House of Cards TV Series Colorist Color Space Decisions

While watching the House of Cards TV series in my living room on an HDTV through Netflix, I noticed how different the color space is compared to my client's monitor in the studio using a profiles calibrated to sRGB for the web In a 32 inches Sony HDTV.

Of course, on the HDTV in the living room, the color are more saturated than in the client monitor and the Apple laptop.

However, my dilemma is: since the series has been made with the Internet distribution through computers and HDTVs in mind, what are the color space criteria chosen by the colorist to accommodate these two different type of clientele devices?

They will stick to a reference (rec709) because they can't control the viewing conditions. Listen to this podcast for info on this subject. https://soundcloud.com/mixinglight/mail-bag-ep1-part2

They will stick to a reference (rec709) because they can't control the viewing conditions. Listen to this podcast for info on this subject. https://soundcloud.com/mixinglight/mail-bag-ep1-part2

Tero, thanks for your response.

I listened the podcast. At the end they are suggesting to check the color grading in an iPod, and the suggestion
is repeated in a fallowing podcasts they have. They agree that the iPod is the viewing device closed to the Rc709. Which means they are discarding color grading in sRGB for the web. This is a revelation.

I would like to hear the version of other full-time colorists in this forum about this subject, because it's hard to believe that postproduction houses like the one where Walter work in LA are spending US$17,000 in a postproduction monitor just to match the iPod when the delivery color space is the web. Or is it?

It doesn't matter if you do stuff for the web or TV or cinema. If you use a proper calibrated monitor you and the client can be sure that the footage is dealt with properly. Also the more expensive monitors have features that your normal monitors don't/won't have. If you're calibrating with some off the shelf ******* TV it will only look good on that.

It doesn't matter if you do stuff for the web or TV or cinema. If you use a proper calibrated monitor you and the client can be sure that the footage is dealt with properly. Also the more expensive monitors have features that your normal monitors don't/won't have. If you're calibrating with some off the shelf ******* TV it will only look good on that.

I think within this conversation we are mature enough to understand that a calibrated monitor is a must and that Matching the delivery color space is another criteria we should add, for example from input in Davinci Resolve Color Management to the final intended output. Then, it does matter what is the delivery color space, no just to have in general a calibrated monitor. Calibrated to what; sRGB, Rec.709, PCI-P3, etc?

The question is how to split the differences between an HDTV display (Rec709) and the Web (sRGB) as the colorists in House of Cards have to deal with?

Your monitor/projector should be calibrated to what you're doing. If you're doing TV it will be rec709. Rec709 and sRGB are, for all intents and purposes, pretty much identical.

Your monitor/projector should be calibrated to what you're doing. If you're doing TV it will be rec709. Rec709 and sRGB are, for all intents and purposes, pretty much identical.

Thank you for all your responses to the subject.

Rec709 and sRGB are, for all intents and purposes, pretty much identical.

There will be more variation among individual displays then there is difference between these two colorspaces.

jPo

Your monitor/projector should be calibrated to what you're doing. If you're doing TV it will be rec709. Rec709 and sRGB are, for all intents and purposes, pretty much identical.

Not sure I completely agree with this assertion, at least from a 'deliverables' point-of-view.

Legal Range video levels (Rec709 / 16-235) are substantially different from Full Range video levels (Data / 0-255). And watching content for one particular range, on a display device that is operating in a different range, is readily apparent to even an untrained eye.

In the past, you had a better chance knowing the kind of device your content was going to be consumed on (computing platform versus consumer broadcast platform). Now, as the OP noted, the convergence of content delivery mediums has made that task more difficult -- if not impossible.

The question is how to split the differences between an HDTV display (Rec709) and the Web (sRGB) as the colorists in House of Cards have to deal with?

I don't think the differences are big enough to worry about, because the variability of sRGB computer displays is so vast anyway, and there's no real standard being adhered to (at least for out-of-the-box displays). Normal Rec709 stuff looks OK to me on my own computer displays (compared to my calibrated Rec709 monitor).

I went through this around 2001-2004 when Sony Pictures was struggling to promote all their theatrical features on the web. We did roughly a hundred tests using different kinds of color and gamma spaces trying to "anticipate" the losses inherent on YouTube and other services (including Sony's own websites). After four or five months, everybody pretty much threw in the towel and said, "let's just make the trailers look as good as they can for broadcast TV, and just assume it'll translate OK for home computers."

You could make an argument on 2.2 gamma vs. 2.4 gamma, but it's not a massive change. My general inclination for broadcast anyway is to lean 1% or 2% brighter anyway, on the belief that it's better if the overall picture is slightly brighter than slightly darker. But that's my preference. Ultimately, it's the client's decision.

i use the standard bt1886 (rec 709 with 2.4 Gamma) for color, is difficult to argue against a standard. then two export one at bt1886/2.4 for Broadcast and one at sRGB 2.2 gamma, just because.

Legal Range video levels (Rec709 / 16-235) are substantially different from Full Range video levels (Data / 0-255). And watching content for one particular range, on a display device that is operating in a different range, is readily apparent to even an untrained eye.

Video levels and data levels should look identical if they are interpreted/scaled correctly. Check this article out: http://www.lightillusion.com/data_tv_levels.html

Legal Range video levels (Rec709 / 16-235) are substantially different from Full Range video levels (Data / 0-255). And watching content for one particular range, on a display device that is operating in a different range, is readily apparent to even an untrained eye.

Video levels and data levels should look identical if they are interpreted/scaled correctly. Check this article out: http://www.lightillusion.com/data_tv_levels.html

yep, if something is "different to untrained eyes" there is s mistake somewhere....

In a related subject, I have found an interview with the Lead Colorist of House of Cards, Season 2, Laura Jans-Fazio ; where she explains the color grading workflow of the show and the future of choice for color grading. An interesting reading:
https://library.creativecow.net/wilson_ ... oduction/1

Legal Range video levels (Rec709 / 16-235) are substantially different from Full Range video levels (Data / 0-255). And watching content for one particular range, on a display device that is operating in a different range, is readily apparent to even an untrained eye.

Video levels and data levels should look identical if they are interpreted/scaled correctly. Check this article out: http://www.lightillusion.com/data_tv_levels.html

And that's the way how it should be. Anything else=bugs in software/workfow etc!

i use the standard bt1886 (rec 709 with 2.4 Gamma) for color, is difficult to argue against a standard. then two export one at bt1886/2.4 for Broadcast and one at sRGB 2.2 gamma, just because.

But as most TV's are closer to 2.2 than 2.4, do they compensate for that in broadcast or just run it as 2.4?

But as most TV's are closer to 2.2 than 2.4, do they compensate for that in broadcast or just run it as 2.4?

For traditional broadcast: No.

The "new black" is 2.4 and that is what we grade for. If a client has a need to generate the 5-ring circus of every version and display format known to man, then fine, as long as that is a billable, that is a fair request, although generally a waste of time. Time = money. Einstein proved that.

The reason 2.2 exists at all is that that is what the natural transfer characteristic is for legacy phosphor CRT displays. Since those are functionally no longer relevant, it is now up to the display user to get it right. Most of the time a newer "smart" TV should have this as a default setting, but the reality is that consumers are now in the same knowledge gap as untrained or inexperienced software users who are not now or ever have been aware that multiple standards are in deployment and it is up to them to figure it out for themselves, even if it means diving into the "Advanced" menu, whether its on a home TV or in the "Colorspace Management" preferences of an application.

Strict 0-100 levels as they correspond to RGB / Y'CbCr digitized range values (0-1023 / 64-940) don't really have anything to do with the gamut expressed as a colorspace. If it was that easy, life might have a chance of being pleasurable.

jPo

i use the standard bt1886 (rec 709 with 2.4 Gamma) for color, is difficult to argue against a standard. then two export one at bt1886/2.4 for Broadcast and one at sRGB 2.2 gamma, just because.

But as most TV's are closer to 2.2 than 2.4, do they compensate for that in broadcast or just run it as 2.4?

the standard is 2.4, regardless of the nubi you found out there.

if you take a guess, anybody can ask to redo your work for free because it does not look good in their monitors, if you are on the money with the standard, they will have to pay you if they want changes....

if you take a guess, anybody can ask to redo your work for free because it does not look good in their monitors, if you are on the money with the standard, they will have to pay you if they want changes....

There's a famous story about a major A-list director and a mega-billion-dollar movie where the facility wound up setting up about a dozen consumer-grade monitors in a room so that the director could evaluate what his movie would look like on an "average" display. They referred to this room as "Best Buy." Trying to predict how things are going to change on different monitors is an endless descent into misery and pain.

In this case, they did sorta/kinda calibrate every monitor to Rec709 (or at least as much as the settings would allow), so it was ballparkish. But it's still the wild west out there, and getting worse.

It is my understanding that Netflicks does several encodes and it can determine what device it is being run on and will deliver the proper version. I don't know if this is strictly to lower bandwidth for the appropriate device or if color space is also factored in

It is my understanding that Netflicks does several encodes and it can determine what device it is being run on and will deliver the proper version. I don't know if this is strictly to lower bandwidth for the appropriate device or if color space is also factored in

BWAHSHAHAHA HG A...I'm choking

test quiz: i have a perfectly calibrated grade 1 monitor and Netflix send me the image. looks grexactly!

then i have the absolutely identical monitor but i proudly turn the saturation to 0.

what Netflix will do then?

It is my understanding that Netflicks does several encodes and it can determine what device it is being run on and will deliver the proper version.

Yes, I have been told they have up to 50 or 60 versions of every feature film (literally), encoded for different bitrates, potential monitor sizes, and so on. If you watch a movie on an iPhone over LTE, it's not going to be the same data stream you get if you watch the movie over a 300Mbps connection on a 5-foot display. I had always assumed they were just encoding on the fly, but nope, they maintain all these versions separately on massive servers.

But color depth and gamma space and all that stuff... I think it's all the same. I have sometimes seen "crushed" versions of movies on Netflix, Amazon, and iTunes, and I just assumed that somebody screwed up the encoding somewhere. About 90% of the time, it looks fine on a normal Rec709 display. I think the last streaming project I saw was 11/22/63 on Hulu, and it looked fine.

I'll again say that Sony Pictures had a lot of high-level meetings about this 10 years ago, and they definitely considered making versions of trailers encoded especially for YouTube, some especially for Vimeo, some for Hulu, some for iTunes, etc., and ultimately they just decided it was such a crapshoot they gave up. At least, that was what I was told then. Nowadays, I dunno. You could compare a 1-minute test in Resolve, then optimize one for Rec709 delivery and optimize a second one for sRGB, and see what looks best for you.

But as I've said before, I walk into the Apple Store at the Grove in LA and see 15 Macs on display, and every one of them has a totally different picture. If even Apple can't get it right at an enormous store like this, who will? If you buy a dozen brand-new iPads and put them side-by-side, I guarantee you, they'll all look different, too.

In an effort to try to get some consistency myself, I uploaded a 30 second 1920x1080 25P test pattern to Vimeo yesterday which I had initially corrected in Resolve at Rec.709 with gamma 2.4. I next rendered it out to a 16bit Tiff Image Sequence Master before uploading the Sequence to Vimeo via my NLE which has a 'special' direct Vimeo upload connection, said to be ideally optimized for Vimeo streaming.


Here is the result which looks fine on all my screens here including our Sony 55" X9005B 4K TV which is factory set to gamma 2.4.

In an effort to try to get some consistency myself, I uploaded a 30 second 1920x1080 25P test pattern to Vimeo yesterday...

are you sure, you do not violate intellectual properties of light illusion by using this test charts?
but it's nice, that you want to help other users...

i prefer to use the standard SMTPE color bar and simple gradient generators for this kind of tests. if you are able to read waveform and vectorscope measurements, it's very easy to check and correct the levels using these test patterns.

It's no that new but it's still related to this subject. It seems to be that Apple has decided to change the color gamut in the new iPad. A quote from Apple site:

"A color standard big enough for Hollywood.
The 9.7-inch iPad Pro display uses the same color space as the digital cinema industry. This wider color gamut gives iPad Pro up to 25 percent greater color saturation than previous iPad models. So colors are more vivid, true to life, and engaging."

Then, another new consumer device to match.

...are you sure, you do not violate intellectual properties of light illusion by using this test charts?

Steve Shaw recommends their use and LightIllusion are credited in the text but if there's an issue, I'll happily remove the file...

This is an interesting MixingLight.com podcast that relates to this thread:

MixingLight MailBag Episode 30 Part 3: How Do You Handle Color/Gamma With Web Deliverables?

I find Walter's approach works nicely; grading in 2.4 then rendering a 2.2 version for web.

As most people have already said if it's done on a good setup at 2.4 then it really should still look good anyway on a properly setup computer display, but I find export a 2.2 version helps retain some of the original contrast. I only do this if it's me that's going to be encoding and doing the final web upload, if I'm handing over something that is destined to go to a film festival, web upload, maybe tv, then delivering the standard 2.4 is safest.