Michael's DNG comment

[deejjjaaaa]

Hi,

I'm not really sure how an ICC profile belongs to a raw file. I got the impression that cameras cannot really be profiled in the iCC sense, but I may be wrong.

it was long understood that ICC is just a container for a data that raw converters are using... and they do not need to use that data exactly as ICC prescribes

[deejjjaaaa]

Does a CR2 or a NEF?

Not sure what your point is there...

I was answering to the statement that DNG advantage is to have camera profile embedded in DNG file itself (by manufacturer)... my point is that is a half baked solution... DNG has to allow several dcp / icc profiles to be embedded simultaneously to start with... plus as most of raw converters do not follow Adobe's dcp model it will be again an argument against DNG adoption for manufacturers who are using non Adobe raw converters as OEM... the data stored in icc profiles is not standard in the sense how most raw converters are using it, regardless of what ICC (color.org) wants/thinks.

[deejjjaaaa]

LR and ACR support that in DNG images via XMP, but so far as I am aware, its not documented.

Sandy

so much for a standard that we shall use

[opgr]

From the documentation:

Camera Profiles
DNG 1.2.0.0 and later formalizes the concept of a “camera profile” and allows multiple camera profiles to be embedded in a single DNG file. A camera profile consists of a set of tags (both existing in DNG versions earlier than 1.2.0.0 and newly defined in DNG version 1.2.0.0), some of which are optional.

I presume this is what you guys were talking about? the tag to look for is called ExtraCameraProfiles if I am not mistaken. It appears that some of you try to vehemently argue against DNG while perhaps not being informed properly?

[sandymc]

From the documentation:

Camera Profiles
DNG 1.2.0.0 and later formalizes the concept of a “camera profile” and allows multiple camera profiles to be embedded in a single DNG file. A camera profile consists of a set of tags (both existing in DNG versions earlier than 1.2.0.0 and newly defined in DNG version 1.2.0.0), some of which are optional.

I presume this is what you guys were talking about? the tag to look for is called ExtraCameraProfiles if I am not mistaken. It appears that some of you try to vehemently argue against DNG while perhaps not being informed properly?

I could be wrong, but I don't think that tag is actually used when ACR or LR store multiple profiles. But maybe I just haven't got LR set up right to see that.

Sandy

[opgr]

I could be wrong, but I don't think that tag is actually used when ACR or LR store multiple profiles. But maybe I just haven't got LR set up right to see that.

Why would ACR or LR be meddling with your DNG files? They are consumers of DNG, not producers…

If you are referring to the calibration tab in those programs, why would you think that those settings should be stored in any DNG file selected for processing? And more importantly, when would you want them to append that if at all?

[BJL]

Ah, so you be happy having a copy of the Mona Lisa and trashing the original painting? As we've seen since the beginning of the digital revolution, raw digital captures keep getting better and better because of the advances of raw processing capability. No having access to the original would really be a big problem for archivists.

Realistically, most images that reach the hands of archivists are probably going to be in a final output format like JPEG, or worse yet, prints that must be scanned, rather than any kind of raw format. So archivists would be relatively thrilled to get images in a format like 16-bit TIFF without the lossy compression and 8-bit per channel limitations of JPEG. (Or in any raw format for which they know the specifications.)

As a practical matter, I think Erik Kaffehr has a good strategic idea in this post in a related thread. My version: only buy a camera if and when it either uses an acceptable format or good software is available that supports its raw file format including lossless conversion into a suitable format. Note that this rules out things like formats with encrypted white balance information that third party software cannot reliably read.

(Aside: I would not compare a TIFF to a human-made copy of a painting. A large, high quality print made under the control of the photographer is a closer analogy, assuming that the photographer is in control of the conversion to TIFF, even if that just means choosing the settings on the camera.)

[sandymc]

Why would ACR or LR be meddling with your DNG files? They are consumers of DNG, not producers…

If you are referring to the calibration tab in those programs, why would you think that those settings should be stored in any DNG file selected for processing? And more importantly, when would you want them to append that if at all?

ACR and LR by default store ALL of their data on adjustments, profiles you use, etc inside DNGs. It's one of the advantages of DNG - no sidecar files, so if you copy a DNG from one system to another, all of your adjustments go along automatically. Although, not everyone views having your images modified every time you edit them as an advantage.

Sandy

[aduke]

ACR and LR by default store ALL of their data on adjustments, profiles you use, etc inside DNGs. It's one of the advantages of DNG - no sidecar files, so if you copy a DNG from one system to another, all of your adjustments go along automatically. Although, not everyone views having your images modified every time you edit them as an advantage.

Sandy

There's an interesting point in there. All of the adjustments are in the DNG, but that does not necessarily make them meaningful to a raw converter other than the last one used to process the DNG. With LR4, the program knew that it was not the version to last modify the DNG but did know enough about the previous version to use the settings. If the new processor did not know the meaning of the settings, what would it show and do?

Alan

[digitaldog]

All of the adjustments are in the DNG, but that does not necessarily make them meaningful to a raw converter other than the last one used to process the DNG.

That would be true if the data wasn't DNG too. If you wanted to pop E6 film in C41 chemicals, you'd be in the same boat.

Another raw converter that could understand the DNG would apply a different default rendering.

[opgr]

ACR and LR by default store ALL of their data on adjustments, profiles you use, etc inside DNGs. It's one of the advantages of DNG - no sidecar files, so if you copy a DNG from one system to another, all of your adjustments go along automatically. Although, not everyone views having your images modified every time you edit them as an advantage.

Ah, yes, in that case program settings are likely stored under a private tag, not under several public tags.

[jeremypayne]

Although, not everyone views having your images modified every time you edit them as an advantage.

I find that to be a pretty big disadvantage of the DNG model.

I like the fact that every time I do an incremental backup of my library only the XMPs with changes get backed up.

[aduke]

That would be true if the data wasn't DNG too. If you wanted to pop E6 film in C41 chemicals, you'd be in the same boat.

Another raw converter that could understand the DNG would apply a different default rendering.

I'm sorry that I wasn't clear. I wasn't talking about cross-processing, but rather thinking about how, in 50 years, when ACR has been forgotten and replaced by something else, the raw converter that reads DNG's knows what the numbers in the following lines from a DNG are supposed to mean:

   <crs:FillLight>58</crs:FillLight>
   <crs:Vibrance>0</crs:Vibrance>
   <crs:HighlightRecovery>47</crs:HighlightRecovery>
   <crs:Clarity>+50</crs:Clarity>

Alan

[digitaldog]

I wasn't talking about cross-processing, but rather thinking about how, in 50 years, when ACR has been forgotten and replaced by something else, the raw converter that reads DNG's knows what the numbers in the following lines from a DNG are supposed to mean:

   <crs:FillLight>58</crs:FillLight>
   <crs:Vibrance>0</crs:Vibrance>
   <crs:HighlightRecovery>47</crs:HighlightRecovery>
   <crs:Clarity>+50</crs:Clarity>

It doesn’t know. But the raw converter that understands DNG will have it’s own set of metadata rendering instructions.

[aduke]

That I'll buy. Having the adjustments in the DNG makes it easier to ensure that the adjustments follow the image, but it doesn't do anything to help the new generation understand what the adjusted image looked like. There is, of course, the thumbnail in the DNG also, which does provide a record.

Thanks for the clarification.

Alan

[Fips]

That I'll buy. Having the adjustments in the DNG makes it easier to ensure that the adjustments follow the image, but it doesn't do anything to help the new generation understand what the adjusted image looked like. There is, of course, the thumbnail in the DNG also, which does provide a record.

But this is not a disadvantage of DNG per se (not implying that you meant to say that). It's just like with film. What's put into the archive is the mostly unmodified negative and if you want to preserve what the photographer intended to show, to also have to archive an additional print - which would be a developed TIFF or JPG in our case.
Maybe at some point in the future, when computers haven gotten fast enough to handle it, DNG will allow the preview image not only to be full size but also stored with a lossless compression and full colour depth. This would eliminate the problem you pointed out.

[aduke]

There are two  disadvantages to this. Jeremy Payne pointed to the real one. How important this is is debatable, perhaps quite important if your backup process will send the entire file to backup every time you modify the image, much less so if backup only notes the difference between the previous version and the new version.

The second disadvantage is very theoretical. Its the care that must be taken to protect the contents of a file kept in an archive. You certainly do not want anybody to be able to change the archived version, especially inadvertently through trial changes in a raw converter. This certainly does not warrant changing the design of the DNG file.

Alan

[Alan Goldhammer]

I like the fact that every time I do an incremental backup of my library only the XMPs with changes get backed up.

I hope you also back up new RAW images!   (I do the same thing in terms of XMPs)

[jeremypayne]

I hope you also back up new RAW images!   (I do the same thing in terms of XMPs)

Yes, of course ... but the point is that the RAWs only have to get backed up once.

I have a vague notion that someday I will archive my "selected" RAWs inside a DNG, but I haven't ever gotten around to that ...

[johnbeardy]

You don't need to keep backing up your DNGs - it's incomplete and redundant. Back them up once when they're new, keep backing up your Lightroom catalogue, and 100% of your work is covered.