Any advantage to shooting 16 bit TIFF?

Started Dec 14, 2011 | Discussions thread
apaflo
Veteran MemberPosts: 3,854
Like?
Re: One marginal advantage
In reply to cinemascope, Dec 17, 2011

cinemascope wrote:

The simple FACT that the TIFF file is bigger means it has MORE data,

It does not have more information though. That is basic to the understanding of data communication and information theory. See "A Mathematical Theory of Communication" by Claude E. Shannon. Specifically what you want to look at is
Shannon's use of the term "entropy".

http://cm.bell-labs.com/cm/ms/what/shannonday/paper.html

you're just not happy to accept that simple fact because YOU don't like the data it contains.

I haven't been able to figure out what you mean by that statement. The information in a TIFF file defines a single very specific image. Liking the data has nothing to do with how much information it contains.

But of course that the RAW file will always be a better source of information anytime, since only it contains the original information that was captured.

The information in the raw sensor data clearly is greater than the information in a TIFF image. That is why is is a better source.

To say the colour is Bayer encoded is just fluff. It still doesn't change the fact that the pixels gather light intensity in a monochrome fashion, with no colour information except for deviations in the light intensity due to the colour filter array. It is still monochrome at the end of the day, if you like it or not.

If it is in fact monochrome just exactly where does the color information in the derived TIFF image come from? You are confusing the issue of what each data byte represents with what and entire encoded data set can contain. Each data byte is indeed a single monochrome sample, but since different data bytes represent different colors, the entire data set contains color information. That is not in any way different than the fact that RGB image data sets also have bytes that represent different colors.

The data in the RAW is 14 bits per pixel, period. That is what is in the RAW file at any point in time. 126 bits from MULTIPLE pixels MIGHT be used LATER to calculate a single RGB pixel. Your "per sensor location" is just an abstraction you created, and it explains well what information can be extracted from the RAW file, but you cannot think that your abstraction is somehow embed in the RAW file. The file still has 14 bits per pixel. What will be done later with those 14 bits is different. If 126 bits will be used later to calculate a single RGB pixel that's something else.

You are confusing each pixel in an image with each location on the sensor. Pixels are parts of an image, and the raw data from a sensor does not contain bytes that relate to a single pixel. Pixel data is derived from multiple sensor locations, and when that is done the result is an image.

RAW files have 14 bits per sensor location. The number of bits used to generate a single pixel varies, but generally the fewest is from a 3x3 matrix of 126 bits. However, most of the pixels in an image are derived from 4x4 or 5x5 matrixes and hence on the average there are vastly more than 126 bits per pixel.

Also there are many ways to interpret the data in a RAW file, and yours "126 bit" approach is just one of them. Look at dcraw for example. It provides one with several different "mathematical" ways to translate a RAW file into an RGB one.

And as I said, the 3x3 matrix is generally the minimum sized matrix used.

A TIFF will have DIFFERENT information, information DERIVED from the original RAW information. Yes, it can have more data, if more information is DERIVED from it. Same happens if I read a company's 4 page financials and write a 20 page report on it.

You can use more words, but if that is the only source you cannot have more information. Entropy is the operative word. And a word of caution too, to consider that it makes no difference is half the words are redundant or just random noise, the effect on entropy is exactly the same: twice as many words to communicate the exact same information. That would also be twice as many data bytes to contain the same information!

Of course data will get scrambled, and that is why you cannot reverse the per pixel RGB information in the TIFF file back into the monochrome information in the RAW file.

You can't reverse it because the TIFF doesn't have enough information to regenerate the RAW file.

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow