Eric wrote:
[snip]
You miss my point Barry, if a manfuactere isn't doign their image
pipeline similary to adobes, then adding those extra fields is more
work, and also might not yield the same quality their own methods
do. If in their eyes it doesn't yield the same quality, they
aren't going to do it. You and I debating doesn't change the fact
that they haven't adopted yet, and IMHO this is one of the major
reason they haven't.
I don't believe that DNG would make a significant difference to the pipeline. Part of the change to DNG would be the wrapping, (which you appear to accept), and the rest is largely constant data. After all, that extra data is equivalent to the camera-model data that raw converters add to their products to support a new camera model.
The most performace critical part of the pipeline is surely the generation and storage of the sensor data, and that should be similar.
I believe the main reason that most camera manufacturers haven't adopted DNG yet is "not invented here", probably combined with the learning curve which would probably make their first camera to use DNG cost a bit extra to develop. (Nikon may have an extra motive - use of DNG would inhibit their restrictive practices).
[snip]
Exactly, some mfgrs beleive their color profiles/pipelines are best
and right now ACR doesn't fit the bill.....Even with the recent
addition of some "optional" tags realting to alternate pipelines,
they really can't be used as the DNG spec says they can be ignored
by 3rd parties, and most liklye will by ACR as well.
Silkypix has its own profile data, but when provided with a DNG file offers an option of which to use. ACR only uses the DNG version in that case, because it is actually the same as its own anyway. Rawshooter uses its own and ignores the DNG version, so, in effect, it treats DNG as your "Tier 1" anway. All of these are options, with differing merit.
Having the big 2 (or 3 depending on your bent), embedding/writing
meta data, settings and previews in a common fashion would be a lot
more useful then then nothing.
[snip]
There is discussion in the Pixmantec forums about support for XMP metadata within DNG files. The users there realise that it is a separate development from support of the profile data in a DNG file - which it is, of course.
If Bibble wanted to treat a DNG file simply as a different way of wrapping the data, but continue to use its own profile, allowing users to adopt a DNG-based workflow yet still use Bibble, what would be the problem?
--
DNG is better than sliced bread;
DNG images don't become toast.