RedWideGamutRGB: discuss!

[Elsie N]

As someone not steeped in post knowledge, please forgive if I ask simple questions. I'm just trying to understand how RWG will affect me.

For instance, when it's in the camera firmware, will I be able to monitor in RWG and having the look I want, then be able to open up that set of metadata in an NLE (Premiere for instance) and see things how they were shot?

And if that be true, then can I convert that footage into an ACES workflow and still have it look the same?

And further, if I exhibit in the ACES protocol, will it look pretty much at its best without changes, in a REC 709 only capable viewer?

 

[Matthew Riggieri]

I just finished up grading a trailer for a feature film that I am heavily involved with. Because it is a personal project and not commercially, I decided to try out RWG/LG310 behind a custom transform LUT. I will say it was a pleasure to grade and the skin tones and general colors were super accurate. This was shot on Dragon. The depth of color available in RWG is amazing. I need to work on my transform LUT, as blues were a little super saturated, but beyond this, I think I will be working this way with RED footage form now on. The color separation was truly astonishing.

https://vimeo.com/181108134

 

[Marcus Friedlander]

I'm not a regular poster here, so forgive me for butting in but i have a few observations and i'm sure i will be corrected if these aren't, well, correct.

RWG is simply a gamut that is big enough to contain all of Reds camera RGB with the added benefit that Red will calibrate their conversion to RWG so the end results will be a close as possible, based on the capability of the sensor. This includes a log curve to record the range the camera can see too. This is an awesome move to simplify post. It's not unlike what ACES is doing for cameras at large but this is purely within the Red family.

However sticking a generic LUT on to 709 is sort of pointless, especially when people are pixel peeping and scrutinising the results - it's not fair on RWG.

The issue with other gamuts in the past is that the gamut is smaller than the camera could see and therefore at some point there was remapping of out of range colours into this smaller gamut. That's the issue with most colour science and grading - the decisions on how you massage a colour back into gamut can affect the whole image. IMHO this partly why cameras have a 'look', it's the mapping of what the sensor saw into the manufacturers interpretation of what it was.

The 709/sRGB primaries, whilst being the default target for most of us, are really quite small. There's a thing called pointers gamut that was an attempt to map real world colours into a standardised gamut and if you look that up you can see that some of what we see around us is outside the 709 primaries. Even on a macbeth chart.

Red (the colour) is very difficult because 709 doesn't have much tonality for Red. This is one of the reasons why it's so difficult with dSLRs and other cameras to get decent Reds. So it's a good illustration of what i mean. The whole sRGB thing was, AFAIK, intentionally choose to get nice easily saturated images on the internet. But we can capture so much more these days.

The task of getting a full red into a smaller gamut is down to the application and workflow you are using. You can simply clip the red, you can scale whole sections of colour to maintain distances between other colours but not matter what you're doing you are artificially changing the image the camera saw. If you're scaling colours in then there's no set way of doing that and no matter what you do you *will* affect all the colours to some degree.

In addition to this, each gamut has it's set of primaries - how red is that cameras red and so on.

So to swing this back onto topic.

RWG is awesome but the task of creating an image in a smaller colorspace is fundamentally part of what a colourist is doing. The yellow/green/red skintones can be an artefact how how you go about mapping a huge space into a smaller one. And also ensuring that the application you are using is aware of the primaries of the source material. The number of times i've seen someone try to grade Slog without any colour space conversion is shocking. The Slog primaries are quite different from 709 and *need* a conversion. Hence 'Slog2' skin is yellow.

So IMHO i think we should be looking at workflows and manual grading and, for the time being, not looking for generic LUTS unless you need a quick way to get some results.

So I don't think it is correct to say that RWG produces yellow or green skin tones. It's what you do to that source afterwards (unless of course there is a shift when Red turn camera RGB into RWG)

Does that make sense?

cheers
Paul

Great post! I'm not the most knowledgeable post guy in the world, so many someone else will have something to say, but this all sounded logical to me!

Thanks for sharing!

 

[M Most]

RWG is awesome but the task of creating an image in a smaller colorspace is fundamentally part of what a colourist is doing.

Not really. It is part of what a color scientist does. One of the primary functions of properly designed color transforms is gamut mapping, i.e., taking one color space and mapping it into another with a perceptually similar result. The job of the colorist is to manipulate the image within whatever container is chosen as the working space to produce a creatively appropriate result. These two functions are not the same. The first is done by using proper mathematical transforms, and, based on various color theories, come up with a resulting image that displays "properly" on the target medium. The second is done by using artistic approaches to manipulating various balance and range controls, using color and area isolations creatively, and creating an artistically pleasing and creatively appropriate result. The first is done by a mathematician (albeit one well versed in color theory) who is interested primarily in scientific accuracy, and the second is done by a colorist (i.e., an artist) whose primary concern is the storytelling value of the result.

As I said, these are two very different jobs that require two very different approaches.

 

[paulcurtis]

Not really. It is part of what a color scientist does. One of the primary functions of properly designed color transforms is gamut mapping, i.e., taking one color space and mapping it into another with a perceptually similar result. The job of the colorist is to manipulate the image within whatever container is chosen as the working space to produce a creatively appropriate result. These two functions are not the same. The first is done by using proper mathematical transforms, and, based on various color theories, come up with a resulting image that displays "properly" on the target medium. The second is done by using artistic approaches to manipulating various balance and range controls, using color and area isolations creatively, and creating an artistically pleasing and creatively appropriate result. The first is done by a mathematician (albeit one well versed in color theory) who is interested primarily in scientific accuracy, and the second is done by a colorist (i.e., an artist) whose primary concern is the storytelling value of the result.

Yes i understand what you mean, you're totally right. My point is that in the special case of RWG (and reading the posts here which are judging based on 709 results) the colourist will most likely be the one who needs to do this work. Perhaps there's a core grade and a trim for 709. But sometimes that trim is more than just a technical conversion.

There are no perfect mathematical transforms gamut mapping, there's a degree of creativity even through the math. Do you scale perceptively, are you trying to maintain a very specific corporate colour? Is there a precise creative relationship between some colours. It's not always math.

The conversation here about very precise rendering of skin tones is affected by how the RWG is mapped into different space. You can use X number of off the shelf mathematical solutions but perhaps they're not doing the best, in a given situation. They're black boxes and inside you don't know what's happening to the tonality of the source.

As i said you are obviously correct, but in this special case here with people able to access RWG the post workflow is important and it can cause no harm if people understand what's happening to their colours?

cheers
Paul.

 

[Sabovic Adis]

Not really. It is part of what a color scientist does. One of the primary functions of properly designed color transforms is gamut mapping, i.e., taking one color space and mapping it into another with a perceptually similar result. The job of the colorist is to manipulate the image within whatever container is chosen as the working space to produce a creatively appropriate result. These two functions are not the same. The first is done by using proper mathematical transforms, and, based on various color theories, come up with a resulting image that displays "properly" on the target medium. The second is done by using artistic approaches to manipulating various balance and range controls, using color and area isolations creatively, and creating an artistically pleasing and creatively appropriate result. The first is done by a mathematician (albeit one well versed in color theory) who is interested primarily in scientific accuracy, and the second is done by a colorist (i.e., an artist) whose primary concern is the storytelling value of the result.

As I said, these are two very different jobs that require two very different approaches.

Yes i understand what you mean, you're totally right. My point is that in the special case of RWG (and reading the posts here which are judging based on 709 results) the colourist will most likely be the one who needs to do this work. Perhaps there's a core grade and a trim for 709. But sometimes that trim is more than just a technical conversion.

There are no perfect mathematical transforms gamut mapping, there's a degree of creativity even through the math. Do you scale perceptively, are you trying to maintain a very specific corporate colour? Is there a precise creative relationship between some colours. It's not always math.

The conversation here about very precise rendering of skin tones is affected by how the RWG is mapped into different space. You can use X number of off the shelf mathematical solutions but perhaps they're not doing the best, in a given situation. They're black boxes and inside you don't know what's happening to the tonality of the source.

As i said you are obviously correct, but in this special case here with people able to access RWG the post workflow is important and it can cause no harm if people understand what's happening to their colours?

cheers
Paul.

Is there a particular reason not to go with proper R3D to REC.709 trans...eh...form, code...something...the 709/709 (plus sauce not special or secret at all, the Alchemy) in, say, DaVinci? Why the detour to RWG/Log3G10 then looking for the "proper" 709 LUT, if it's gonna end up being 709 anyways?

Sorry, I had to ask, I'm a bit slow. :rofl:

BUMP! :biggrinjester:

 

[Mike Finegan]

Graham / others
Can we know the use case for adopting RED WCG? Is this a stepping stone to BT.2020?
The example that started this was a blue outside the color volume of a particular RED color space / gamut combination - right?
It is easy to say leave the LUT creation to mathematicians, but really that is impractical. All color grading shops have created LUT as their secret sauce, right?

I don't see the motivation to map results back to rec 709 - which is a much smaller color volume and adapted to CRT not UHD WCG devices now available. The 4K UltraBluRay and SUHD tv are all targeting BT2020 (even if less than 100%) I believe. Why not at least pick p3 for post? So much is lost mapping to rec709.
(is it possible to show where that cobalt color that caused the issue lies in CIE chart compared to BT2020, P3, Rec709? Would P3 color space be able to accurately display that color?)

1) which is the best RED color space / gamma to work with for most forward compatibility / BT2020 delivery - say Amazon, Netflix 4K HDR on SUHD TV or UHD Bluray or HDR digital cinema?
2) can't RED provide a basic LUT since they created this color volume and will be most familiar with its strengths and weaknesses? No warranty, no guarantees, but if every user has to create / buy custom LUT to use - that is not so helpful

 

[PatrickFaith]

This is the old Red One MX with REDWideGamutRGB/Log3G10 with Goran's LUT. Only processed minimally in Redcine-X, no sharpening. Contax 85mm exposure F2.8. No IR filter

RED ONE really needs to be in a camera hall of fame, is just a amazing achievement that shows the power of RAW and the continuous improvements to color science.

 

[M Most]

Graham / others

It is easy to say leave the LUT creation to mathematicians, but really that is impractical. All color grading shops have created LUT as their secret sauce, right?

I didn't say that. I said that gamut mapping was a mathematical transform that is not done by a colorist, it is done by a color scientist.

Would P3 color space be able to accurately display that color?)

P3 s a specific color space that was designed for and defined by a specific display technology (in this case, DLP projection). It should not be looked at as some kind of generalized, "better" color space than Rec709 or anything else. Wide gamut color spaces are designed to exceed the limits of current and near future display technologies so as to accommodate all of them. That is the reasoning behind the definition of ACES AP0 as well - although Graeme will probably tell you that a good part of his motivation for creating the RWG space was to "fix" some of the issues with certain values creating clipping in the ACES transforms.... :wink5:

1) which is the best RED color space / gamma to work with for most forward compatibility / BT2020 delivery - say Amazon, Netflix 4K HDR on SUHD TV or UHD Bluray or HDR digital cinema?

None of them. You pick the color space you're working in and the color space you're viewing in based on the viewing device and the delivery requirements. You don't do that expecting that some other device is going to come along that the grades you just did will "look good" in. New color spaces require new grading, whether that's accomplished by a base transform to accomplish the basic gamut mapping and white point accommodation, or simply a re-grade in the new space using displays that are appropriate for that space. That's why you either hold on to the camera original files, or transform into a very wide gamut space (like ACES) that will likely be suitable for future displays.

2) can't RED provide a basic LUT since they created this color volume and will be most familiar with its strengths and weaknesses? No warranty, no guarantees, but if every user has to create / buy custom LUT to use - that is not so helpful

Patience. Have patience.

 

[luigivaltulini]

This is the old Red One MX with REDWideGamutRGB/Log3G10 with Goran's LUT. Only processed minimally in Redcine-X, no sharpening. Contax 85mm exposure F2.8. No IR filter

Here is iPhone reference shot.

hehe Steve, the Red one still produces beautiful images, there is no 'doubt

 

[paulcurtis]

Graham / others
Can we know the use case for adopting RED WCG? Is this a stepping stone to BT.2020?
The example that started this was a blue outside the color volume of a particular RED color space / gamut combination - right?
It is easy to say leave the LUT creation to mathematicians, but really that is impractical. All color grading shops have created LUT as their secret sauce, right?

Ideally we try to work in the widest space we should and then deliver to different targets. If you *know* that you're only going to be viewed on 2020 then you can cut corners, but i'm not sure how many people can really claim that's the case.

That's the basis of ACES. Work in ACES and use a different ODT to map to lesser gamuts. (But i'm still saying IMHO that there's an element of creative work involved in targeting a particular destination even if you are going to slap a mathematical transform on the end). With HDR and a multitude of standards i don't think anyone should realistically expect no tweaking for a given destination for example.

The new iOS supports extended RGB which uses >1 floating point values to create a wider internal colorspace. Why is this quite cool? Well the iPadPro and 7 displays are factory calibrated to a wider space (near P3) and you hold an old iPad running iOS 10 and an iPad Pro and view a full range image and visually see the difference. These devices are pretty awesome previewing 'sanity checks'. I can't think of any other consumer devices where it's so easy to see a wider colour space like this.

But here's the thing, what are you using in your workflow? If you're using Premiere then it's internal primaries are sRGB. Even when using lumetri, the image is transferred from sRGB 2.2 (i think) gamma to linear, colour work done and then transferred back to 2.2. This isn't the most efficient way of doing this and hopefully one day this will evolve, nirvana would be colorspace management in there. But it means with the best will in the world you're going to struggle to manage colorspaces in that application. I have no idea how viewing R3Ds is working. I know that for DNGs there are a number of thorny issues. It's an issue because a lot of what goes on behind the scenes is automagical and quite difficult to understand. Hence the baselights of the world which offer total control. (I'm a Nuke guy, although used Baselight Editions, most of my colour work though is manual in Nuke)

Then an argument is why not just use a massive colorspace? The reason is that the math can fall down - basic grading operations work differently if you have an unbound colorspace. Especially CGI work. See http://nbviewer.jupyter.org/gist/sagland/3c791e79353673fd24fa for examples. This is a good extreme demo of how the spaces affect basic operations too http://ninedegreesbelow.com/photography/unbounded-srgb-as-universal-working-space.html

As Mike said even ACES have toned back the colorspaces they're using. And i presume this is the motivation for Graeme and Red - to create the ideal container for the cameras.

Not wanted to get away from the topic. My point is that yes you have this wonderfully ideal colorspace. But the applications you are using can affect what you can do, so when examples are posted they ought to be complete with some info about the workflow - in case that got in the way and changed the results subtly. And at the end of the day if you're viewing this in a browser on 90% of the devices out there then you're viewing in sRGB space. So when the image was created was there a final mapping to sRGB done - either manually or by whatever export was done to get the image. My workstations working space is 709 and i'm always being burnt when sending things off or uploading to Facebook or whatever because i forget to translate to sRGB in photoshops export... Super easy to do (or maybe i'm just a bit useless at it)

cheers
Paul

 

[rand thompson]

Hello, First post here. I tried the RED RWG colorspace, but instead of using the LOG3G10 I used the LOG3G12. Was that wrong?

[/URL][/IMG]

 

[Stacey Spears]

Log3G10 is the successor to 3G12. Nothing wrong with 3G12.

 

[rand thompson]

Log3G10 is the successor to 3G12. Nothing wrong with 3G12.

Thank You.

 

[Marc Wielage]

Hello, First post here. I tried the RED RWG colorspace, but instead of using the LOG3G10 I used the LOG3G12. Was that wrong?

I would love to see the XRite color chart full-frame and in focus. That's generally how I ask for color charts to be shot, when I'm in a position to ask. It's useful to see what effect specific color settings, LUTs, and adjustments have on a known chart when you can look at monitors and scopes.

 

[Steve Sherrick]

I would love to see the XRite color chart full-frame and in focus. That's generally how I ask for color charts to be shot, when I'm in a position to ask. It's useful to see what effect specific color settings, LUTs, and adjustments have on a known chart when you can look at monitors and scopes.

Mark, here's a chart. Sorry, not in perfect focus, not full frame, and not current crop of cameras. But here's an R3D nonetheless and I'll see what I have on drives for more recent cameras. Goran's LUT is in there too. The official RED one is not available yet.
https://www.dropbox.com/sh/aoclvb7vs9py2e1/AADgxC-OoUldSP_E3rCVfGzka?dl=0

 

[rand thompson]

I would love to see the XRite color chart full-frame and in focus. That's generally how I ask for color charts to be shot, when I'm in a position to ask. It's useful to see what effect specific color settings, LUTs, and adjustments have on a known chart when you can look at monitors and scopes.

I use this exact same chart to manually color correct all of my gh4 camera footage. However, this was sample footage from RED's site. I just now need to get a good color calibrated monitor to better judge color with.

http://www.red.com/sample-r3d-files

 

[rand thompson]

I use this exact same chart to manually color correct all of my gh4 camera footage. However, this was sample footage from RED's site. I just now need to get a good color calibrated monitor to better judge color with.

http://www.red.com/sample-r3d-files

I don't know if this helps , but here's a crop of the Xrite chart from the photo above. Sorry for the quality.

[/URL][/IMG]

 

[Stacey Spears]

I don't know if this helps , but here's a crop of the Xrite chart from the photo above. Sorry for the quality.

Technically that is a DataColor chart, not X-Rite. As long as one has the known RGB values for the chips, it is all good. I really like that particular DC chart as it comes in a hard case that folds up to protect itself. Also like the cube that sits on top to capture the black hole and specular highlights.

 

[rand thompson]

Technically that is a DataColor chart, not X-rite. As long as one has the known RGB values for the chips, it is all good. I really like that particular DC chart as it comes in a hard case that folds up to protect itself. Also like the cube that sits on top to capture the black hole and specular highlights.[/COLOR]

I think I misunderstood Marc's post. The way I understood it was that he usually asks for full frame shoots of Xrite color charts , however if I could provide a closeup view of the DataColor chart in the grade above, That would be helpful also.