And just because you found it on the Internet, it's true?
Cambridgeincolor often uses language very imprecisely. They really should fix that! Wikipedia is well know for inaccuracy. Here is what that article says (analyze this for logic, rather than content):
"Raw image files are sometimes called digital negatives, as they
fulfill the same role as negatives in film photography: that is,
the negative is not directly usable as an image, but has all of
the information needed to create an image."
IOW, it's not an image after all.
jcboy3 wrote:
While there are numerous algorithms for demosaicing the sensor data, that does not change the fact that it is an image.
Invalid logic, but that isn't the point anyway.
There are numerous images that can
correctly be generated from any given RAW file with any given algorithm. They all represent a valid application of that given algorithm. (And that it is not a reversible process, in that none of the images can be used to re-generate the RAW file.)
jcboy3 wrote:
Perhaps you can provide a source for your statement that image data defines just exactly one single image.
Perhaps you can provide a source that says otherwise. Isn't it inherent in every image standard and every definition of "image"? An image is an image, not a collection of images.
Or more to the point, each pixel has exactly one location, one tone level, and one color value. And each pixel is unique, it affects those values for that one location in the image and does not affect any other location. That is, as a set, the pixel data absolutely defines an image. Just one image. Anything that changes the image is either an edit or an error.
Compare that to raw sensor data. Each sensor location value affects multiple pixels of any generated image. None of the values is unique to any pixel. And no set of values for one location actually defines the location, the tone level or the color value for that location in a resulting image generated from the raw data. And when correctly processed there can be many different and each quite correct images produced.
jcboy3 wrote:
On the other hand, if I use Lightroom to view a RAW file, then there is in fact a unique image based upon the demosaicing algorithm used by Lightroom. There is no random event; I will get the same result every time. So in that case, there is exactly one single image associated with that file.
That is not true. Lightroom can show you thousands of distinctly different images, each derived from the exact same raw sensor data, and each image is "correct". You will only get the same result every time if you tell it you want that one single result. That doesn't make the other incorrect, and the others do not require editing the raw sensor data to get a different result.
With an image file there is only one way to display the data correctly. Anything else necessarily means the data is edited. (Not that every display provides exactly the same image, but that is the intension and the differences can be measure and are "errors" in display.)
jcboy3 wrote:
The derivation of an RGB image for display purposes is the same for all image formats. JPG files are compressed and must be converted for viewing. TIFF files can be compressed, and can be higher resolution than supported by displays. So it's all the same, nothing unique about the RAW file. The ONLY difference is that an image is not saved as a RAW file; it's only used for input.
JPEG and other images, not just TIFF images, can also be higher resolution than supported by displays. The image format has nothing to do with resolution.
RAW files are clearly different than image files. The data necessarily must be processed to generate an RGB image. The data does not define just one RGB image. Without editing the data there are many possible RGB images. But once an RGB image is formed, it is just one image and any change requires editing the image data to display something different.
jcboy3 wrote:
As far as WB, well what can I say. The "as shot" WB is a fact, it comes from the RAW file; it is preserved in a DNG file. It is not a "default" value but is set by the camera. Might be constant if the WB mode is set (e.g. "daylight"), but if auto WB is used, or a custom WB is set in the camera, or if WB is adjusted in camera, then the value will be what is used by the camera to create a JPG.
WB data that is in the RAW file has not been applied to the raw sensor data. That particular WB data need not ever be applied! It is used as the default by some RAW converters, but even with those the default can be changed to virtually any valid WB setting.
Whatever WB data is in the RAW file is just non-essential metadata. Your RAW converter can totally ignore it with no consequence.
jcboy3 wrote:
WB is in the data. Experiment. Change settings in your camera; look at the "as shot" WB in your editor, change it in your editor and export the file as DNG and other formats and see what happens when you re-import. It's all there to discover. No amount of dogmatic pronouncement will change this.
No need to experiment, I've looked at the source code to the RAW converter that I use (and a couple others).
The raw sensor data is not affected, ever, by WB. It isn't until the RAW converter generates RGB image data that WB is applied. And if that is then changed, while still working with a RAW converter, the entire RGB data set that had already been adjusted for WB will be discarded, a totally new data set generated from the raw sensor data, and only then will WB be once again applied.
It is, at that stage, never converted from one WB adjustment to another. But if it is saved to a file or otherwise transfered to an editor that does not have access to raw sensor data, any change to WB is necessarily an adjustment to what exists. That has a very different effect too, due to the artifacts created at lower bit levels.
Essentially, you really do not want to change WB of an 8-bit JPEG if there is any way to avoid it. Less important with a 16-bit TIFF, but even then it is best avoided.
jcboy3 wrote:
The difference between WB adjustment for JPG and TIFF can also be easily demonstrated. If I adjust WB and export to TIFF, then when I re-import the image the adjusted WB will be displayed as an "as shot" value. For JPG, however, the WB setting is not preserved and the "as shot" value will show 0 for both temperature and tint. That is clearly a difference, as I was stating.
Clearly an artificial distinction between what your editor shows for one and what it shows for the other, and not a distinction between how WB is applied to either type of file.
At the point when WB is adjusted in an editor the data being worked on is not in either JPEG or TIFF format, but rather in the internal RGB format of the editor. After WB is applied to the RGB data it can be written to a file in either JPEG or TIFF format, which will not change WB.