|
|||||||||
|
Thread Tools | Search this Thread |
December 23rd, 2009, 11:53 AM | #1 |
Inner Circle
Join Date: Jun 2005
Location: Cincinnati, OH
Posts: 8,425
|
At what point do you notice lost quality when lowering bitrate?
I have a project that is 135 minutes, but I want to add 10 minutes of footage, and don't want a noticeable loss is quality, but I don't want to divide things up over two discs. Usually that is what I do, but in this case there isn't enough video to justify two discs. Too much for one, not enough for two.
I sell my service largely on high quality images so this is important to me, any noticeable loss in quality is not acceptable, as much of the reception footage is shot in the dark, and therefore must be preserved as best as possible. How much can you lower the default bit rate when rendering widescreen SD footage and not see any difference? |
December 23rd, 2009, 11:55 AM | #2 |
Inner Circle
Join Date: Jan 2006
Location: Minnesota (USA)
Posts: 2,171
|
That REALLY depends on the footage. As a practical matter, chances are you probably will not notice a difference between 135 minutes and 145 minutes (of the same footage), encoded properly for a DVD, so long as the 135 minutes looks real good.
|
December 23rd, 2009, 12:09 PM | #3 |
Major Player
Join Date: Sep 2008
Location: East Bay Cali
Posts: 563
|
its all linear, the less bits you have the worse it looks, but that is so Highly dependant on the quantity of movement and how many of the "pixels" get changed up on the whole frame, when using a proper 2 pass VBR.
Dvds are originally sort of 2 Hour items, if you wanted to pick a standard that would be it. if you have 2 hours of clean still stuff , piece of cake. if you have 2 hours of every pixel changing half the time it certannly isnt enough data already, but it is still within the 2 hours. mabey your math should be in percentages? your only adding in 10 more minutes. Vary rates for stuff that doesnt need it? non zooming slideshows dont need hardly any data, the zoom slideshows changes almost every pixel and needs more data than. Compression of dark stuff goes to heck , because it some of the first things tossed. High color stuff needs more data, low color not so much slow motion doesnt need as much data, but by slowing down the data you see every flaw the highlights parts are the coolest, always give them more data. use 2 disks, WTH if your distribution is less than say 8 copies , then it doesnt make any difference . does the customer get daunted by thinking the thing is LONG? not if you tell them you put it on two in a High Data Rate so it would look wonderfull. built in intermission, good thing. When gramma gets a copy and its 2 discs OH MY she will leave it on the counter longer :-) but if you have two different things on each disk, then they will watch one now and one later.
__________________
----------------sig----------------- Re-learning everything all over again, one more time. |
December 23rd, 2009, 01:35 PM | #4 |
Inner Circle
Join Date: Jan 2006
Location: Minnesota (USA)
Posts: 2,171
|
Compressed image quality does not have a linear relationship to bitrate. As bitrate increases returns in image quality diminish. When you start getting over 6Mbps or so for (most) video (properly) encoded for DVD, there's very little gain in image quality.
It really depends on the footage, as to how low the bitrate can go, and still get a good image. Going over 2 hours does start getting dicey on a single layer DVD, but with some footage, if there's little motion and little noise, you can certainly exceed 2 hours, and still get nice imge quality. There's a couple things you can do, if you are right on the hairy edge of acceptable image quality, with a particular video. You can apply some visually lossless noise reduction, to help the codec's efficiency. Lightly denoising can result in an image that visually looks no different from the original, but to the compressor, those little stray pixels that are not noticeable to the human eye can require allocation of a lot more bits for encoding. A good way to denoise for boosting compression efficiency, is to use MSU's denoising filter with Virtual Dub (use google), at the low setting. The image will look the same to the human eye, but much cleaner to the codec. The other thing you can do, is use dual layer DVDs. You can use almost the maximum allowable bitrate (for DVD video), with 145 minutes of video on a dual layer DVD. |
December 23rd, 2009, 01:40 PM | #5 |
Inner Circle
Join Date: Jan 2006
Location: Minnesota (USA)
Posts: 2,171
|
Btw, Virtual Dub and MSU's denoising filter are both freeware. I don't know of any NLE that bundles a better denoising filter with their product (most suck), although there are certainly some good plugins out there. Even so, I don't really know of a plugin that is clearly better than MSU's denoising filter, and some of them are awfully pricey. MSU's filter is really pretty dang good (and the price is right).
|
December 23rd, 2009, 02:04 PM | #6 |
Inner Circle
Join Date: Jun 2005
Location: Cincinnati, OH
Posts: 8,425
|
I wont' thank you individually, thanks everyone, great information. Initially I will lower the bitrate as little as possible and see what happens...
|
December 23rd, 2009, 02:37 PM | #7 | |
Inner Circle
Join Date: Sep 2003
Location: Portland, Oregon
Posts: 3,420
|
Quote:
Video noise is a real killer for compression. Even invisible noise greatly affects compression. Dual layer DVD - so many worries go away! I'll only add that as a working pro, only you can decide how good is good enough. Prep some sample footage. Run some short tests with that at target bitrates... you'll know!
__________________
30 years of pro media production. Vegas user since 1.0. Webcaster since 1997. Freelancer since 2000. College instructor since 2001. |
|
December 23rd, 2009, 04:09 PM | #8 |
Inner Circle
Join Date: Jun 2005
Location: Cincinnati, OH
Posts: 8,425
|
Wait, I just re-read the posts, and saw Marty mention VBR 2 pass.
I never have used 2 passes. I read (in this forum) years ago a discussion as to when it is beneficial, but completely forget what I read. I vaguely recall that it seemed to not be of much benefit under normal circumstances. Anyone care to comment? Also, it occured to me I do not know the proper way, if there is one, to lower bit rate. What I generally do is change the default settings to 9,000,000, 5,900,000 and leave the last setting as it is. I changed nothing else. If that is not enough, I lower it a tad more as needed. Incidentally, this setting produced exactly the right size for the DVD that I needed, though I haven't viewed it yet. |
December 23rd, 2009, 04:39 PM | #9 |
Major Player
Join Date: Jan 2007
Location: New York City
Posts: 523
|
Variable bitrate - Wikipedia, the free encyclopedia
Basically, on the first pass, the encoder 'watches' the video and determines which parts can be heavily compressed and which must be lightly compressed. On the second pass, it uses this information to do the compression. VBR is, well, more intelligent. Depending on the content, this can reduce the file sizes and increase the compression dramatically. Large amounts of movement will not compress well under VBR. Whenever possible, I try to use VBR.
__________________
Andy Tejral Railroad Videographer |
December 23rd, 2009, 04:39 PM | #10 |
Major Player
Join Date: Sep 2008
Location: East Bay Cali
Posts: 563
|
imo 2 pass is a nessiary for anything important. But mostly for about everything we do.
it would work "best" for stuff that has great needs for Bits in one place, and in another will not need bits. Because i use dissolves for example, It is a requirement because every pixel changes at that one span of time. so VBR steals bits from some place that doesnt need it and uses it there. it would be almost useless for talking heads stuff, locked shots, anything where the pixels dont all change up. it is really useless if your at 9 using the max rate the whole dvd. If your project is short you wouldnt need it.
__________________
----------------sig----------------- Re-learning everything all over again, one more time. |
December 23rd, 2009, 04:45 PM | #11 |
Inner Circle
Join Date: Jan 2006
Location: Minnesota (USA)
Posts: 2,171
|
You definitely want to do 2 pass VBR encoding! If you encoded the 135 minute version at a constant bitrate, the 145 minute version (2-pass VBR encoded) will be visually better quality. Depending on how good the 135 minute version looks in the first place though, it may not be noticeable (if the 135 minute version already looks great anyway).
I haven't touched Vegas in a coon's age, but basically, you want to set the average bitrate to whatever is appropriate (for maximum utilization of the DVD capacity). I'd set the upper bitrate limit to 8000Kbps though, for maximum assurance of playability on DVD players. Even though the specs allow for a little higher bitrate, pushing the upper limits of DVD video specs can potentially cause compatibility difficulties with a few players (crappy ones), in real world usage, and visual quality really just won't be impacted (with any reasonably typical footage). |
December 23rd, 2009, 04:55 PM | #12 |
Inner Circle
Join Date: Jan 2006
Location: Minnesota (USA)
Posts: 2,171
|
Where VBR makes the most difference, is when portions of the footage have a lot of motion and other portions have little motion (and/or more noise/less noise in portions). The codec can allocate more bits to the high motion portions, and waste less on the low motion portions that just don't really need it. Whenever you are encoding at less than maximum CBR bitrate, using VBR will improve the result visually. For some footage, the difference can be dramatic, while with some footage it can be minimal (depending on both the footage itself, and how hard you are compressing it in the first place).
|
December 23rd, 2009, 05:17 PM | #13 |
Inner Circle
Join Date: Jan 2006
Location: Minnesota (USA)
Posts: 2,171
|
Marty is right about transitions. Transitions are essentially the same as 100% motion, to MPEG-2 video compression.
|
December 23rd, 2009, 05:41 PM | #14 |
Inner Circle
Join Date: Jun 2005
Location: Cincinnati, OH
Posts: 8,425
|
Hmmm, a bit to mull over. I used huge number of slow dissolves in this project. Not usually my style, but it really fit the project. I've got tonite to try with and without and see how they work.
|
December 23rd, 2009, 05:49 PM | #15 |
Inner Circle
Join Date: Jan 2006
Location: Minnesota (USA)
Posts: 2,171
|
If the 135 minute version looks real good, don't sweat it. The 145 minute version will almost assuredly look just fine too (with 2-pass VBR encoding).
|
| ||||||
|
|