Comment by jacobolus | Hacker News Reader

Comment by jacobolus | Hacker News Reader

Parent Full thread

How much software fully supports the 10-bit Eizo or NEC displays on Windows these days? Do typical GPUs now support 10-bit color, or is it still only expensive workstation GPUs?

My impression was that only a few niche applications have bothered (notably, Photoshop has had support for a few years), since 10-bit displays are so rare/expensive, but I’m not an expert.

As for Linux, as far as I know the stable version of GIMP still only supports 8 bits/channel color, so displaying at higher bit depth is right out. I guess maybe some specialized video or 3d rendering software could use a 10-bit display? I doubt typical Linux apps support it.

* * *

Edit: In doing some web sleuthing, there seem to be endless issues with compatibility. (As you’d expect from a new technology that requires upgrading every part of the pipeline for support: applications, operating system, video card drivers, video cards, display interface, and the displays themselves)

I see this on a random web forum: “I got an answer from an engineer at Adobe, why Photoshop CC cannot display 10bit/color under Windows 8, 8.1 and 10: There is no OpenGL-path with 10bit/channel like in Windows 7 any more :(”

Adobe’s troubleshooting site says “Note: 30-bit display is not functioning correctly with current drivers. We are working to address this issue as soon as possible”

Linux is a first class platform for many of the atypical apps that would benefit from higher bit depths.

I’m seeing a lot of web discussions like this: https://devtalk.nvidia.com/default/topic/771081/30-bit-depth...

Can you find an example of someone talking about using 10 bits/channel color displays on Linux in practice, and having it behave properly? What kind of work are they using it for? (Maybe my google-fu is deficient, but all I can find are people complaining about compatibility problems and Nvidia marketing materials.)

In the last few months, I've used 30-bit color on Gentoo in the context of microscopy and data visualization.

Enabling it with a modern NVidia card is dead easy. Basically, in xorg.conf, replace occurrences of 24 with 30. However, not much makes use of the extra colors. Although Qt5 will correctly output 24-bit color in 30-bit mode, Krita is the only Linux application of any kind I found that can actually display additional colors. It's easy to test - show a full screen gradient.

No OpenGL compositors (eg kwin, compiz) support 30-bit color to any extent. Total fail, here. Tearing like it's 1998.

Notably, with Qt5, it does work to make 10-10-10-2 and 10-10-10-0 GL contexts and render to them with your own old-school GL calls, nv_path_drawing, and compat & core profile shaders. However, you do have to manage the GL context yourself - QOpenGLWidget has no 10-bit suppport. You can still use Qt's GL abstractions, just not QOpenGLWidget. Easiest approach is to use QWindow in a widget container (created with that QWidget.createContainersomethingorother static member function), as was commonly done with Qt 5.3 (ie, before QOpenGLWidget).

Back in 2002 I worked on some plugins for a color grading system that supported 4K.

Internally it processed color in 12 bits per channel and I think it supported 10bit color output.

This was running Linux with NVidia GPUs and was totally amazing to see 4K footage being graded in realtime.

That system is now called Baselight and AFAIK it is one of the best: http://www.filmlight.ltd.uk/products/baselight/overview_bl.p...

Compositing and grading for VFX/film. Most high-end VFX studios use Linux.

With what software?

Davinci Resolve is what Company 3 uses (The Martian, Spectre, Jurassic World, etc). It's available on Mac and Linux, but the big shops use the Linux version as they can expand it allowing for more real time nodes.

Baselight is another major platform for color grading that runs on Linux.

I worked for a company that performed stereoscopic conversion for Hollywood films. Their primary tool was NUKE [1] running on Linux. It's also available for Mac and Windows, but because NUKE encompassed their entire workflow, the underlying OS wasn't especially important.

[1] http://www.thefoundry.co.uk/products/nuke/

Resolve, Nuke, Flame, etc

If there's one problem this thread makes me want to solve it's the undiscoverability of the unknown. 99.9+% of software is stuff most of us have never heard of, but we don't know what to search for because we don't know how to know what we don't already know. 99% of business opportunities we might have are invisible to us because we don't know how or where to look. 99+% of the dating population is out of reach of each other because of path dependence problems -- their lives simply won't ever intersect.

If I had my own startup accelerator, tonight (and only tonight ;-) ) this would be my RFS#1. Solve the knowledge bootstrapping problem. Google clearly failed to link 30-bit to VFX.

I've been thinking about this problem on and off for several years (including a startup with an alpha version a couple of years ago).

It's especially important in learning.

How do you organize and present information in a way that finding things most important/natural/easy/interesting to you isn't limited by dissipation of such information through old-fashioned network effects.

No matter what, I think you must first build a knowledge graph. One example that was recently posted to HN is https://news.ycombinator.com/item?id=10468732. Maybe it could be crowdsourced or generated by mining Wikipedia.

Then you need some way for people to find out where they are on that knowledge graph, and a recommendation engine for where to go next. I also think it would be useful to mix in a few random topic nodes that are some number of hops away from a person's current knowledge, plus occasionally a completely random topic of the day.

But that's just the most trivial beginning to an enormous problem.

I'm glad there are others who think the same way! Like always, probably more topic-specific solutions will lead the way to a hopefully universal collection of knowledge.

One way to think of it is as a very early precursor to the civilization bootstrapping kits included on starships in Vernor Vinge's Zones of Thought novels (particularly The Children of the Sky).

More or less what I thought reading down this thread. 'New' tech in the consumer market is tech from 10+ years in specialized trades (4K in 2002 from an earlier comment, touchsreen on your Win8 box? Had touchscreen on the retail POS 10 years ago[1]).

So what is around now that I don't want to wait 10 years for?

[1]cslight differences may occur

Linux workstations have been the preferred platform for high-end finishing applications like NUKE, Flame and Baselight for years.

Generally it's NVIDIA hardware on Red Hat.

Krita, a KDE project competing with GIMP, supports a lot more colorspaces than GIMP. Also, next to all KDE apps support it.

8 bit? 16 bit? 10 bit? CMYK? YPbPr? XYZ? Krita has them all.

A funny note: During Longhorn (PDC 2003), some MS devs indicated that the color pipeline for everything in Windows was moving to a 128-bit pixel struct. Some floating point part, I think. Course that never shipped but it sounds cool.

I think Carmack mentioned that in some extreme-but-possible cases, 64-bit isn't enough to represent the results correctly.

I have to assume that's 128-bit precision in 3D rendering -- partly because it's Carmack, but largely because 64-bit color would be... well beyond human perceptual limits.

This StackExchange post claims ~2.4 million colors for the limit of human color vision; it cites CIE color science: http://photo.stackexchange.com/questions/10208/how-many-colo.... 64 bit per channel color would produce ~2e19 different colors; if it's 64 bits per channel it would produce some even-more absurd number like ~8e57. (Besides the ridiculous numbers, I assume those numbers are per-channel, because it's not divisible by 3.)

And yes, higher bit precision is really important for drawing 3D graphics. Look at the price premium you pay for Nvidia's Quadro cards, which can process in higher bit precision than your stock cards... they're typically 5-10x a consumer card: http://www.nvidia.com/object/quadro.html

Mostly advertised for high end CAD or VFX, where bit precision errors can result in odd display artifacts or incorrect rendering.

I think the context was that after doing multiple layers of texture mapping on a 3D object, a 128-bit color space wasn't enough to be fully accurate. I'll see if I can find what I'm most likely miss misremembering.

Thanks for the SE link. 2.4M can't be right or we wouldn't be able to see banding all over the place right? The link says including luminosity, it might be up to 100M. So 24 bit wouldn't cut it but 30 would?

Like all things perceptual, an exact answer is maddeningly difficult to pin down (individual to individual variation, subjective issues like ambient light, etc).

Even at 10-bit you could in theory see banding if your color discrimination were good enough, I think. It depends on the material... but the 10-bit displays I've seen look awfully, awfully good. (And the ability to represent more dynamic range starts to matter -- you have to have a wide scale to show off the bit depth, too. A monitor I saw at NAB this year has two backlights and can show off a lot more light... campfires glowed, headlights looked like they were bright, very real... that's the real direction for displays, I think.)

Probably 4 32-bit floats per pixel for ARGB, it's what most GPUs use internally for shader computations since DirectX 10.

Reply on news.ycombinator.com