Question for Adam Wilt (and his response).

Toke Lahti · January 30th, 2006, 06:39 PM

Quote:

Originally Posted by Antoine Fabi

which codec is working in 10 bit 4:2:0 ?

Maybe infinity will have this as an option.
Funny thing is that 4:2:2 codecs are designed for interlaced picture, where vert rez is more important than horz rez. In progressive picture they are equally important, so when human vision is more accurate for luminance than for chrominance, having same vert luminnance & chrominance rez is waste of data. Much more important would be getting more tones for color correction.

Shannon Rawls · January 30th, 2006, 06:53 PM

Quote:

Originally Posted by Toke Lahti

4:2:2 codecs are designed for interlaced picture, where vert rez is more important than horz rez. In progressive picture they are equally important

Is that right? Well hot diggity. I wonder why everybody is so hellbent on "verticle" resolution when it comes to these cameras? I was under the impression horizontal rez is not all that important based on the way humans view images.

If this is the case, then that's good news for cameras with high horizontal rez that wish to do a film-out or one day be recorded to HD-DVD/BLURAY

- ShannonRawls.com

Toke Lahti · January 30th, 2006, 07:05 PM

Quote:

Originally Posted by Shannon Rawls

I was under the impression horizontal rez is not all that important based on the way humans view images.

Human eye's fovea's center, where the sharpest vision is, is a round thing with equal amount of horizontal and vertical cells. BTW, there are only couple of hundreds of them per dimension, so it is quite funny how much resolution we need to reproduce natural view. If we could just nail all audineces eyes on one point of picture all else could be just blurred. But everybody's eyes seem to wonder through the pictures with their own path and pace...

Of course for chroma keying and compositing 4:4:4 is always the best and it gets worse when you go further from that, but with natural image, there should be no way to separate progressive 4:2:2 picture from progressive 4:2:0 picture.

Antoine Fabi · January 30th, 2006, 07:24 PM

Quote:

Originally Posted by Toke Lahti

Of course for chroma keying and compositing 4:4:4 is always the best and it gets worse when you go further from that, but with natural image, there should be no way to separate progressive 4:2:2 picture from progressive 4:2:0 picture.

Oh !!!!

Here i completely disagree.

I mean, i can easily "see" the difference between 4:2:2 and 4:2:0
and i can easily "see" the difference between 4:2:2 and 4:1:1

Even watching the footage at a good distance from the monitor, i can actually see that 4:2:2 looks more "solid", but smoother (more natural) at the same time.

Robert Lane · January 30th, 2006, 11:30 PM

Quote:

Originally Posted by Toke Lahti

...but with natural image, there should be no way to separate progressive 4:2:2 picture from progressive 4:2:0 picture.

If you think that's true, take a look at Adams site and look at the direct color comparison he shows on the site. While 4:4:4 is king, there is a huge difference between 4:2:2 and 4:2:0. The difference is not only noticeable, it's downright ugly!

Barry Green · January 31st, 2006, 03:54 AM

Quote:

Originally Posted by Robert Lane

While 4:4:4 is king, there is a huge difference between 4:2:2 and 4:2:0. The difference is not only noticeable, it's downright ugly!

Just shot some stuff in 4:2:2 and 4:1:1; I think the difference is noticeable enough (and ugly enough!) that it deserves its own thread...

Toke Lahti · January 31st, 2006, 05:14 AM

Quote:

Originally Posted by Barry Green

Just shot some stuff in 4:2:2 and 4:1:1; I think the difference is noticeable enough (and ugly enough!) that it deserves its own thread...

I wasn't talking about 4:1:1.
That is about ugliest chroma resolution for progressive picture.
So are you guys seeing differences with 4:1:1 or 4:2:0?
Or just between 4:4:4 and everything else?

If you are seeing difference between 4:2:2 and 4:2:0 progressive natural images with full framerate playback, I'd guess that you are looking from too short distance, so that you are able to separate single luminance pixels.

Otherwise you guys have just trashed the theory behind component video and our perception has evolved in couple of decades.

I still haven't heard a single argument, why human eye should not notice lower chroma resolution horizontally, but should notice it vertically.

Antoine Fabi · January 31st, 2006, 10:24 AM

Toke,

4:2:2 is twice the color resolution of 4:2:0 or 4:1:1, twice !

It's not about satuation level, the 4:2:2 color will always appear smoother, richer, more natural than 4:2:0 and 4:1:1.

Yes, even at a good distance from the monitor, you can see the difference.

Toke Lahti · January 31st, 2006, 02:55 PM

Quote:

Originally Posted by Antoine Fabi

4:2:2 is twice the color resolution of 4:2:0 or 4:1:1, twice!

So you are saying that your eyes resolve chroma as sharply as luminance?
Or that your eyes can see the difference vertically but not horizontally?

Well, maybe this old story about component chroma resolution compression is just a legend. Maybe I'll have to do some tests with my own eyes.
Some of these old "laws of nature", that you learn very early, you just take for granted. Should question everything...

Antoine Fabi · January 31st, 2006, 04:41 PM

Quote:

Originally Posted by Toke Lahti

Maybe I'll have to do some tests with my own eyes.

Yes Toke !

David Heath · February 1st, 2006, 07:02 PM

I'd like to give Toke some support. In principle, the most desirable system would call for chroma to have the same resolution as luminance - 4:4:4. For many systems it is unfortunately necessary to use techniques to achieve bitrate reductions, compression is obviously one, reducing the chroma resolution relative to luminance is another.

If the latter technique is employed, it makes sense to do it symmetrically in the vertical and horizontal directions for all the reasons Toke says - it suits the way the eye works. Hence 4:2:0 seems a far more logical choice than 4:1:1, and I believe the latter originates from when digital recordings were seen as 'islands' in an analogue world - it was reckoned to survive repeated A-D conversions better than 4:2:0, especially true when interlace is taken into account.

It may be argued that extra chroma res can't do any harm, so 4:2:2 must be better than 4:2:0, mustn't it? I can see the reasoning there, but the bitrate to allow it must come from somewhere, and in 4:2:2 systems where bitrate is constrained that has to mean that the whole image is more highly compressed than if 4:2:0 was used. Hence, is it not conceivable that a better compromise may result from less vertical chroma resolution if it allows lower overall compression, or even 10 bit working?

Heath McKnight · February 3rd, 2006, 04:24 PM

I've been working with the FX1 and Z1 for a year now, and the 4:2:0 color space is really nice. I am looking forward to using the HVX soon and see how the 4:2:2 looks. Until then, I'll wait to pass judgment, but until then, I like the Z1's color space, and the HD100 has a nice image, too.

I'm the kind of guy who likes to put a camera through its paces then look at footage on my HDTV and a calibrated HD monitor before I decide what's right.

heath

David Heath · February 3rd, 2006, 05:11 PM

Quote:

Originally Posted by Heath McKnight

I've been working with the FX1 and Z1 for a year now, and the 4:2:0 color space is really nice. I am looking forward to using the HVX soon and see how the 4:2:2 looks. Until then, I'll wait to pass judgment, ...........

Whatever your final conclusion is, the differing colour spaces are only one aspect of the differing recording systems and codecs. With one an intraframe 100Mbs system, the other 25Mbs interframe, and differing luminance resolutions, all those alone may make more difference than the differing colourspaces. And that's before we even start thinking of the cameras having different lenses, chips, processing.......

Practically though, I'll agree with your second paragraph!

Heath McKnight · February 3rd, 2006, 05:18 PM

I'll know more shortly.

heath

Toke Lahti · February 10th, 2006, 04:19 AM

I'd say that hvx is just the first step out of chains of locked tape formats.
With this camera you can choose two resolutions and variable frame rate.
Maybe in next model, you could choose different codec, adjust compression ratio, color depth and sampling.
1280x720@10bit@4:2:0 would need only 25% more with same compression ratio than today's 960x720@8bit@4:2:2.
So 10bit colors with 24pN and bitrate of dvcpro50. Doesn't that sound good?

February 3rd, 2006, 04:24 PM	#42
Heath McKnight MPS Digital Studios Join Date: Apr 2003 Location: Palm Beach County, Florida Posts: 8,531	I've been working with the FX1 and Z1 for a year now, and the 4:2:0 color space is really nice. I am looking forward to using the HVX soon and see how the 4:2:2 looks. Until then, I'll wait to pass judgment, but until then, I like the Z1's color space, and the HD100 has a nice image, too. I'm the kind of guy who likes to put a camera through its paces then look at footage on my HDTV and a calibrated HD monitor before I decide what's right. heath __________________ My Final Cut Pro X blog

February 3rd, 2006, 05:18 PM	#44
Heath McKnight MPS Digital Studios Join Date: Apr 2003 Location: Palm Beach County, Florida Posts: 8,531	I'll know more shortly. heath __________________ My Final Cut Pro X blog

January 31st, 2006, 10:24 AM	#38
Antoine Fabi Regular Crew Join Date: Mar 2004 Location: sherbrooke (Quebec) Canada Posts: 108	Toke, 4:2:2 is twice the color resolution of 4:2:0 or 4:1:1, twice ! It's not about satuation level, the 4:2:2 color will always appear smoother, richer, more natural than 4:2:0 and 4:1:1. Yes, even at a good distance from the monitor, you can see the difference.

February 1st, 2006, 07:02 PM	#41
David Heath Inner Circle Join Date: Jan 2006 Posts: 2,699	I'd like to give Toke some support. In principle, the most desirable system would call for chroma to have the same resolution as luminance - 4:4:4. For many systems it is unfortunately necessary to use techniques to achieve bitrate reductions, compression is obviously one, reducing the chroma resolution relative to luminance is another. If the latter technique is employed, it makes sense to do it symmetrically in the vertical and horizontal directions for all the reasons Toke says - it suits the way the eye works. Hence 4:2:0 seems a far more logical choice than 4:1:1, and I believe the latter originates from when digital recordings were seen as 'islands' in an analogue world - it was reckoned to survive repeated A-D conversions better than 4:2:0, especially true when interlace is taken into account. It may be argued that extra chroma res can't do any harm, so 4:2:2 must be better than 4:2:0, mustn't it? I can see the reasoning there, but the bitrate to allow it must come from somewhere, and in 4:2:2 systems where bitrate is constrained that has to mean that the whole image is more highly compressed than if 4:2:0 was used. Hence, is it not conceivable that a better compromise may result from less vertical chroma resolution if it allows lower overall compression, or even 10 bit working?

February 10th, 2006, 04:19 AM	#45
Toke Lahti Major Player Join Date: Oct 2004 Location: Helsinki, Finland Posts: 158	I'd say that hvx is just the first step out of chains of locked tape formats. With this camera you can choose two resolutions and variable frame rate. Maybe in next model, you could choose different codec, adjust compression ratio, color depth and sampling. 1280x720@10bit@4:2:0 would need only 25% more with same compression ratio than today's 960x720@8bit@4:2:2. So 10bit colors with 24pN and bitrate of dvcpro50. Doesn't that sound good?