Linear Workflow and Rec. 709

[dennisqdw]

Hi all, max 2016 has spurred me to do further research into linear workflow settings. I found this interesting webpage (http://www.aversis.be/tutorials/vray...orkflow_01.htm) which suggests that there is a difference between an image in the vray buffer as seen with the sRGB switch on, and an image rendered with embedded gamma of 2.2. Although it's a minor difference. HDTV color space however is Rec. 709, and from what I've read the gamma for that is about 1.9. This would be an even larger difference than either viewing with sRGB or embedded gamma of 2.2. I understand of course that rendering 16 bit or 32 bit exr images gives me the color space flexibility I'd need when the files are ported to AE or something similar. However, I'd obviously like to look at rendered images while in max that will be as close as possible to the end result. So what are my choices here? Should I render to an embedded gamma 1.9 and not use the sRGB switch? I'm certain that smarter people than me have already resolved this.
'

[dennisqdw]

One more bit of information. Rendering an image in max (using the "standard" linear settings) and viewing it with the sRGB switch on, and comparing that image to the way it looks in AE using sRGB workspace, there is a minor difference. The AE image looks just a little bit lighter in the shadows. Not sure how this relates to my main question, but I thought I'd mention it anyway.

[vlado]

You can use LUTs or OCIO/ICC profiles in the V-Ray VFB to preview your output accurately, instead of sRGB.

Best regards,
Vlado

[dennisqdw]

Thanx I'm setting that up now. --ds

[dennisqdw]

Ok another question. I can see that I can load the icc profile. But according to your docs "This correction can be controlled with environment variables. Setting the VRAY_VFB_ICC_FILE allows to automatically specify an ICC profile." I did this, but I'm not sure how I know if it's doing anything. I restarted max and opened the frame buffer, but it doesn't list the profile in the ICC slot of the frame buffer. Is it happening in a "hidden" way nonetheless? I guess I can "save the preset" but I thought I'd give the variable method a shot.

[Zdravko.Keremidchiev]

There seems to be an issue with that variable, I add it in our bug tracking system.
For now please use the VFB ICC option.

[dennisqdw]

Ok, good to know. Glad I could help .

[im.thatoneguy]

There is a common misconception that you should preview your renders with a Rec709 LUT on your computer monitor. This is completely wrong. If you want to see what your render looks like, use your monitor's LUT which is probably sRGB. When you output your final output apply a Rec709 LUT on top of the linear output. It will look *wrong* on your computer monitor just as looking at a linear exr without a lut will look *wrong* on your computer monitor. That's because before display your computer monitor applies an inverse-sRGB LUT to linearize the data for display (after all photons are 'linear'). In the case of most televisions they will apply an inverse Rec709 LUT before outputting linear data (assuming all content is Rec709, just as a computer monitor assumes all content is sRGB). If you feed a TV sRGB the TV will apply an inverse rec709 LUT internally which will make the image look crushed and bad. Conversely if you feed a computer monitor Rec709 it will look all washed out and dull. You can color correct your Rec709 render on your computer monitor to look "Good" but then on a TV it'll look crushed again, because you essentially just color corrected it back to your monitor's native gamma sRGB.

The best way to work is to keep everything linear and simply apply whatever Gamma you will be viewing the content on. If you're working on an sRGB monitor use an sRGB viewing gamma. If you're working on a television use a rec709 gamma. If you are going to be presenting your render on an sRGB monitor apply an sRGB gamma correction, if you will be presenting your render on a television use a rec709 gamma.

It gets even more complicated when you start dealing with "Smart" image formats. For instance AVID's DNxHD carries a gamma flag in it for sRGB or Rec709. So does ProRes. So if you render out an animation in sRGB and you apply the sRGB flag in the Quicktime container a 'proper' prores player should convert dynamically your sRGB ProRes to a Rec709 ProRes on a TV (or vice versa). But not all players properly do this. That's why when you watch some H264s which are often encoded in Rec709 they look all washed out, that's because the player isn't converting the Rec709 video file to sRGB and is presenting it incorrectly (I'm looking at you Quicktime for Windows).

[joconnell]

On a totally different note - thanks a million for writing and releasing Nuke 'em - I use it a few times every month, will buy you a beer if we ever meet in person

[dennisqdw]

I would of course agree with you entirely. Except that I use one of those HP LP2480zx 10 bit color monitors where you can control the color space of the imagery. And one of my options is rec. 709. On the other hand I do recognize the difficulty in that potentially some of my clients will be viewing in sRGB. So I have to plan accordingly, and perhaps just going that route would make everything much easier on me. And the difference is almost negligible to most people anyway. But I'm a little obsessive sometimes
But thanx for the response. I had kind of forgotten all about this post.

[im.thatoneguy]

@Joconnell, Thanks! Btw, I just updated it for the 2016 PhysicalCamera.

You can definitely use a monitor in rec709 space but then *everything* has to be rec709 space which causes problems in applications which don't manage color very well like... say... windows Since most applications default to sRGB I think it's easier to work in sRGB until you output your masters and just trust the math to handle the conversion. Viewing Linear converted to rec709 on a rec709 monitor should look identical to viewing linear converted to sRGB on an sRGB monitor. There are far less problems with applications like Nuke, Maya or Max's UI on rec709 which will display improperly.

[dennisqdw]

You raise a good point about sRGB being easier to work in, and indeed I did work in that until just recently, and I very well may go back, and yes there are some apps that do have rec709 issues, or don't even offer it as an option. One quibble the color gamuts of rec709 and sRGB *are* different. Not that most people would notice, unless doing side by side comparisons. And so Viewing Linear converted to rec709 on a rec709 monitor *would not* look identical to viewing linear converted to sRGB on an sRGB monitor. And that's the whole problem with different color spaces -is that they are different. With some -not all- of my work I (knowing that it will be going to HD bluray) I want to ensure the best possible color match. Knowing of course that the end viewer is likely to not have anything calibrated anyway ;-(

[im.thatoneguy]

One quibble the color gamuts of rec709 and sRGB *are* different.

Nope, they share the same white point D65 and primaries:

		xW     yW     xR   yR   xG   yG   xB   yB
ITU-R BT.709 	0.3127 0.3290 0.64 0.33 0.30 0.60 0.15 0.06 
sRGB		0.3127 0.3290 0.64 0.33 0.30 0.60 0.15 0.06

[dennisqdw]

Hi, at one point I had read that there is a small difference between the two, and here are two sources (http://www.helios.de/web/EN/color_fidelity.html#rec709) (http://liftgammagain.com/forum/index...ge-3#post-8935) the latter gives a technical explanation as to why even thought the primaries match, there still is a difference. And additionally there is a "potential" gamma difference. But everyone seems to agree that whatever differences there are are negligible.

[im.thatoneguy]

Juan is saying that video playback software is inconsistent in applying the correct gamma transform for rec709 footage on an sRGB computer (hence washed out quicktimes on windows) which is true, but that's a software playback issue, not a gamut issue. If you are rendering to linear and monitoring in sRGB then outputting Rec709 and use the correct output LUT and the correct viewing LUT you should come out the same. Also he's backwards, Rec709 used to just be "whatever TVs on average are" which at one point used a somewhat arbitrary Sony television that was popular in studios as the recommend a specific gamma. However it's slowly evolved over the years seeing as nobody uses CRTs. NTSC we were always told in film school stood for "Never Twice the Same Color". Rec709 is less consistent in my experience than sRGB. sRGB *does* have a consistent gamma (ignoring past PC vs OSX differences which were rectified in like Snow Leopard). Also in practice if you are broadcasting in Rec709 whenever I see my commercials on TV it's always a cringe fest since it goes through the commercial dupe house and then out to the individual TV stations for each market and then onto a Best Buy calibrated TVs. I find that sRGB displays are actually more reliably calibrated than televisions which are completely all over the place. In my opinion unless you're going to DCI or ACES there is essentially no reason to even have a perfectly calibrated monitor as long as you are in the generally correct colorspace since no two viewers will see it the same anyway. But that's the fatalist in me.