OpenRAW

You have a good point regarding the use of flat files containing ASCII data, rather than 'binary' files.

Currently the method to use might be XML, which is a bit more 'open' than DNG.

Regards

Alan

(Software Devloper)

Alan, as a software developer you understand the concept of a top-down decomposition of elements.

You start from the total image, then decompose into blocks, then further decompose, until you get to the valuable data such as sensor data, metadata, etc. (Rather like talking about the "document tree" of a web page, where the visible material is at the "leafs").

"XML" and "DNG" are at different levels of that decomposition. They can't be compared with one another, anymore than you could compare an HTML document with HTML text, or a procedure with a label.

DNG is the top-most level. It only means something when talking about the complete image. It decomposes into tags, because it is based on TIFF 6.0, and can even be compatible with TIFF/EP. TIFF tags are very open indeed. TIFF 6.0 has been around for 12 years or more, and is one of the most widely-used image formats around. TIFF/EP is an ISO standard.

Within those tags the contents can have other syntaxes. Some of these could be based on XML. So XML can be thought of as a possible language for marking up the content of some of the contenst of tags within a TIFF 6.0, or TIFF/EF, or DNG, file. In fact, parts of DNG use XMP for metadata. XMP is XML compatible, I understand.

Could you build a complete Raw image file using XML from the top down? Possibly, I don't know, although I assume you could. But all important cases I know start with TIFF (or TIFF-like) fields at the outer-most level. Most or all of the camera Raw formats do so, because they tend to be based at least loosely on TIFF/EP. Obviously the TIFF 6.0 format that you can set a camera to, or that you might use as input and/or output for a photo-editor, is so based. XMP or XML appears at a lower level within that tagged structure.

The tagged structure of DNG is based on an ISO standard, and is completely open. DNG then defines rules and particular tags, and that then needs a specification. If you started with XML as the mark-up language for the content, you would still have to assign meanings to the fields until you have described the whole file, and that would need a specification. The difference is that, unlike DNG, it wouldn't be similar to any other file format.

As far as I know, DNG doesn't itself define how to hold sensor data, which is the obvious example of binary data. It simply uses the TIFF/EP standard for this, which is also what typical Raw formats do. I'm not sure there is an existing ASCII data way of holding sensor data, although it would obviously be easy to define one. I'm not sure the size implications would be. It sounds like a bigger, slower, format.

It is NOT true that "All trough the years of computers' existence, ONLY one kind of format remains; the classical FLAT FILE, nothing more than a text file with ascii data in it". Flat files with text may well have a long life, but it hasn't always been ASCII. (I worked a lot with EBCDIC). And the instant you depart from natural-language text, you are imposing rules that may have a short life. (How many people know the EGTRAN language nowadays? My first programming language).

Perhaps an XML-based language would be more "elegant". No standards body would use that, instead of a TIFF-like standard, because of the start-up and conversion cost.