 and Y'=17 as visibly different, in-loop deblocking or overlap transform aren't going to help, since the output of the decoder is still 8-bit 4:2:0.

Much of the value of PEP and VC-1 for HD optical came from the outboard dithering we had to address this issue, but I think that an encoder that integrated the 10-bit to 8-bit conversion could do even better psychovisual tuning. In the end, dithering's a pretty lossy process, essentally randomizaing the LSB. IF the codec was able to account for both the source frequencies (what was really a band and what was a gradient or noisy region in the source), it'd be able to figure out much less costly ways of preventing blocking and banding.

And even that dithering's pretty rare. The only reasonably mainstream professional compression tool that includes good dithering is Inlet's Fathom.

While 10-bit isn't an option for the recompression workflow, 10-bit 4:2:2 sources are very common inprofessional content. And if we could really take advantage of 10-bit sources for content delivered as compressed, a lot more could be authored accordingly. All the major NLEs can do 10-bit 4:2:2 or beyond, and we've got nice visually lossless authoring codecs like ProRes, DNxHD, and Cineform that make it much less expensive than it was a few years ago when v210 and expensive RAIDs were the only real option.

Quote:
Btw x264 can by now with very low effort compress down your "What happens in Vegas" DNLA EE sample by Half in 1.5x Realtime (1pass) on 2 cores (K8) and keeping the grain structure (and most of the VC-1 compression artifacts intact) in most of the scenes visually (without using PSY-RD or any RD @ all) (especially the hardcore 2 woman stand in front of the door background grain) (The most visual problem still are the fades but soon hopefully this will be also history)
Im about to compare it with The Island Trailer that you did back then with the Rhozet Carbon Encoder (Mpeg-2, blocky high motion scenes)
I have permission to provide the Island source, if you like. Unfortuantely it came from an 8-bit DPX sequence.

Quote:
Maybe you could create a new Sample with EEV3 using higher compression efficiency and maybe unrestricted to anything @ 2 mbits 720p
Fine idea! I've got a big backlog of blog tutorals and samples from when I was working on the book, but it's all checked in now, thank goodness. The new dynamic motion vector cost makes a big improvement on film grain encoding, pretty much eliminating false-positive motion vectors due to noise. This both looks a lot better, and saves bits that can be used elsewhere.

I've got a bunch of new trailers using the Smooth Streaming VC-1 VBR encoder that uses variable bitrate, chunk/GOP duration, and frame size. Hopefully they'll get posted this week.

I never use unrestricted for any real-world content, though, since there's always some upper bound of decode complexity and peak bitrate that should be applied. In practice, with a high enough buffer constrained and unconstrained give the same output, so there's no downside to having some constraint. My default (with many exceptions, particularly as ABR goes down) is a 1.5x ABR/PBR ratio. That provides probably 80% of the theoretical max quality advantage of VBR, without having complexity/VBV much harder than CBR.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book

Last edited by benwaggoner; 17th August 2009 at 08:15.
benwaggoner is offline   Reply With Quote
Old 19th August 2009, 10:57   #11  |  Link
CruNcher
Registered User
 
CruNcher's Avatar
 
Join Date: Apr 2002
Location: Germany
Posts: 4,949
Thx to you and DreamWorks for Sharing the 8 Bit source of The Island Trailer
__________________
all my compares are riddles so please try to decipher them yourselves :)

It is about Time

Join the Revolution NOW before it is to Late !

http://forum.doom9.org/showthread.php?t=168004
CruNcher is offline   Reply With Quote
Old 17th November 2009, 23:30   #12  |  Link
Jonaldinho
Registered User
 
Join Date: Aug 2008
Posts: 49
Hi Ben, i've been using your settings with fantastic results. I'm finally getting a faster processor this week so i can encode all my blu-rays to be played back on my Xbox 360. I'm wondering wether to encode to 720p or 1080p. For 1080p I would encode them to fit on DVD DL,for 720p DVD-R, DVD DL for longer movies. Would 6mbps be too low for 1080p? Would i be better off going down to 720p? What about custom resolutions like 1600x900. One setting i don't fully understand is the overlap filter, do i just use it for low bitrate? How low? Hope you can help, thanks.
Jonaldinho is offline   Reply With Quote
Old 19th November 2009, 15:01   #13  |  Link
Jonaldinho
Registered User
 
Join Date: Aug 2008
Posts: 49
I've done a few tests with The Dark Knight Blu-Ray source at a bitrate of 5687 Kbps, the lowest the bitrate would be on a DVD DL for a 3hr movie. I encoded to 1920x1080, 1440x1080 (Anamorphic) & 1280x720. All with the Overlap filter off. I noticed in the imax scenes the picture looks excellent, then the normal scenes look a bit blocky and blurry sometimes, even at 720p. I'm gina try with overlap filter on now, never used it. It made me think though if the quality is so good in the imax scenes (& some 2.35:1 scenes) and bad in other scenes is it because i've got the peak bitrate to high (14500 Kbps). Does this comment you wrote too CruNcher mean choose a peak bitrate 1.5x the average bitrate:

''My default (with many exceptions, particularly as ABR goes down) is a 1.5x ABR/PBR ratio''

So in this case 5687*1.5=8531

I've compared screenshots and the sharpness is still always much more noticable with the 1080p encodes vs the others, even with a low bitrate.

Sorry for so many questions, I want to choose settings I no will work well before i encode everything. Thanks.
Jonaldinho is offline   Reply With Quote
Old 21st November 2009, 01:54   #14  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 3,981
Quote:
Originally Posted by Jonaldinho View Post
Hi Ben, i've been using your settings with fantastic results. I'm finally getting a faster processor this week so i can encode all my blu-rays to be played back on my Xbox 360. I'm wondering wether to encode to 720p or 1080p. For 1080p I would encode them to fit on DVD DL,for 720p DVD-R, DVD DL for longer movies. Would 6mbps be too low for 1080p? Would i be better off going down to 720p? What about custom resolutions like 1600x900. One setting i don't fully understand is the overlap filter, do i just use it for low bitrate? How low? Hope you can help, thanks.
The overlap filter is an in-loop deblocking filter for intra blocks, weighted to QP. It's turned off automatically at low QPs (below 4 maybe?), but at higher QPs will reducing blocking on I-frames and intra-coded blocks of other frames. It's almost always a net positive at "interesting" bitrates, although wasn't generally wasn't used for HD DVD as it can cause a slight increase in softness.

It's a much simpler design than the VC-1 in-loop deblocking filter used for inter blocks, or the H.264 in-loop deblocking filter, which is why it's allowed in Simple Profile.

I'd definitely have it on for 6 Mbps 1080p encodes from 1080p source. It can make for a pretty dramatic improvement in the quality of your most complex scenes, particularly if your peak bitrate is relatively low.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 21st November 2009, 02:01   #15  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 3,981
Quote:
Originally Posted by Jonaldinho View Post
I've done a few tests with The Dark Knight Blu-Ray source at a bitrate of 5687 Kbps, the lowest the bitrate would be on a DVD DL for a 3hr movie. I encoded to 1920x1080, 1440x1080 (Anamorphic) & 1280x720. All with the Overlap filter off. I noticed in the imax scenes the picture looks excellent, then the normal scenes look a bit blocky and blurry sometimes, even at 720p. I'm gina try with overlap filter on now, never used it. It made me think though if the quality is so good in the imax scenes (& some 2.35:1 scenes) and bad in other scenes is it because i've got the peak bitrate to high (14500 Kbps). Does this comment you wrote too CruNcher mean choose a peak bitrate 1.5x the average bitrate:

''My default (with many exceptions, particularly as ABR goes down) is a 1.5x ABR/PBR ratio''

So in this case 5687*1.5=8531
Well, you might go in with WMSnoop and see where your bitrate is going. But in general, for Xbox use I'd use a higher PBR/ABR ratio since you have such a strong decoder. I'd probably use 12 Mbps minimum for 1080p there, maybe even 15. The 1.5x ratio is to make transfer time per minute more predictible for progressive download and to go easier on lower-end processors, neither of which are the case here.

Still, it's an interesting case since the aspect ratio varies, and I suppose the IMAX scenes could be starving the rest of the movie. But those aren't so many, and they're really the money shots, so if they're gorgeous, it sounds like things are heading in the right direction .

There's rarely a quality hit from using a high peak bitrate; that's just a max, and the encoder might never get that high if optimal rate-distortion doesn't mandate it. I've done 100 ABR 1000 PBR encodes where the peak didn't get much past 300.

Quote:
I've compared screenshots and the sharpness is still always much more noticable with the 1080p encodes vs the others, even with a low bitrate.
Cool. And Overlap should definitely help at those bitrates. For 1080p, I don't think I'd even think of turning it off below an ABR of at least 10 Mbps.

Quote:
Sorry for so many questions, I want to choose settings I no will work well before i encode everything. Thanks.
No apology needed; these are fun questions!
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 22nd November 2009, 00:37   #16  |  Link
Jonaldinho
Registered User
 
Join Date: Aug 2008
Posts: 49
Cool. Thanks a lot for all your answers. I really appreciate it. I will take all your answers on board. I only encoded the first 10 mins of The Dark Knight which more than half were taken up by the imax scenes so when I encode the whole movie i'm confident it will look great at not a really high, but higher bitrate of 6841 kbps. There is definitely an improvement with the overlap filter aswel. I'm confident using your settings so once my new processor arrives, i can crack on.
Jonaldinho is offline   Reply With Quote
Old 13th August 2010, 11:10   #17  |  Link
Jonaldinho
Registered User
 
Join Date: Aug 2008
Posts: 49
I wondered if anyone knows why I can't open an avisynth script with Expression Encoder 3? I can't do it however basic it is eg Directshowsource("movie.m2ts"). I've tried with avisource, import etc. I have all the correct decoders installed, I can open a script and convert to x264, but Expression Encoder just crashes and says it's stopped working then closes. I need it to input yellow and blue anaglyph video and would rather not encode it 1st. Thanks.

EDIT: In the compatiblity settings the only decoders that are checked are ac3filter, ffdshow video, haali media splitter (AR) and directvobsub (auto-loading version).

Last edited by Jonaldinho; 13th August 2010 at 11:13.
Jonaldinho is offline   Reply With Quote
Old 17th August 2010, 00:19   #18  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 3,981
There were some stability issues with EEv3 and AVISynth chains that take a long time to return a frame. Do you have SP1 installed? Also, EEv4 (free download for WMV encoding features) does much better with AVS sources.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 17th August 2010, 10:20   #19  |  Link
Jonaldinho
Registered User
 
Join Date: Aug 2008
Posts: 49
Do you mean Service Pack 1? If so then yes. I've tried with version 4 and it still crashes. To get around the problem i've been encoding to x264 near lossless first, qp 1 (qp 0 won't decode properly) but would prefer to just pass it straight to expression encoder.

While on the subject of Version 4 please could you tell me what the new setting Motion Vector Cost is about and what would produce the best quality. Also, if I enter a value in the min or max packet size will it ignore the set bitrate and just produce a file size to the selected value. Thanks.
Jonaldinho is offline   Reply With Quote
Reply

Tags
360, expression encoder, vc-1, wmv, xbox

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 22:27.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, vBulletin Solutions Inc.