1. Is this practical? I don't know ennough about the low level
aspects of RAW data to know. Are there parts that are tied to
specfic sensor hardware that are needed such that general purpose
data and algorithms aren't feasible? (Like a single universal
PostScript driver apparently isn't.)
My understanding from Thomas is that yes, a universal format could
be produced that would allow camera manufacturers to put in their
own secret sauces but allow others to at least get to the data to
de-morize the RAW data to at least produce an image.
2. A lot of camera hardware (within a class) is pretty much
identical these days. (8 meg prosumer digicams that all use the
same Sony sensor, for example.) But image quality isn't. The main
reaons for this is because of algorithm tweaks when decoding the
RAW data. How much of this do you expect the camera manufacturers
to disclose? If they don't disclose this, then we are back to the
"universal" RAW decoders being inferior to the manufacturer's RAW
decoders.
As I said, my understanding is that given enough effort, a company
can reverse engineer what they need to build an image. That's what
Thomas has had to do in most cases for Camera RAW and I'm sure the
same is true for the other converters on the market. It's difficult
and time consuming. If you've played with various converters using
the same RAW file, you'll see that you end up (at least with
default settings) vastly different quality and color appearance.
This is much like giving a color neg to 5 different labs. The
difference here is that each lab has no difficultly popping a neg
in their enlarger, figuring out what filter pack to use and
processing the paper. With RAW, the enlarger, the paper and the
process is totally proprietary so it's a lot more work for multiple
companies to even consider supporting a product to do this.
You can take Kodak, Agfa or Fuji color neg and produce a print in a
lab. You don't need to know all about how each company manufactured
the film to make a print. You can print using paper made by any of
the companies or use chemicals made by any of the companies but the
basic process is documented and standard. You'll get different
results (some may say Kodak paper is better or Fuji RA4 chemistry
is better). But you don't have to first figure out how to mix up
some proprietary chemistry to process the film (it's C41), you
don't need to figure out what proprietary paper and chemistry to
use. So we need some kind of somewhat open standard for processing
RAW data like we have for processing C41 film. Manufacturers can
still use their competitive advantages to build a better mouse
trap. OR better, they can say to themselves that Adobe has a killer
RAW processor, so they don't spend the time and money to supply
what is usually a pretty poor piece of software to go with their
cameras. 99 times out of 100, the supplied software from all the
major Japanese camera vendors is just awful.
3. The word "open" is heavily loaded these days for propriatory
software vendors. If they start granting open data standards,
then who knows, maybe you'd be asking for open source next. Or
for Adobe to free their color management patents so that other
image editing programs can implement color management.
We did this with ICC profiles. Not that the ICC was a perfect body
politically. The ICC was a large group of companies that decided to
produce an open standard for communicating color using an open
standard for ICC profiles but built in private tags so that each
manufacturer could add their own proprietary options. Yet on a Mac
or PC, any ICC savvy application would still function using the
profiles. That's akin to what we need.
I'm also mindful of "be careful of what you wish for." The current
"open document" format is XML, which is at least 10 times fatter
and slower than any of the propriatory document or data formats.
The alternative is having years of legacy images that can't be
opened. That's worse in my mind. I'll take slow but processed data
over data I can't use 10 years or more from now.
If Adobe can't make their PSD format be truly open, then why should
we expect the camera manufacturers to do so?
They did with TIFF. It's a much older and simpler format anyway. If
you had the option of making an open, standard image file format,
why PSD? You've got layers, alpha channels, clipping paths, and
lots of other complicated stuff. TIFF has been updated by Adobe sot
that IF you wanted to have a TIFF with layers, you can but don't
expect older software to deal with it. But it's backwards
compatible to a very high degree. You can setup Photoshop to save
out a layered TIFF and still make a setting in Photoshop so that
someone running Photoshop 1.0 can open the file.
I'm not arguing that a universal RAW format is a bad idea, but I'd
like to know more about the details. The devil is always in the
details.
There are no details per say. That's why the site and this group
has come together to discuss this and try and work out the details.
THEN use our power and influence in the industry to make it happen.
We need a 900lb Gorilla and Fortunately one exists that has an
interest in seeing this happen. Then we use our buying power and
influence. Any manufacturer of a digital cameras that doesn't
support an open RAW format, no matter how great gets a point taken
off the review of the product.
