[Openicc] Questions about color pickers and graphics libraries under LINUX

Chris Murphy chris at colorremedies.com
Wed Feb 13 15:58:08 PST 2008


On Feb 13, 2008, at 10:55 AM, Bob Friesenhahn wrote:

> On Tue, 12 Feb 2008, Chris Murphy wrote:
>> better. HDR displays are appearing now. The adoption rate may be slow
>> at first, but when people start seeing HDR capture on these displays,
>> it's really stunning. Scene-referred captures on a scene-referred
>> display!
>
> To me, HDR means that the device/format is able to capture/display/ 
> store values which are much whiter than white and blacker than black.

Eh, yeah I understand that, but that's the language of bits and  
encoding data. But that's not necessarily a clear way to describe it  
in the context of how devices actually work.

First it's not very clear at what dynamic range or contrast ratio  
something is HDR. I don't know if there is a spec somewhere that  
defines these things, but I haven't come across that yet.

But in an ICC context, things are very output-referred centric. Most  
of the workflow assumes things are output-referred, i.e. they are  
rendered to look-good when reproduced on a device with output/print  
contrast ratio (or dynamic range). The ICC spec defines the contrast  
ratio for the perceptual reference medium as being 288:1. That's  
really very high and for most situations the dynamic range of print  
is actually a lot lower than this.

A display that can do 10000:1 contrast ratio in a practical  
environment is maybe not quite HDR, but in more of a gray area, I  
don't know what to call it. Don't know if we need a name for it. But  
perhaps we do since that's about the dynamic range of a single Raw  
capture from a DSLR, unrendered, scene-referred. When we stack even a  
few of these together with perhaps only 1/4 stop differences between  
the first and second, and second and third bracketed shots, then  
composited together, it is called HDR. 32bpc floating point is used  
to encode, although doesn't Microsoft have a 16bpc integer  
implementation for scRGB (which is HDR by definition)?

So I don't think that encoding alone is not the indicator of what is  
or isn't HDR. It's about actual dynamic range.


>   In other words, the range is broader than the human eye can  
> handle at once.

I don't know that this works. Our non-adapting dynamic range is  
greater than some capture devices; if I bracket multiple shots with  
such a camera, and composite them together, that is considered an HDR  
image even if it's only a 1/4 stop bracketing, and I would save the  
file using an HDR format, and I'd most likely be saving it 32bpc  
floating point if I'm using Photoshop.
>
> Perhaps by HDR display you are referring to a wide contrast ratio  
> (not quite the same)?

60:000:1 contrast ratio? That's not just wide contrast ratio. That's  
HDR. Dolby bought Brightside, and have said they'll have a new  
protype in Q1 2008.

5000:1? 10:000:1? Well OK, maybe we need a different term for this,  
it's not LDR by any means as it's misleading to lump that in with the  
vasty majority of images in the world (they're all effectively at  
288:1 at best).

Chris


More information about the openicc mailing list