Is B.T 709 conversion produce more colours

[Aston]

While converting from RGB to YCbCr color space we can use following two options.
1. 601-quotients
2. 709-quotients

I understood that 601 for standard type and it's old. Latest HDTV's use 709 type. Is this right ?? Is this right ??

Can 709 quotients produce more colours than 601 type while colorspace conversion ??

[hello_hello]

The different methods relate to the formula used for converting RGB to YUV and then back again to RGB. On playback, HD uses R.709 and SD definition uses R.601, so when converting RGB you need to use the appropriate one. Renderers/players will often make the choice of which to use based on the video's resolution. If the wrong one is used, the colours will look a little off (reds get a bit darker while greens and blues get a little brighter, I think... or the other way around). However if you take a HD YUV video and convert it to SD YUV, and you convert from 709 to 601 in the process (which you should) then both the HD and SD video should display using the same colours because the former will be converted to RGB on playback using 709 while the latter with 601. So basically if you're converting RGB to HD YUV then use R.709, if you're converting to SD use R.601.

If you have MPC-HC and ffdshow installed and want to see the difference in action.....
Open a video with ffdshow decoding and maximise MPC-HC. Open the ffdshow video configuration and enable the resize filter. If it's a SD video (ie 854x400) change the resizing to 1920x1080. With MPC-HC maximised I'm pretty sure the video will just continue to fill the screen as you enable/disable the ffdshow resize filter, but the colours will change a little. It's not a huge difference so it's more obvious when playing some video than others, but if the original video is SD and you use ffdshow to resize to HD, if the colours change to r.709 as a result then the video is displaying incorrectly.

Some programs or filters etc use one or the other without giving you an option. If that happens (say you're converting RGB to HD YUV) and the program converts RGB to YUV using R.601, then you'd need to manually apply an appropriate conversion yourself if possible to correct it. I'm pretty sure if you feed RGB video directly to the XviD and x264 encoders, they use R.601 by default. I think x264 has an option to tell it to use R.709 (I'd have to check) but I don't think XviD does. On the other hand, if you happen to be converting YUV video to RGB and back again while editing, it doesn't matter which you use as long as you use the same method both times.

[Aston]

Hi hello_hello,

1. From your comments what I understood is, R-709 conversion will not produce more colors but for HD video conversion(720p RGB to 720p YCC) it will give better image compare to R-601 conversion. Is this right ??
2. And for 4k2k resolutions also these R-709 coefficients are enough are do we have another coefficients for this 4k2k. And these R-709 coefficients produce better image up to which range (resolution) ??

[sanlyn]

Narendra, you have it completely incorrect. The "601" and "709" standards don't deliver "more" color or "less" color, they are simply different color matrices. Referring to BT.601 as "old" and BT.709 as "not old" is absurd, they are simply different and they are have different uses. It is equally absurd to say that one is "better" than the other. It is likely that in the future, all video whether standard def or HD or whatever will use the same matrices, and that will be yet another "newer" system for interpreting and displaying color from various sources. The "BT" matrices are used, for example, to convert between YUV video data storage and RGB video display.

BT601 and BT709 are not the only color matrices. Try looking up SECAM and a few others, and matrices that are used in the space program and various scientific pursuits. When converting back and forth for proper storage media or for proper display hardware, one should use the matrices that is appropriate for matching the source with the display. To say that one is "better" than the other is not correct.

[hello_hello]

Hi hello_hello,

1. From your comments what I understood is, R-709 conversion will not produce more colors but for HD video conversion(720p RGB to 720p YCC) it will give better image compare to R-601 conversion. Is this right ??

No, it's not about better, it just the method used for converting the YUV video to RGB, Why the standard was even changed from R.601 to R.709 for HD video is beyond me, it just adds an additional area where things can go wrong and as far as I know, offers no real benefit. The process of converting RGB to YUV isn't technically lossless regardless of the method used, but that aside (for the sake of the discussion we'll assume it is) if you encode RGB video using one method and then convert it back to RGB using the same method, then you're back to where you started whether you used R.709 or R.601. So when converting RGB to YUV, the decision as to which method to use comes down to which method will be used when the encoded video is converted back to RGB for display on the monitor TV. It should always be R.709 for HD and R.601 for SD, so you'd also do it that way when encoding.

The colours aren't "better" or "worse" as such according to the method used. If one method is used to convert YUV to RGB while a different one is used to convert it back to RGB for display, then the colours will be a little "wrong". If the same method is used each time, they'll be "correct".
Think of it this way.... if you take a file and compresses it with a zip algorithm, you have to uncompress it the same way to get the original file back. It's not exactly the same thing, but it's the best analogy I can come up with at the moment.....

2. And for 4k2k resolutions also these R-709 coefficients are enough are do we have another coefficients for this 4k2k. And these R-709 coefficients produce better image up to which range (resolution) ??

To the best of my knowledge the higher resolutions still use R.709 when decoding YUV video.

[Aston]

Why the standard was changed from R.601 to R.709 for HD video ??

[hello_hello]

Have a read of this. It looks like I was wrong when I said neither system is better, because according to this article, technically R.709 is better. Not enough to make up for the hassle of having two methods of converting colours though, according to the author.

Everything else I said still holds true, and is confirmed in the article. The correct colorimetry should be used according to the video definition, when converting RGB to YUV, as well as when up-scaling or down-scaling.

http://www.poynton.com/papers/SMPTE_98_YYZ_Luma/index.html

It's actually something I've been meaning to test for quite a while, but as I pretty much always use my PC for playback and I know exactly what it does... ie when the video is converted using the correct colours and when it isn't.... and I know how to correct them on playback manually, I've never tested my TV or Bluray player to see what they do.
For instance if I play a standard definition video using the Bluray player with it set to upscale to 1080p, does the video display using the same colors as it would if the Bluray Player was outputting PAL/NTSV resolution and the TV was doing the upscaling? And what about the media player in the TV? Does it use the correct colors according to resolution?

At some stage I might try to test the above, although unlike when I've testing it all using a PC for playback, I can't just run two videos simultaneously on the TV and easily switch between them, so I guess I'll have to try to find a video where it's obvious when the colours are wrong.

Here's an example for you. The first pic is an encode taken from HD video while reducing it to SD without converting the colors. So while it's being converted to RGB on playback using R.601 because it's SD, it should be done using R.709 for it to display correctly.


[Attachment 16714 - Click to enlarge]

Here's the same screenshot again, but the colours have been corrected so it'll display as it should.


[Attachment 16715 - Click to enlarge]

Because they're animation it's impossible to tell which is correct unless you know.... in that neither are "true to life" in terms of colour.... but I picked it as animation has less "shades" than real-life video, it should make the difference easier to see.

[jagabo]

When rec.601 is decoded as rec.709, and vice versa:
https://forum.videohelp.com/threads/329866-incorrect-collor-display-in-video-playback?p...=1#post2045830

[sanlyn]

@hell0_hello, yes, it does make a difference if video is not correctly matched by matrix. But if you go to the trouble of changing Rec601 SD video to rec709 SD to make it "better" (skeptical there, but I'm open to it), problems occur with playback components.

Not long ago we had a TV station engineer check some DVD players for our local movie fan club (8 old farts and two young sprinters who luv old movies). He sez our DENON and OPPO players output 601 for SD and 709 for HD disc. I don't know what my old Toshiba and Cambridge Audio players do, but I assume it's 601 for everything. My TV's have an Auto setting for colorspace. I've never changed it.

As you know from another current thread, Rec601 -> Rec709 conversions aren't so simple.

Historically, both "Rec" matrices have been changed/upgraded several times.

[hello_hello]

@hell0_hello, yes, it does make a difference if video is not correctly matched by matrix. But if you go to the trouble of changing Rec601 SD video to rec709 SD to make it "better" (skeptical there, but I'm open to it), problems occur with playback components.

I don't know if we're talking about different things, but I wasn't saying you should change from one to the other to make it better, only to make it "correct".

The second screenshot would be how the original HD video should display using R.709. After it's downscaled to SD and re-encoded, on playback it'll probably be converted to RGB using R.601, so the colours won't be correct, as per the first screenshot., The second screenshot therefore shows it displaying correctly if it's converted to R.601 while it's encoded.

[sanlyn]

Well, yes, I think that by "better" we both mean "correct". Which would be, well, better than incorrect.