Video format Q

[Tafflad]

If I click on one of my video clips in the Vegas it gives the file data .. For example:

Video: 1920x1080x12, 59.940 fps etc.

Now I know what 1920 x 1080 is, and I know what fps is ... but what does the '12' stand for ?

[smrpix]

Bit depth.

[jagabo]

The video uses chroma channels that are half the size (each dimension) of the luma channel. So your 1920x1080 video has a 1920x1080 luma (greyscale, Y) channel and two 960x540 chroma (color, U and V) channels. Each channel consists of 8 bit values. So there are 12 bits per pixel on average.

[Cornucopia]

Yes, to put it in a more tabular format:

(some common ones)
RGB @ 12bit/channel = 12 + 12 + 12 = 36bits/pixel.
RGB @ 10bit/channel = 10 + 10 + 10 = 30bits/pixel.
RGBA @ 8bit/channel = 8 + 8 + 8 + 8 = 32bits/pixel.
RGB @ 8bit/channel = 8 + 8 + 8 = 24bits/pixel.
YUV 4:4:4 @ 8bit/ch = 8 + 8 + 8 = 24bits/pixel.
YUV 4:2:2 @ 8bit/ch = 8 + 8/2 + 8/2 = 8 + 4 + 4 = 16bits/pixel.
YUV 4:2:0 @ 8bit/ch = 8 + 8/2/2 + 8/2/2 = 8 + 2 + 2 = 12bits/pixel.
YUV 4:1:1 @ 8bit/ch = 8 + 8/4 + 8/4 = 8 + 2 + 2 = 12bits/pixel.

BTW, you'd get much more detailed (and usually more correct) info using MediaInfo.

Scott

[Tafflad]

attached is media info off a clip from my camera

it shows bit depth is 8 ... yet the file format shows as 12, or am I confusing things.

So when you see a reference to 16-bit colour or 24-bit colour is this is figure that equates to that ?
Inf that is teh case looks like my nice new camera is not too good at only 12 bit

[Cornucopia]

You seem to be confusing things.

Your MI readout (thank you very much, BTW)shows YUV, 4:2:0, 8bits as "bit depth". Looking at my list you see that works out to an equivalent 12bits/pixel. Note also that it lists the bit/pixel/frame average as 0.280. This is due to the AVC compression - ~50:1!

The 8bits is the bits/color channel coding standard.The 12bits is the overall calculated average bits/pixel (when not compressed), and the 0.280bits is the actual average bits/pixel/frame (when taking into account the compression). That last one often explicitly includes the "per frame" reference because interframe codecs will vary quite a bit between their I, B, & P frame types.

3 different numbers, yet they refer to different aspects of the file, so they are not in disagreement.

Scott

[Tafflad]

Thanks for the explanation.

Came about as I noticed friends video clip with same frame size has a '24' where mine were 12.
So presumably means his video clips have a lot more colour information. Even though both providing the same 1080p video.

[newpball]

Thanks for the explanation.

Came about as I noticed friends video clip with same cane size has a '24' where mine were 12.
So presumably means his video clips have a lot more colour information. Even though both providing the same 1080p video.

Yes unfortunately color resolution reduction is prevalent!

Color compression is obviously something that should be handled by the fine surgical knife of using human color perception by the CODEC but instead the machete was used and color information was simply cut across the board. Brilliance in the making!

[Tafflad]

Strange as its a new Canon camera .. but at least the images look good at full HD
I'll know for next one to check colour bit depth ... although just looked at specs for camera .. it does not give this information.

[poisondeathray]

Don't worry about it - ALL consumer level cameras record subsampled 4:2:0. Even BD is 4:2:0

That's actually one of the things that the "engineers" got right. Chroma subsampling is an excellent way of reducing bandwidth on most material if human eyes are watching it. Human eyes are not as sensitive to color information. It is fine for an end delivery format in most scenarios.

To put it another way, if you had the same bandwidth , but recorded 4:4:4, the image quality would be visibly lower in most consumer recording bitrate range scenarios, because more bitrate is "wasted" in attempting to preserve the color information.

[Tafflad]

Thanks for answ

[newpball]

Don't worry about it - ALL consumer level cameras record subsampled 4:2:0. Even BD is 4:2:0

That's actually one of the things that the "engineers" got right. Chroma subsampling is an excellent way of reducing bandwidth on most material if human eyes are watching it. Human eyes are not as sensitive to color information. It is fine for an end delivery format in most scenarios.

To put it another way, if you had the same bandwidth , but recorded 4:4:4, the image quality would be visibly lower in most consumer recording bitrate range scenarios, because more bitrate is "wasted" in attempting to preserve the color information.

Wrong, doing this kind of compression at the CODEC level is much more efficient as you can determine the amount depending on the situation.

Just look at CODECs like Redcode!

[poisondeathray]

Don't worry about it - ALL consumer level cameras record subsampled 4:2:0. Even BD is 4:2:0

That's actually one of the things that the "engineers" got right. Chroma subsampling is an excellent way of reducing bandwidth on most material if human eyes are watching it. Human eyes are not as sensitive to color information. It is fine for an end delivery format in most scenarios.

To put it another way, if you had the same bandwidth , but recorded 4:4:4, the image quality would be visibly lower in most consumer recording bitrate range scenarios, because more bitrate is "wasted" in attempting to preserve the color information.

Wrong, doing this kind of compression at the CODEC level is much more efficient as you can determine the amount depending on the situation.

Just look at CODECs like Redcode!

Redcode isn't a "consumer" format. There is not enough bandwidth on consumer media like SD cards to record compressed raw even

Sure, if you had SSD's , external recorder, the right camera, full pipeline you can even record RAW (true raw, not redcode compressed RAW). You get what you pay for, these are completely different markets and price ranges . Acquisition and final delivery format are 2 different things - even the next gen UHD BD will be 4:2:0.

[newpball]

Don't worry about it - ALL consumer level cameras record subsampled 4:2:0. Even BD is 4:2:0

That's actually one of the things that the "engineers" got right. Chroma subsampling is an excellent way of reducing bandwidth on most material if human eyes are watching it. Human eyes are not as sensitive to color information. It is fine for an end delivery format in most scenarios.

To put it another way, if you had the same bandwidth , but recorded 4:4:4, the image quality would be visibly lower in most consumer recording bitrate range scenarios, because more bitrate is "wasted" in attempting to preserve the color information.

Wrong, doing this kind of compression at the CODEC level is much more efficient as you can determine the amount depending on the situation.

Just look at CODECs like Redcode!

Redcode isn't a "consumer" format. There is not enough bandwidth on consumer media like SD cards to record compressed raw even

Sure, if you had SSD's , external recorder, the right camera, full pipeline you can even record RAW (true raw, not redcode compressed RAW). You get what you pay for, these are completely different markets and price ranges . Acquisition and final delivery format are 2 different things - even the next gen UHD BD will be 4:2:0.

You are completely missing the point.

Due to human color perception one can take advantage of compression of colors, however a compression method that cuts resolution situationally is obviously much better than one that simply cuts resolution like a machete knife for all situations.

A 4:4:4 CODEC doing situational color compression can obviously reach a higher quality/compression ratio than one that simply starts with a machete cut 4:2:0.

That this obvious observation even needs to be discussed.....

[poisondeathray]

Due to human color perception one can take advantage of compression of colors, however a compression method that cuts resolution situationally is obviously much better than one that simply cuts resolution like a machete knife for all situations.

That this obvious observation even needs to be discussed.....

You already missed the point.

If cost was no object, sure it's better to have more options and flexibility if you are able to record it. But how does it help you if you can't even record the data ?

These things cost more money and bandwidth. Pretty obvious real life considerations, not some hypothetical discussion