This content has been marked as final. Show 11 replies
This has been covered here ad nauseam. Please do a forum search.
In a nutshell, the color temperature is a Canon tag that ACR ignores, but the Canon software reads it.
The Photoshop Camera Raw plug-in can read the
white balance settings of some cameras. Leaving the White
Balance pop-up menu set to As Shot uses the cameras
white balance settings. For cameras whose white balance
settings are not recognized by the Photoshop Camera Raw
plug-in, leaving the White Balance pop-up menu set to As
Shot is the same as choosing Auto: the Photoshop Camera
Raw plug-in reads the image data and makes a guess as to
the white balance.
Where could I find this list?
Obviously that quote from the user guide appears to be in conflict with what Thomas Knoll, the creator of both Photoshop and of Adobe Camera Raw wrote in the link provided above:
>"Adobe uses a difference camera profile than Olympus does, so the mapping between neutral grays [and] color temperature numbers is different.
b This is true for every camera model that Adobe Camera Raw supports."
Let's hope someone from the ACR team jumps in to clarify.
I don't see any contradiction here. Using the camera's WB setting does not imply, that the same number has to be displayed by different programs. Regard this as for example using different co-ordinate systems, Cartesian vs. polar vs. spherical system - these are different approaches of describing a point in some space. Different numbers can describe the location of the very same point, depending on the system.
The point is not that the same numbers appears in camera and in ACR, but that the same effect is achieved. If the camera's own raw processor produces an image containing a grey subject, then the result of ACR too has to show that subject as grey.
>but that the same effect is achieved.
Correct and different profiles will necessarily produce different numbers, as Thomas says.
Thanks for your input, G.
Ramon seemed to have the answer in hand, then G Sch above chimed in with some random comment about coordinate systems. Weirdly, Ramon then agreed with G Sch's nonsense and thanked him for it. Suddenly the thread has suffered an ineluctable defenestration.
- camera maker's control for XXXX Kelvin wrong?
- Adobe's control for XXXX Kelvin wrong?
- the use of the designator "K" in these contexts wrong, as it implies physics reference for the measure while the camera and ACR just do their own thing?
By the logic used in this thread, 1/250sec shutter doesn't have anything to do with a time standard, nor does F4 mean an aperture, it's just a coordinate in a locally defined system, la la la. So why bother to even code it in EXIF? What's the point of providing a control in terms of K if K isn't normalized?
The question was answered at Ramon's first post: ACR doesn't read the 5D white bal metadata. The camera K setting is used for in-camera processing and by Canon utilities. But note that the raw data are white bal agnostic but white bal results are subject to a camera profile which may differ between OEM and ACR, and at which point there is room for discrepancy for interpretation of color. Which one is right? I can't say. It's important to realize the results for a given K setting may differ between OEM and ACR because of this. Contrary to what G Sch writes above, the same K setting ought to give the same results if a "Kelvin" setting is to have real meaning, but the seems to be impracticable if the developers don't agree on the characterization of the gear.
>Weirdly, Ramon then agreed with G Sch's nonsense
I did no such thing. :)
I agreed with a single phrase in that post, the one I quoted:
> but that the same effect is achieved.
> the use of the designator "K" in these contexts wrong, as it implies physics reference for the measure while the camera and ACR just do their own thing?
'K' or "degrees Kelvin" is not an accurate measure of the color of light. EG, 6500K is taken from Kelvin's measure of the approximate color of black body radiation -- IE, it is not an accurate measure of the color of light under slightly overcast skies. EG, 6500K is not the same as D65.
Digital cameras need to simplify the Wb numbers without refering to how complex the 'K' number is relative to slight variations of tint. I'm not a color scientist, but refering to one Wb number relative to another is more complex than what a digital camera's menu options, or EXIF, general provides for a user or a raw developer.
That's my take ...
> The question was answered at Ramon's first post:
> ACR doesn't read the 5D white bal metadata.
Actually, ACR does.
> The camera K setting is used for in-camera processing
> and by Canon utilities.
There are no camera K settings. There are scalar multipliers that control the gains required to the R, G, and B camera values to achieve white balance. This is related to the correlated color temperature, but the CCT itself is generally not stored.
> But note that the raw data are white bal agnostic
> results are subject to a camera profile which may
> differ between OEM and ACR.
True. Every raw converter applies its own math given the camera multiplier values to obtain the white balance. Similarly, each raw converter has its own math to compute CCT Kelvin values, assuming they are shown and displayed to the user in the first place. Since each raw converter has separate math used, the resulting displayed Kelvin values may be different.
> Contrary to what G Sch writes above, the same K
> setting ought to give the same results if a "Kelvin"
> setting is to have real meaning
That's assuming the camera profiles applied are the same in all cases. But they're not the same, since each raw converter has its own set of profiles. So the results won't be the same.
>Contrary to what G Sch writes above, the same K setting ought to give the same results if a "Kelvin" setting is to have real meaning, but the seems to be impracticable if the developers don't agree on the characterization of the gear.
A Kelvin temperature will accurately describe the spectral power distribution (SPD) of a black body radiator. For non-black body light sources, correlated color temperature (CCT)is used and for every CCT there are infinite numbers of SPDs. Refer to the illustration on Wikipedia showing lines of constant correlated color temperature. Unfortunately, color temperature does not fully describe the color of illumination--D65 describes a specific SPD while 6500K can have a large number of different SPDs.