«

Barry and Paul,

Great discussion points. My reading of what you've written is is:

1. DNG is getting nice traction
2. Nikon is likely to be very resistant, Canon a bit less resistant.
3. Reverse engineering RAW formats is a hurdle, DNG conversion requires
that reverse engineering. (Though I get the sense Adobe's RAW
converters are comparable to Canon's? - for example?)
4. Vendors are still sorting out the metadata standards a bit (XMP in
DNG).

It sounds like if either Nikon or Canon went to in-camera DNG the
industry would have to follow along pretty darned quickly. I'm hopeful
Canon will cave first; one of the reasons I went with the XT rather
than a D50 was my perception that Nikon was even more proprietary than
Canon (could be wrong though!).

For my purposes I'll stay with RAW for this next two years, then plan
to convert all my RAW to DNG @ 2007-2008. I figure by then metadata and
RAW conversion issues will be sorted out, but that's not so far away
that the Digital Rebel XT RAW format will have been forgotten. After
that time I'll convert to DNG on import as a part of my workflow and
discard the RAWs -- and my Rebel XT replacement ought to do DNG in
camera.

I can live with that plan!

(I'm no pro, so I don't feel so bad about tossing the RAWs. I realize a
pro would keep both RAW and DNG.)

john

meta: standards, RAW, DNG, Adobe, Nikon, Canon, format, image, digital
photography, jfaughnan, jgfaughnan

"David J Taylor" <david-> writes:

>I save all my most important images in PNG. It does as well as or better
>than standard lossless compression algorithms. OK, what I save don't
>happen to be camera images (I use JPEG for that), but they are full tone
>images (they happen to be greyscale). To imply that PNG is only for
>graphics would be misleading - it's equally good for greyscale or colour
>images.

PNG works OK for archiving images, though it wasn't really intended for
that. The compression method chosen takes much longer to encode than
decode, which makes very good sense on the web (where one person will
encode it and thousands of people will decode it), but less so for
private storage.

However, it will never make a good way to store raw data from
Bayer-sensor cameras, because it has no way of storing just one colour
component at each pixel location. Even if you took the raw data and fed
it to PNG as a greyscale file, compression would be poor because
adjacent pixels are not well correlated with each other (like they would
be in a real greyscale image).

Dave

"John Faughnan" <> writes:
> 1. DNG is getting nice traction

I'd say "some" traction; "nice" is yet to be seen.

> 2. Nikon is likely to be very resistant, Canon a bit less resistant.

Maybe. Who knows.

> 3. Reverse engineering RAW formats is a hurdle, DNG conversion
> requires that reverse engineering. (Though I get the sense Adobe's
> RAW converters are comparable to Canon's? - for example?)

Conversion can be done without reverse engineering, by using an SDK
licensed from the camera vendor instead. That's probably what Adobe
does. Adobe likes this just fine, since it shuts out alternatives
like GIMP. The real threat is that future native raw formats will
keyed by the camera's serial number and will only be convertable by
SDK's that authenticate against the DRM hardware that will be built
into future PC's. Reverse engineering and writing free converters
will then become pretty much impossible. That has to be what some
parts of Adobe are really hoping for.

> It sounds like if either Nikon or Canon went to in-camera DNG the
> industry would have to follow along pretty darned quickly. I'm hopeful
> Canon will cave first;

I doubt that either one will switch to DNG, at least any time soon.

In article <>,
Paul Rubin <http://> wrote:
>Conversion can be done without reverse engineering, by using an SDK
>licensed from the camera vendor instead. That's probably what Adobe
>does.

Given the lack of white balance support for some of Nikon's camera
last year, I sort of doubt that Adobe is just using the camera manufacturers
SDK.

>Adobe likes this just fine, since it shuts out alternatives
>like GIMP.

Mplayer is quite good at using standard Windows codecs. And I doubt that
GIMP is really a competitor to worry about at this moment.

>The real threat is that future native raw formats will
>keyed by the camera's serial number and will only be convertable by
>SDK's that authenticate against the DRM hardware that will be built
>into future PC's. Reverse engineering and writing free converters
>will then become pretty much impossible.

Those kinds of DRM solutions tend to be a support nightmare.

You may be able to do it for professional cameras, because professionals
have to buy cameras anyhow. You may be able to do it with low budget
cameras, because people accept just about anything as long as the price
is low enough.

DRM is likely to kill large parts of the slightly more serious amateur market.
That group will simply buy a low budget camera and give up on photography
as a hobby.


--
That was it. Done. The faulty Monk was turned out into the desert where it
could believe what it liked, including the idea that it had been hard done
by. It was allowed to keep its horse, since horses were so cheap to make.
-- Douglas Adams in Dirk Gently's Holistic Detective Agency

Thanks for the info about Leica and Hasselblad supporting DNG.

Paul Rubin <http://> wrote:
> PNG is used in about the same way as GIF was (and to some extent still is).

Not really. GIF never allowed > 256 colors in most implementations.
GIF as a photo storage format is totally inadvisable. Large, too!

PNG's only limitation as a photo storage format is 8-bits per color
in most implementations. Compared to 8-bit JPEG 2000, PNG is not much
larger, encodes faster, and can be displayed easily by most software.

Dave Martindale wrote:
[]
> However, it [PNG] will never make a good way to store raw data from
> Bayer-sensor cameras, because it has no way of storing just one colour
> component at each pixel location. Even if you took the raw data and
> fed
> it to PNG as a greyscale file, compression would be poor because
> adjacent pixels are not well correlated with each other (like they
> would
> be in a real greyscale image).
>
> Dave

Agreed. As I don't store RAW data, that doesn't bother me.

Cheers,
David

Paul Rubin wrote:
> "Barry Pearson" <> writes:
> > There are 3 cameras and 4 digital backs that output DNG as their native
> > raw format. This is irrelevant! If you convert directly from the memory
> > card, you don't CARE what the camera's native raw format is:
> > http://www.barry.pearson.name/articl...y1.htm#cameras
>
> The native raw format is extremely relevant because if if it's
> something other than DNG, you need conversion software; and in cases
> where the native raw format is proprietary, that means you either have
> to rely on reverse engineering, or else use a proprietary converter.

Fair comment. I was really writing the above in response to the OP's
"What's the growth path like? What are the main things holding it back
from even greater acceptance?". There is a view sometimes expressed
that photographers won't take up DNG until the cameras output it. But
the growth doesn't depend on that. The blocker for some photographers
is really the camera manufacturer's software, not the camera itself.
Hence my statement above.

What I have written at OpenRAW is that we do need cameras to converge
on DNG, and in the meantime we want camera manufacturers to publish
their raw formats so that high quality DNG Converters can be produced,
or else the camera manufactuers can provide their own DNG converters.
(Hasselblad-Imacon appear to be the only manufacturer to provide its
own DNG converter for (some of) its own non-DNG raw formats).

(I call reverse-engineering "shoddy engineering". Perhaps you would
agree with that!)

[snip]
> What is needed is for camera vendors to 1) support open formats (such
> as DNG) as native raw formats; and 2) announce their commitment to
> using open formats and stick to the commitments. High end camera
> vendors like Leica, Hasselblad, and (iirc) Phase One are using DNG,
> but some semi-mass-market vendors like Nikon seem to be moving in the
> other direction.

Precisely. (Except that Phase One haven't gone to DNG yet, although
they have some plans in place. And Samsung & Ricoh have used it on one
camera each). My guess is that Nikon will be the last to change! And
Canon the last but one. But we'll see. Certainly, Nikon are the one who
have caused most trouble for other raw converters and for the DNG
Converter by their behaviour. Lightroom doesn't support the WB fully
for some recent Nikons, because its architecture doesn't work well with
the mini-SDK eventually provided by Nikon and used in other Adobe
products.

--
Barry Pearson
http://www.barry.pearson.name/photography/
http://www.birdsandanimals.info/

John Faughnan wrote:
> Barry and Paul,
>
> Great discussion points. My reading of what you've written is is:
>
> 1. DNG is getting nice traction
> 2. Nikon is likely to be very resistant, Canon a bit less resistant.
> 3. Reverse engineering RAW formats is a hurdle, DNG conversion requires
> that reverse engineering. (Though I get the sense Adobe's RAW
> converters are comparable to Canon's? - for example?)
> 4. Vendors are still sorting out the metadata standards a bit (XMP in
> DNG).

"3": yes, it is a hurdle. In fact, reverse-engineering the file format
isn't typically the delaying factor when new cameras come out. The
delay is caused by the need to generate camera calibration data, which
needs Adobe and others to obtain cameras for themselves and try them.
Apparently Adobe simply buy them in the normal way - it is amazing to
read in the Adobe forums about the problems they have!

I believe XMP within DNG is the next "battle ground". It is one of the
attractive aspects of DNG for me, and plenty of others. (There is a new
book by Peter Krogue, commonly known as "The DAM Book", about using
Digital Asset Management on your images, and he uses XMP with DNG).

[snip]
> For my purposes I'll stay with RAW for this next two years, then plan
> to convert all my RAW to DNG @ 2007-2008. I figure by then metadata and
> RAW conversion issues will be sorted out, but that's not so far away
> that the Digital Rebel XT RAW format will have been forgotten. After
> that time I'll convert to DNG on import as a part of my workflow and
> discard the RAWs -- and my Rebel XT replacement ought to do DNG in
> camera.

I think many photographers are delaying like that for a year or two to
see what happens. But I put various personal and descriptive metadata
in the DNGs, so that in years to come I will be able to find my images.
It will be too late to wait until then! I put the data in immediately
after each shoot.

[snip]
> (I'm no pro, so I don't feel so bad about tossing the RAWs. I realize a
> pro would keep both RAW and DNG.)

Some would. I've seen Jeff Schewe say otherwise, and I think some
opinion leaders like that are prepared to toss the originals. The
typical downside is that you can't use the camera manufacturers
software then.

--
Barry Pearson
http://www.barry.pearson.name/photography/
http://www.birdsandanimals.info/

Paul Rubin wrote:
[snip]
> Conversion can be done without reverse engineering, by using an SDK
> licensed from the camera vendor instead. That's probably what Adobe
> does.

No. The typical SDK does its own raw conversion, and delivers an RGB
image instead of the sensor data. That is no use for raw converter
suppliers, although it can be useful for viewers, etc. Adobe and the
others do a full reverse-engineering.

An exception is for the Nikon WB issues, where Nikon now provide, and
Adobe use, a "mini-SDK" provided by Nikon that just handles the WB. The
rest of the file is handled directly.

> Adobe likes this just fine, since it shuts out alternatives
> like GIMP.
[snip]

If Adobe wanted to shut out alternatives, they would not have launched
DNG! Adobe want all companies to use DNG, and make it as easy as
possible by publishing the specification, publishing a world-wide
royalty-free license for anyone to use DNG, and providing a free DNG
Converter.

--
Barry Pearson
http://www.barry.pearson.name/photography/
http://www.birdsandanimals.info/

Barry Pearson <> wrote:
> Paul Rubin wrote:
>> What is needed is for camera vendors to 1) support open formats (such
>> as DNG) as native raw formats; ...
> Precisely.

If it takes time to convert RAW to DNG in-camera, and RAW is already
too slow (usually slower than JPEG, probably due to I/O requirements),
how is this going to help photographers?