Perceived wisdom has it that you get the greatest reduction in MPEG-2 file size when using variable bitrate encoding rather than constant bitrate for footage that features lots of movement.
In fact, although it may seem counter-intuitive, the truth is almost exactly the opposite of that!
If you don't believe me, you can try it for yourself at home.
I took a sample 60 second long avi file that contained lots of movement, including some panning of the camera. In order to reduce the constant audio component of the rendered file size, I set audio to MPEG-2 256 kbps. (although I'm in NTSC land, I'm working with the VS9 trial which won't let me use Dobbly)
I then rendered to MPEG-2 at 4000kbps, 6000kbps and 8000kbps with both constant and variable bitrates.
The variable bitrate files were respectively 92%, 93% and 88% the size of the corresponding constant bitrate files.
Another 60 second AVI clip which I subjectively perceived as containing less movement gave rise to figures of 85%, 83% and 78% for the same tests.
Now taking this to it's logical limit, a video becomes just a slideshow. Repeating the exercise for an image given a duration of 60 seconds resulted in figures of 59%, 56% and 43%.
Using another image file resulted in 72%, 56% and 43%.
Then I got to thinking that the ultimate condition was not an image but a simple colour clip. Irrespective of the chosen colour, the same test resulted in values of 11%, 8% and 6%.
So at a maximum variable bitrate of 8000 kbps, the 60 second colour clip rendered to an MPEG-2 file was almost 16 times smaller than an 8000 kbps constant bitrate file! In fact if you rendered a colour clip at constant bitrate, 8000 kbps, you could only fit about 80 minutes worth of it onto a DVD5 disc. (But it might be more watchable than some of my videos!)
Two things are apparent.
1) The filesize reduction benefit of using variable bitrate is greater with increasing bitrates.
2) The benefit is greater as the video contains less movement and more uniformity.
The Real Advantage of Variable Bitrate
Moderator: Ken Berry
The Real Advantage of Variable Bitrate
JVC GR-DV3000u Panasonic FZ8 VS 7SE Basic - X2
-
THoff
Re: The Real Advantage of Variable Bitrate
I've never heard anyone with any qualifications make that statement -- it doesn't make sense.2Dogs wrote:Perceived wisdom has it that you get the greatest reduction in MPEG-2 file size when using variable bitrate encoding rather than constant bitrate for footage that features lots of movement.
VBR encoding is always relative to an average bitrate, so whether you use 6000Kbps CBR or VBR doesn't make any real difference -- one second of video will take 6000Kbps, and that's that.
That being said, VBR encoding will probably achieve better quality than CBR encoding because it can use otherwise unused bandwidth around and use it for more complex frames.
Re: The Real Advantage of Variable Bitrate
So I've already covered the render speed and simplicity aspect of choosing VBR over CBR, but here you appear to be suggesting that VBR is suited to highly variable video, which otherwise contains large amounts of simpler scenes. Pretty much what I was saying in my post, I would have thought.OK, perhaps I didn't really choose the right words. I should have said something like "it seems to be a commonly held belief". I think that might fairly represent a lot of peoples' views on VBR. I've read in more than a few posts that "using VBR might enable you to put another 10 minutes on a DVD" or words largely to that effect.THoff wrote:I've never heard anyone with any qualifications make that statement -- it doesn't make sense.2Dogs wrote:Perceived wisdom has it that you get the greatest reduction in MPEG-2 file size......
Now I would guess that most posters on this forum are hobbyists, who do it for fun. I have a degree in structural engineering - and that's a qualification, isn't it?
How about this next statement for some prime USDA approved nonsense:I think you may need to explain yourself there.VBR encoding is always relative to an average bitrate, so whether you use 6000Kbps CBR or VBR doesn't make any real difference -- one second of video will take 6000Kbps, and that's that.
If the goal is to achieve the highest possible video quality output fitted to a disk, then I would argue that VBR encoding would likely give superior results to CBR, particularly if there is little movement and much uniformity in the source material. (though I admit that I have yet to analyse the actual video quality of my test renderings....)
This last part sounds like you regretted the earlier attack:I like the word "probably"!That being said, VBR encoding will probably achieve better quality than CBR encoding because it can use otherwise unused bandwidth around and use it for more complex frames
In a post entitled "Pixels On DVD" you posted to New Guy: Tue Jul 05, 2005 2:11 am
Now anyone with any qualifications might know from practical tests that strangely enough, Video Studio takes almost exactly the same time to render a file from AVI to MPEG-2, whether CBR or VBR, and virtually regardless of the chosen bitrate. OK, sure, two-pass encoding will take almost twice as long, but your inference was that VBR, period, was slower than CBR.2. Increase the bitrate. A bitrate of 4000Kbps is going to give you VHS quality at best, and in your case you have specified that as the maximum bitrate, and what really determines quality for VBR encoding is the average bitrate. CBR encoding is quicker and simpler -- use VBR only once you run into disk capacity problems, and then use two-pass encoding.
As for the relative complexity or simplicity of the processes - I sure don't know exactly what series of 0's and 1's are happening within my pc while it's doing it - but I don't need to. Even without qualifications, I find it no more difficult to select "Variable" than I do to select "Constant"!
Now here's an interesting one. Post subject: VBR -vs- CBR Posted: Tue May 17, 2005 11:07 am.
You posted:It depends on the situation.
CBR is easier and faster to encode, and causes fewer problems when fast forwarding or reversing. If you aren't running out of disk space or have to lower your CBR bitrate below about 6000Mbps to make it fit, then stick with CBR.
VBR is useful to fit longer videos onto a disk, or to encode highly variable video where one scene is really simple (static backgrounds, little movement) and the next is highly complex (explosions, special effects, fast panning). In that case VBR encoding can be used to use a specific average bitrate, and the encoder is allowed to shift data around to compensate for the more or less demanding video frames. VBR doesn't really become effective unless you use multi-pass encoding, and unless the minimum and maximum bitrates aren't close to the average bitrate.
The discussion of the "average bitrate" would be all well and good if it was possible to specify it. With Video Studio, we are not able to. (I'd love to stand corrected on that one- and I know that you can edit the ini files to bring up the advanced properties, but they do not "stick") So all we can do is look at results from practical tests. It strikes me that you must be a firm believer in the old adage, "it's fine in practice but it doesn't work in theory".
P.S. where can I get one of those 20-pass gizmos?
JVC GR-DV3000u Panasonic FZ8 VS 7SE Basic - X2
-
skier-hughes
- Microsoft MVP
- Posts: 2659
- Joined: Thu Jul 21, 2005 10:09 am
- System_Drive: C
- 32bit or 64bit: 64 Bit
- motherboard: gigabyte
- processor: Intel core 2 6420 2.13GHz
- ram: 4GB
- Video Card: NVidia GForce 8500GT
- sound_card: onboard
- Hard_Drive_Capacity: 36GB 2TB
- Location: UK
Interesting reading, but as I don't use VS, I don't know how much you can change the VBR settings.
For my VBR, using canopus procoder, I can set the minimum, average and maximum bitrate. Now, just changing the max would have various results if no change was made to the other two, although min is nearly always zero anyway. If you changed the average bitrate, what did you change it to?
How about this, from another forum, which states CBR isn't actually constant anyway
CBR - Constant Bitrate
Goal: Constant bitrate through encoding
The bitrate does not vary up or down - it remains constant.
This produces constant size but variable quality.
1-pass VBR - Single pass Variable Bitrate
Goal: Constant quality through encoding
The bitrate varies from the base video bitrate up to the defined max video bitrate.
When the complexity of the video increases, the bitrate will be increased up to the max video bitrate.
This produces constant quality but variable size (assuming the max bitrate is sufficient to reproduce the complex video sections).
2-pass VBR - Two pass Variable Bitrate
Goal: Set average bitrate with variable instantaneous bitrate
The bitrate varies up to the defined max video bitrate, with the overall average maintaining the defined base video bitrate.
When the complexity of the video increases, the bitrate will be increased up to the max video bitrate but bitrate in areas of lower complexity will be decreased in order to "free" bits for the complex sections.
This produces constant average quality and constant size (assuming the max bitrate is sufficient to reproduce the complex video sections and the defined average bitrate is not so low that bitrate peaks create poor quality in lower-bitrate sections).
Very important points:
1) CBR is fastest, 2-pass VBR is slowest (assuming same Speed/Quality settings)
2) 2-pass VBR is not necessarilly better than 1-pass VBR - especially if predictable size is not a concern.
An easier way to think about it:
CBR is like a straight cylinder. Its size is constant throughout.
1-pass VBR is like a balloon - it can increase as necessary but won't get bigger than the max you set.
2-pass VBR is like a bean bag - some parts can have more beans, other parts can have little or no beans, but overall - it's always a set number of beans.
The only comment I would make is that these are general guidelines. Many users come away after reading this that CBR is inferior to VBR. Certainly CBR set at 5000 to fit 2 hours of video on a disc would be inferior to VBR setting of 5000 average and 8000 max. There will be scenes where 5000 is more than enough and CBR will look just as good as the VBR footage at the same segment, but there are going to be scenes where more bitrate is needed, and the 5000 setting will fall short and look inferior to the VBR file at that same point.
Putting it another way, if one has only an hour of video that they are going to put on a disc, then they aren't concerned about file size. They can set the CBR at 8000, and it should be pretty darn close visually to what any VBR setting with a max of 8000 would give, if not better. The VBR encoded file shouldn't really have any segments encoded higher than the CBR file, and may potentially drop to a lower bitrate than what you would like just by the way it analyzed the file. I have had some graphics quicktime files created in India Pro that look better with CBR than they do with VBR because the bitrate dropped fairly low during the encode when it shouldn't. Since you can't put a minimum on how low it can go in Procoder, that can be a problem with some encodes. Impression Pro from Pinnacle struggles with playback of very low bitrates, and for that reason files that I encode from graphics programs (primarily as intros or motion menus for dvd's) I use CBR.
Time and place for each setting made.
section VBR vs CBR: All modern video codecs are variable (in bitrate) in the sense that not every frame uses the same number of bits as every other frame.---- Even codecs labeled "CBR" can vary data rate quite a bit throughout the file.
section Buffered v whole file data rate control: For example, for a five-second buffer, the data rate for each five seconds must be the same. This isn't a question of multiple, discreet five-second blocks, but that any arbitrary five seconds plucked from the file must be at or under the target data rate. This is also called sliding window and CBR.
In order not to infringe the copyright of the author, I shall not copy the whole page here. Just to summarize, the cbr mode is not an absolute constant for every frame. It is a constant over a buffered period along the timeline of the file. The buffer period slides along the timeline of the file and is equivalent to the concept of moving average. But at any one single point, it is not a constant.
For my VBR, using canopus procoder, I can set the minimum, average and maximum bitrate. Now, just changing the max would have various results if no change was made to the other two, although min is nearly always zero anyway. If you changed the average bitrate, what did you change it to?
How about this, from another forum, which states CBR isn't actually constant anyway
CBR - Constant Bitrate
Goal: Constant bitrate through encoding
The bitrate does not vary up or down - it remains constant.
This produces constant size but variable quality.
1-pass VBR - Single pass Variable Bitrate
Goal: Constant quality through encoding
The bitrate varies from the base video bitrate up to the defined max video bitrate.
When the complexity of the video increases, the bitrate will be increased up to the max video bitrate.
This produces constant quality but variable size (assuming the max bitrate is sufficient to reproduce the complex video sections).
2-pass VBR - Two pass Variable Bitrate
Goal: Set average bitrate with variable instantaneous bitrate
The bitrate varies up to the defined max video bitrate, with the overall average maintaining the defined base video bitrate.
When the complexity of the video increases, the bitrate will be increased up to the max video bitrate but bitrate in areas of lower complexity will be decreased in order to "free" bits for the complex sections.
This produces constant average quality and constant size (assuming the max bitrate is sufficient to reproduce the complex video sections and the defined average bitrate is not so low that bitrate peaks create poor quality in lower-bitrate sections).
Very important points:
1) CBR is fastest, 2-pass VBR is slowest (assuming same Speed/Quality settings)
2) 2-pass VBR is not necessarilly better than 1-pass VBR - especially if predictable size is not a concern.
An easier way to think about it:
CBR is like a straight cylinder. Its size is constant throughout.
1-pass VBR is like a balloon - it can increase as necessary but won't get bigger than the max you set.
2-pass VBR is like a bean bag - some parts can have more beans, other parts can have little or no beans, but overall - it's always a set number of beans.
The only comment I would make is that these are general guidelines. Many users come away after reading this that CBR is inferior to VBR. Certainly CBR set at 5000 to fit 2 hours of video on a disc would be inferior to VBR setting of 5000 average and 8000 max. There will be scenes where 5000 is more than enough and CBR will look just as good as the VBR footage at the same segment, but there are going to be scenes where more bitrate is needed, and the 5000 setting will fall short and look inferior to the VBR file at that same point.
Putting it another way, if one has only an hour of video that they are going to put on a disc, then they aren't concerned about file size. They can set the CBR at 8000, and it should be pretty darn close visually to what any VBR setting with a max of 8000 would give, if not better. The VBR encoded file shouldn't really have any segments encoded higher than the CBR file, and may potentially drop to a lower bitrate than what you would like just by the way it analyzed the file. I have had some graphics quicktime files created in India Pro that look better with CBR than they do with VBR because the bitrate dropped fairly low during the encode when it shouldn't. Since you can't put a minimum on how low it can go in Procoder, that can be a problem with some encodes. Impression Pro from Pinnacle struggles with playback of very low bitrates, and for that reason files that I encode from graphics programs (primarily as intros or motion menus for dvd's) I use CBR.
Time and place for each setting made.
section VBR vs CBR: All modern video codecs are variable (in bitrate) in the sense that not every frame uses the same number of bits as every other frame.---- Even codecs labeled "CBR" can vary data rate quite a bit throughout the file.
section Buffered v whole file data rate control: For example, for a five-second buffer, the data rate for each five seconds must be the same. This isn't a question of multiple, discreet five-second blocks, but that any arbitrary five seconds plucked from the file must be at or under the target data rate. This is also called sliding window and CBR.
In order not to infringe the copyright of the author, I shall not copy the whole page here. Just to summarize, the cbr mode is not an absolute constant for every frame. It is a constant over a buffered period along the timeline of the file. The buffer period slides along the timeline of the file and is equivalent to the concept of moving average. But at any one single point, it is not a constant.
-
jchunter
2Dogs,
Well, I believe you because you have done the measurements. Anyone who doubts this has an obligation to do his own measurements and report them here. This forum could do with less speculation and more science.
The wonder is that slide shows take as much space as they do because the compression algorithm should know ahead of time that an unmoving image has a duration of x seconds and is going to is going to need an easily calculable number of Identical GOPs (Group of Pictures). The first GOP is easy to compute because, with unmoving subject matter, the compressed inter-GOP frames are simple to calculate.
It has always bugged me that slide shows can't use the extra bitrate headroom to display more resolution on an ordinary DVD.
Well, I believe you because you have done the measurements. Anyone who doubts this has an obligation to do his own measurements and report them here. This forum could do with less speculation and more science.
The wonder is that slide shows take as much space as they do because the compression algorithm should know ahead of time that an unmoving image has a duration of x seconds and is going to is going to need an easily calculable number of Identical GOPs (Group of Pictures). The first GOP is easy to compute because, with unmoving subject matter, the compressed inter-GOP frames are simple to calculate.
It has always bugged me that slide shows can't use the extra bitrate headroom to display more resolution on an ordinary DVD.
Hi Graham,
you raise many interesting points, but I must state that my tests were only concerned with using VS9 (trial, in fact) so many of the VBR options and settings are simply not available.
The text formating options of this forum are also fairly limited, but I could have done a better job of presenting my results. Maybe I should have created a few charts since a picture is often worth a thousand words and so on.
I do realise that there is variability within CBR - the fact that CBR filesizes varied with video content would be proof of that. The way I presented my figures probably masked it, however.
What truly surprised me was how much more compression you get with VBR when using just an image or colour clip. I can't imagine any rational argument for having to generate large file sizes for a colour clip - which CBR does.
As you say, CBR at 8000 kbps will fit something over an hour of video onto a DVD5, and maybe that's what most people would be doing. Most MiniDV tapes are an hour long, so it sort of ties in with that.
Using compressed audio allows you to push the maximum bitrate up to the maximum DVD-VR compliant 9800 kbps. With CBR, you might only get 60 minutes of typical video onto the disk - less when you add menus. If you use VBR, however, you might just squeeze an hour onto disk.
Most of the time, the difference might be slight - although in subjective tests I've done previously, I do think that the video quality is improved when you use VBR resulting in a file of comparable size to a CBR one.
Anyone doing a slideshow type video might see greater benefits, however. Perhaps you could fit 90 minutes of slideshow material onto a DVD5 using VBR 9800 kbps. (although the idea of sittting thru 90 minutes of holiday snaps might not be that appealing!)
Long credit rolls are certainly much more compressed with VBR, freeing up disc space for the use of a higher maximum bitrate. When I did a previous post on rolling credits, needless to say I test rendered to all kinds of settings and the VBR files looked good.
At the end of it all, I don't lose sight of the fact that the final video quality has much more to do with the quality of the original source material, and I am no great videographer. It's just an interesting little numerical exercise (for me at least) with some practical benefits, in some specific situations.
I have to admit that my initial response to THoff's post was quite snappy and had a more aggressive tone than I intended - but I was a bit miffed at his dismissive tone. I half expected to see the thread removed.
Thanks for your constructive response!
you raise many interesting points, but I must state that my tests were only concerned with using VS9 (trial, in fact) so many of the VBR options and settings are simply not available.
The text formating options of this forum are also fairly limited, but I could have done a better job of presenting my results. Maybe I should have created a few charts since a picture is often worth a thousand words and so on.
I do realise that there is variability within CBR - the fact that CBR filesizes varied with video content would be proof of that. The way I presented my figures probably masked it, however.
What truly surprised me was how much more compression you get with VBR when using just an image or colour clip. I can't imagine any rational argument for having to generate large file sizes for a colour clip - which CBR does.
As you say, CBR at 8000 kbps will fit something over an hour of video onto a DVD5, and maybe that's what most people would be doing. Most MiniDV tapes are an hour long, so it sort of ties in with that.
Using compressed audio allows you to push the maximum bitrate up to the maximum DVD-VR compliant 9800 kbps. With CBR, you might only get 60 minutes of typical video onto the disk - less when you add menus. If you use VBR, however, you might just squeeze an hour onto disk.
Most of the time, the difference might be slight - although in subjective tests I've done previously, I do think that the video quality is improved when you use VBR resulting in a file of comparable size to a CBR one.
Anyone doing a slideshow type video might see greater benefits, however. Perhaps you could fit 90 minutes of slideshow material onto a DVD5 using VBR 9800 kbps. (although the idea of sittting thru 90 minutes of holiday snaps might not be that appealing!)
Long credit rolls are certainly much more compressed with VBR, freeing up disc space for the use of a higher maximum bitrate. When I did a previous post on rolling credits, needless to say I test rendered to all kinds of settings and the VBR files looked good.
At the end of it all, I don't lose sight of the fact that the final video quality has much more to do with the quality of the original source material, and I am no great videographer. It's just an interesting little numerical exercise (for me at least) with some practical benefits, in some specific situations.
I have to admit that my initial response to THoff's post was quite snappy and had a more aggressive tone than I intended - but I was a bit miffed at his dismissive tone. I half expected to see the thread removed.
Thanks for your constructive response!
Last edited by 2Dogs on Fri Dec 09, 2005 3:58 pm, edited 1 time in total.
JVC GR-DV3000u Panasonic FZ8 VS 7SE Basic - X2
Thanks for that, John! I'm a firm believer in practical tests, though I've never respected the saying "if it works, don't unscrew it!" I'm going to fiddle about with a slideshow soon! In fact my daughter will be playing in a Christmas piano recital in a local shopping mall that normally translates to a DVD to all the parents with a slideshow part to it.jchunter wrote:2Dogs,
Well, I believe you because you have done the measurements.
JVC GR-DV3000u Panasonic FZ8 VS 7SE Basic - X2
-
THoff
2Dogs, just to let you know that my post wasn't meant as personal criticism.
But I want to make it clear that VBR encoding is still relative to an average bitrate, so the file size of a VBR and CBR-encoded video of the same duration will be essentially the same, give or take a few bytes.
The encoder is allowed to stray from the average bitrate up to the minimum and maximum that are specified (and you'll sometimes find the encoder exceed even that), but given that it will try to maintain the average bitrate, and a sufficiently long video, the size will be virtually the same as for a CBR-encoded video.
It is also important to recognize that the MPEG encoder included with Videostudio does not give you complete control over the VBR encoding parameters, even if the Advanced button is revealed by tweaking the UVS.INI file -- it does not allow independent control of the Minimum, Average, and Maximum bitrate.
But I want to make it clear that VBR encoding is still relative to an average bitrate, so the file size of a VBR and CBR-encoded video of the same duration will be essentially the same, give or take a few bytes.
The encoder is allowed to stray from the average bitrate up to the minimum and maximum that are specified (and you'll sometimes find the encoder exceed even that), but given that it will try to maintain the average bitrate, and a sufficiently long video, the size will be virtually the same as for a CBR-encoded video.
It is also important to recognize that the MPEG encoder included with Videostudio does not give you complete control over the VBR encoding parameters, even if the Advanced button is revealed by tweaking the UVS.INI file -- it does not allow independent control of the Minimum, Average, and Maximum bitrate.
-
jchunter
Torsten,
OK, but the fact remains that 2dogs ran some tests and measured the resulting file sizes. I think that you owe it to this forum to run similar tests and see what you get. This will either comfirm or disprove his work. We will all be better off knowing the facts.
It may turn out that Ulead has implemented a "nonstandard" VBR....
OK, but the fact remains that 2dogs ran some tests and measured the resulting file sizes. I think that you owe it to this forum to run similar tests and see what you get. This will either comfirm or disprove his work. We will all be better off knowing the facts.
It may turn out that Ulead has implemented a "nonstandard" VBR....
That's OK. I'm pretty thick-skinned really, anyway, but as Heinz-Oz might say, "Peace THoff"!THoff wrote:2Dogs, just to let you know that my post wasn't meant as personal criticism.
But my tests in VS9 do not show this. There is a difference in file sizes.But I want to make it clear that VBR encoding is still relative to an average bitrate, so the file size of a VBR and CBR-encoded video of the same duration will be essentially the same, give or take a few bytes.
All of this is clearly true - except that when using VBR in VS9, the only parameter you can alter is the bitrate, which shows up in the file "properties" when you interrogate it as being the maximum bitrate of the VBR. We have no access to anything that can DIRECTLY control the average bitrate.The encoder is allowed to stray from the average bitrate up to the minimum and maximum that are specified (and you'll sometimes find the encoder exceed even that), but given that it will try to maintain the average bitrate, and a sufficiently long video, the size will be virtually the same as for a CBR-encoded video.
Naturally I completely agree with this - since I pretty well said exactly the same thing in my post!It is also important to recognize that the MPEG encoder included with Videostudio does not give you complete control over the VBR encoding parameters, even if the Advanced button is revealed by tweaking the UVS.INI file -- it does not allow independent control of the Minimum, Average, and Maximum bitrate.
Now to advance the topic, I think I need to investigate how VBR performs with "suitable" source video. I see from many threads on slideshows that people sometimes link the use of VBR to video qualty problems - although it strikes me that there are usually many factors involved, and it is all too easy to see a causal relationship where none exists. Somehow I'll have to rope in a third party into my video quality judgements, so that I can do double blind tests, rather than suffer from my own wishfull thinking.
JVC GR-DV3000u Panasonic FZ8 VS 7SE Basic - X2
-
THoff
I really don't see what the point is of conducting such a test.
Either the average is a certain bitrate, and the bitrate multiplied by the duration of the video yields a certain size video, or the average bitrate is something else, but the file size will still be predictable. In any event, the file size will be essentially the same as for a CBR encoded video.
And nobody who has participated in this thread has tried to back up the statement that I took issue with -- not 2Dogs, or anyone else: VBR encoding itself will not in and by itself reduce the file size when compared to CBR encoding.
Either the average is a certain bitrate, and the bitrate multiplied by the duration of the video yields a certain size video, or the average bitrate is something else, but the file size will still be predictable. In any event, the file size will be essentially the same as for a CBR encoded video.
And nobody who has participated in this thread has tried to back up the statement that I took issue with -- not 2Dogs, or anyone else: VBR encoding itself will not in and by itself reduce the file size when compared to CBR encoding.
-
skier-hughes
- Microsoft MVP
- Posts: 2659
- Joined: Thu Jul 21, 2005 10:09 am
- System_Drive: C
- 32bit or 64bit: 64 Bit
- motherboard: gigabyte
- processor: Intel core 2 6420 2.13GHz
- ram: 4GB
- Video Card: NVidia GForce 8500GT
- sound_card: onboard
- Hard_Drive_Capacity: 36GB 2TB
- Location: UK
-
THoff
-
jchunter
I put five slides in the timeline, 4 seconds each. One was an ISO resolution chart. The data show that file space is saved by using VBR instead of CBR and by cutting the bitrate. However, picture quality is definitely hurt by cutting bitrate. I could not see a quality difference between VBR and CBR, nor could I see any quality difference between a Quality setting of 70% and 100%.
Bitrate __ VBR MB _ CBR MB ___ Resolution
8Mbps ___ 10.6 _____ 20.5 ____ 6 (h)/ 5.5v) fairly clean
6Mbps ____ 9.5 _____ 15.5 ____ 5.5 (h)/ 5.0v) some blurring
4Mbps ____ 8.7 _____ 10.6 ____ 5.5 (h)/ 4.5v) Jaggy, shadow artifacts
2Mbps ____ 5.1 ______ 5.6 ____ 5.5 (h)/ 4.5v) More Jaggy, large shadow artifacts
BTW, Torsten, 9MBPS VBR has a bug: it produces the poor quality of the 2Mbps even though its property shows 9Mbps. CBR 9Mbps file is OK at 22.9MB.
Bitrate __ VBR MB _ CBR MB ___ Resolution
8Mbps ___ 10.6 _____ 20.5 ____ 6 (h)/ 5.5v) fairly clean
6Mbps ____ 9.5 _____ 15.5 ____ 5.5 (h)/ 5.0v) some blurring
4Mbps ____ 8.7 _____ 10.6 ____ 5.5 (h)/ 4.5v) Jaggy, shadow artifacts
2Mbps ____ 5.1 ______ 5.6 ____ 5.5 (h)/ 4.5v) More Jaggy, large shadow artifacts
BTW, Torsten, 9MBPS VBR has a bug: it produces the poor quality of the 2Mbps even though its property shows 9Mbps. CBR 9Mbps file is OK at 22.9MB.
-
THoff
I'm going to test VBR vs CBR encoding at identical bitrates using Procoder when I get home, and post the results.
As was mentioned earlier, Videostudio does not permit full control of the encoding parameters (you cannot freely adjust the min/avg/max bitrate), so for the purpose of making the above comparison it is useless, because you never know what you'll get...
As was mentioned earlier, Videostudio does not permit full control of the encoding parameters (you cannot freely adjust the min/avg/max bitrate), so for the purpose of making the above comparison it is useless, because you never know what you'll get...