Re: RPP raw photo processor 64
Re: RPP raw photo processor 64
- Subject: Re: RPP raw photo processor 64
- From: Ben Goren <email@hidden>
- Date: Sat, 01 Jun 2013 09:12:21 -0700
On Jun 1, 2013, at 8:50 AM, Andre Dumas <email@hidden> wrote:
> So a "scene-referred image" can only be an approximation,
In practice, but not in theory. A digital camera captures the same basic information as a spectroradiometer. But where the spectroradiometer has high spectral resolution and fidelity but low spatial resolution, the digital camera is the opposite. You could couple the exposure settings (aperture / shutter / ISO) of your camera to generate absolute, unscaled XYZ values in the scene, if you so desired.
...but then what?
Your monitor isn't going to reproduce the exact same brightness levels as recorded (nor would you want it to), and your printer certainly isn't.
What you *can* do, and do very well, is map the tonal values in the original scene to the tonal range of your output device.
Assume perfect, even illumination in the scene. Now, further, assume that said illumination is exactly the D50 standard, and that the subject is the proper distance from the illumination for it to be the exact same brightness as defined by the standard. A proper digital photographic exposure and development of this scene is theoretically trivial; all you need to do is ensure that the Lab values represented by the profiled RGB values in the development match the actual Lab values in the scene.
Now, double the brightness of the illuminant. the proper exposure and development can now be obtained with an exposure one stop darker -- twice the shutter speed or half the aperture or the next stop lower ISO setting, or some combination thereof.
Though the original scene is now twice as bright, the resulting digital image is the same as if the light were still at the standard brightness. In other words, objects are still rendered in the digital file with their Lab values matching theoretically-perfect illumination, even though that wasn't the actual illumination.
If illumination isn't even, then you're faced with artistic challenges...but, again, mostly due to the output medium.
Lastly, I'll note that a great deal of practical problems in digital imagery stem from our use of file formats that clip everything brighter than L*=100 to L*=100. The cameras are well capable of recording information about scenes with a much larger tonal gamut than Lab, but none of our working spaces can encode that. So, we wind up with kludges where we scale the brightest element in the scene to the clipping point and let L*=100 get mapped to wherever it lies, frequently even in the L*=80 range. That may well be a practical necessity for printing, but that kind of tone mapping should be done much later in the processing pipeline, ideally as part of the preparation for printing, and not at the time of raw development.
That is, cameras actually do faithfully record scenes with a much greater brightness range than can be reproduced on a piece of paper, but all of our file formats are designed for an idealized piece of paper. So of *course* we're constantly fighting dynamic range -- and we have nobody to blame but ourselves.
Cheers,
b&
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden