Hi,

I recently captured a 50 minute film with 4 tracks of audio at 2:1 SD, 25i, to Unity. Being used to averaging about a GB of drive space for uncompressed capture, I was surprised to find that the video came in at only about 14GB in size.

I just did some test captures  - one minute duration each, and came up with the following file sizes:

1 min 1:1 video: 1 255 452 KB

1 min 2:1 video: 398 370 KB

1 min DV25 420: 213 020 KB

1 min 10:1 video: 106530 KB

1 min 15:1s video: 18466 KB

I know it's probably not a case of simple division, but the 2:1 seems more than three times smaller than 1:1, while the 15:1s which I would've thought would be 30 times smaller, appears to be more than 60 times smaller.

Has this always been the case or is there something new happening with the codecs? Certainly when the Capture Tool is estimating drive space it still seems to use a simple 1:1, 2:1 ratio.

Cheers

I believe that the compression depends on the compexity of the images.  For example very fast moving images with alot of detail won't compress as easily as a simple cartoon.  So to make your test meaningful you would have to try with a variety of different material.

As far as 15:1s goes remember that it's half the height and half the width so that's a quarter frame to begin with. 4 x 15 = 60

regards

Brian

bsuttie:

I believe that the compression depends on the compexity of the images.  For example very fast moving images with alot of detail won't compress as easily as a simple cartoon.  So to make your test meaningful you would have to try with a variety of different material.

As far as 15:1s goes remember that it's half the height and half the width so that's a quarter frame to begin with. 4 x 15 = 60

Of course, that explains the 15:1 s. Although I was under the impression that with the 2:1, 10:1 etc compression ratios it's a fixed compression, so your images will look better or worse depending on the complexity of the media? It does state an actual ratio. Wasn't it the AVR compression ratios that did a variable compression?

It sounds like to two of you are talking about different forms of compression.  Gerrit you are trying to find information about the the amount of space that is used when capturing/digitizing under different resolution settings.  Bsuttie, it seems like you are referring to compression like doing DVD encoding.  In a DVD encoder, the software goes through and looks at the images and determines where more or less compression is needed (assuming you use a VBR setting)  When capturing in AVID the data rates for the different resolutions are the same no matter what is being captured.

i.e.  You could capture a fast moving music video with hundreds of cuts in it at 1:1 and you would use up the same amount of drive space as if you captured just black at 1:1.

Hi Ducktech,

Yes, that's what I mean. Obviously I'm happy as I'm managing to fit more material on the drives, but if the 2:1 compression is actually compressing more in the region of 3:1 or 4:1 as the test seems to indicate, that's not so good.

There is a definite discrepancy between how much drive space the Capture Tool seems to think it needs for a specific clip, and how much it actually uses once captured. The other ratios seem close to what one would expect from them, it's just 2:1 that seems way off.

I think you're getting caught up in the numbers. 

Aside from the all the compression ratios and drive space available you have to ask yourself - Is the image on the monitor acceptable?  You're the only one that can answer that question.  To me the difference between 2:1 and 1:1 is negligible.  I generally work at 1:1 for broadcast, other high-end projects and greenscreen work.  But for most everything else 2:1 gets the job done.  Digitize a couple of shots at 1:1 and at 2:1.  Then evaluate the shots on a HQ monitor.  If they are identical to the naked eye - Does it matter that the 2:1 video takes up a less drive space than you think it should?

You're entirely correct Haze, and we do the same, sometimes working at 2:1 as it's definitely good enough. I just found it curious as the numbers didn't make sense, and with a pretty full Unity the estimated drive space sometimes prevent you from capturing when there obviously will be enough actual space based on these results.

Also, haven't really tried a serious grade on the video, whether any artefacts might show if pushed beyond a certain point.

But yes, if it looks good enough, it probably is.

 When capturing in AVID the data rates for the different resolutions are the same no matter what is being captured.

i.e.  You could capture a fast moving music video with hundreds of cuts in it at 1:1 and you would use up the same amount of drive space as if you captured just black at 1:1.

So why when I just tried it did 10 seconds of programme at 2:1 come out at 62,468KB and 10 seconds of black at 2:1 came out at 4,228KB?

Try it

Brian

ducktech2:

Bsuttie, it seems like you are referring to compression like doing DVD encoding.  In a DVD encoder, the software goes through and looks at the images and determines where more or less compression is needed (assuming you use a VBR setting)  When capturing in AVID the data rates for the different resolutions are the same no matter what is being captured

It's because Avid uses a fixed compression algorithm that file sizes vary.  Bsuttie is quite right, as his subsequent tests show.  Broadly, the more complex an image is under this type of algorithm the more space it will occupy after compression.  An all black frame will inevitably take a tiny amount of space compared to a more standard image.

The comparison with VBR compression is exactly the opposite of what occurs.  One of the purposes of VBR compression is to maintain the frame data size after compression.  To do this it dynamically varies the compression rate.  In that case there will be considerably less variation between an all black frame and a standard image.

jwrl:

 

 

The comparison with VBR compression is exactly the opposite of what occurs.  One of the purposes of VBR compression is to maintain the frame data size after compression.  To do this it dynamically varies the compression rate.  In that case there will be considerably less variation between an all black frame and a standard image.

I might be misunderstanding you but having worked with VBR compression quite a lot using Sonic Scenarist, what it will strive to do is to maintain the average bitrate specified, rather than the frame data size. Higher bitrates would be devoted to complex material, with hardly any to black. So on an encode with a specified average of 6.5 Mb/s, the GOPs during complex material will be significantly larger in size than black, which is why you run into problems when encoding a piece containing a lot of black, as it really lowers the bitrate on these sections, and then in an attempt to reach the specified average bitrate it would overcompensate on complex material, ending up with peaks of 11Mb/s, which might cause stuttering. Reducing the margins between top and bottom ceilings helps limiting these peaks, but in effect you end up wth something more akin to a constant bitrate encode.

You're both right and wrong.  Yes it's actually the data throughput that is maintained.  In an attempt to shorten my explanation I oversimplified.  But in introducing DVD encoding as the basis of your analysis you are not comparing apples with apples.

Bottom line?  Bsuttie is right in his explanation.

Sorry to be going on about this, but it's more for my own clarity. I just found this:

http://www.avid.com/nab2005/storageCompressionChart.pdf

Clearly the Meridien resolutions are indicatied as having a fixed duration per GB of storage space, as opposed to the AVRs which are more flexible, and seemingly fit Bsuttie's explanation better. I also seem to recall Meridien calculators doing the rounds, which would be obsolete if the amount of storage used was down to the source material.

Also found these two explanations in another Avid pdf.

An AVR is referred to
as a constant quality M-JPEG resolution since the same
quantization table (of coefficients) is applied to each frame
of a video clip during digitization. For any given AVR, the
actual compressed data rate will increase as the
complexity of the imagery increases. For example, a head
shot typically results in a low data rate while a crowd shot
from a sporting event will yield a high data rate.

JPEG File Interchange Format – a compression scheme
used by Avid Technology in its Meridien hardware-based
non-linear systems. A JFIF M-JPEG resolution is termed
constant rate since compressing clips of varying
complexity results in a fixed data rate.

Are you running Meridien?  My assumption was that you were running Adrenaline, as bsuttie is.

The tests were on Media Composer software with SDI Mojo. I take it 2:1 on MCsoft and Adrenaline is a completely new codec then to 2:1 Meridien? That would explain it behaving differently. My assumption was it is similar. I seem to recall for example Quicktime export still referring to Avid Uncompressed and Compressed, with possibly a Meridien reference in there as well, might be completely mistaken.

2:1 Meridien = 2:1 DNA