spectral tonemapping possible?
by Ernest Burden
on
Feb 2, 2008 at 5:41:34 pm
I am looking for a way to apply a tone curve to images in Photoshop using spectral data--output curves expressed in the spectral space between about 400nm and 700nm. Is that possible, if so how? Working in 16 or 32bit is an option, if that helps.
An example of how you might use this is to get the spectral curves of a particular lamp type and bias an image to them, simulating the use of that lamp to produce the picture. The spectral curve will pump up certain wavelengths (colors/saturations) while suppressing others.
When you look at RGB spaces and LAB, it's hard to see how you might accomplish a spectral re-mapping. Any suggestions?
Re: spectral tonemapping possible? by Vincent Rosati on Feb 4, 2008 at 3:04:40 am
Holy crap! Are you a rocket scientist or something? :)
I'm only answering this because it makes me sad to see a thread go unanswered, and it may spur a little friendly competitiveness to get this thing answered.
The question sounds vaguely like adjusting curves - Image/Adjust/Curves. However, I'm not sure how you might input the values you mentioned.
By default, the 'curve' is linear and hold 2 points - one input and one output.
I often use an 'S' curve by adding two points. One at the I/O 192 intersection, and the other at the I/O 63 intersection. I push the '192' up a little, and drop the '63' down a little.
Curve profiles can be saved, and loaded into other files.
Are you trying to capture & save some sort of profile from one image and apply it to another? (aka. spectral tonemapping?)
Could you possibly provide a link to a visual example of what you are trying to achieve?
Can this be an artistic process, or does it require automation for, say, legal reasons.
I think we'd all appreciate a list of your workflow and an example, once this solved.
Re: spectral tonemapping possible? by Ernest Burden on Feb 4, 2008 at 1:15:29 pm
I am a rocket scientist. No, I'm not.
The issue is how to use spectral curves to process an image in Photoshop. I already gave the example of using the light output data from a 'blackbody' emitter (a.k.a. a lamp) to bias an image so it would appear to have been lit with that sort of light. We are all familiar with the color temp. settings (I use 6500K) for monitors, well, that's where it comes from.
Another example is to use PS to fake a photographic technique where IR film (infrared) is exposed without a daylight filter, and a second frame is shot on a standard daylight film, then the second is used to mask the first, leaving just the IR portion. Logic suggests that this would only work with a studio shot because clouds move, trees too, so the frames wouldn't line up all that well. Using Photoshop to remove the visible light portion of the image sounds like the perfect solution.
Workflow? I don't have one. I have tried a few things, looked online for plugins, never found anything that works. But PS is so versatile that I bet I'm missing something.
Re: spectral tonemapping possible? by Darby Edelen on Feb 5, 2008 at 7:19:14 am
What exactly are you trying to do?
If you're just trying to change the white point of an image then your best bet would be to capture your images in Camera RAW and adjust the white point on import:
This process, however, is destructive so make a copy of your image.
I don't think you need to be concerned at all with spectral power distributions. Colors are generally coded in computing as tristimulus values, and these are more than enough for accurate reproduction of colors (and are actually more useful). Unless you're working in color science or another scientific field (in which case Photoshop might not be your best tool) why even worry about such things?
[Ernest Burden]"Another example is to use PS to fake a photographic technique where IR film (infrared) is exposed without a daylight filter, and a second frame is shot on a standard daylight film, then the second is used to mask the first, leaving just the IR portion."
You should definitely be able do this in Photoshop, provided you have a method of capturing the IR wavelengths in the first image necessary to do so... I can't think of anyway to 'fake' additional wavelengths of light accurately in the image if they weren't captured there to begin with.
Darby Edelen Designer Left Coast Digital Santa Cruz, CA
