Future of Exif standard

[BogdanH]

Hi,

Following writting isn't ment to be some Exif analysis -just user obsevation. And I hope this forum section is ment for such posts as well.

The first time, I've heard about latest Exif version (v2.3, bublished in April, 2010), was in this forum. While it is normal, that such event get attention here, I'm wondering anyway: how come, that this wasn't mentioned elsewhere? I mean, I'm visiting some well known photo dedicated sites quite regulary. Ok, maybe I've missed that "news" elsewhere...

Before getting to the point, let's see Exif revision history. Not going much into detail, my impression is, that latest major version was v2.0, which dates from 1998! Inbetween, there were four (sub)versions, which didn't contain something really new. Even worse: as I see it, camera makers started to ignore Exif (ExifIFD) tags.

Let's take SceneCaptureType tag (which describes scene that was shot) as example. Per Exif standard, it can contain any value between 0 and 65535, where following values are standardized:

0 -Standard (whatever that means),
1 -Landscape,
2 -Portrait,
3 -Night scene,
Other -reserved.

-and that's it! You took macro photo? Or sport/action shot? Or any other (out of ~65000 possible) photo? Well, tag is there, but we can't use it reasonable -because, in past ten years, nobody bothered to define more than 3(!) tag values.

Now, I'm comming to the point of this post.
Per Exif v2.3, now it's possible to tag data about lenses (LensInfo, LansModel, etc.) inside ExifIFD section. At last -but too late! For years, camera makers are allready writting this info into proprietary (read: secret) MakerNote section. Does anybody believe, that now camera makers will start to use those ExifIFD tags, just because CIPA woke up after all those years? And we blame camera makers, how they are "hiding" ther data -but, per standard, there's no other place than MakerNote available. Ok, camera makers will allways want to keep some data secret... What I am talking about is public (user oriented) info -lens being used isn't ment to be kept that secret, I believe.
Because of all that, it's obvious why Exif v2.3 didn't got much attention... Most are talking XMP anyway.

Thanks for reading,
Bogdan

PS: Sorry for my not perfect english language.

[Phil Harvey]

Hi Bogdan,

You are exactly correct.  CIPA (previously JEITA) seem to be unwilling to add new definitions to existing tags as you mentioned, which makes the old tags essentially useless.

There are other more significant problems too (see my comments on metadata standards), which are unlikely to be fixed.

The real problem is that camera users have no input in this.  Other than someone else taking over the standard, I don't see any solution.  But if someone else takes over the standard, it is unlikely that the camera manufacturers will use the new standard.

[BogdanH]

Hi Phil,

Thank you for link pointing to your page, so I could re-read it. I knew you wrote something about metadata issues "somewhere", but this time I simply couldn't resist to comment -hopefully, I made some things clear for "non-metadata-expert" readers

About your second part: I believe, "standard" has allready been overtaken:
1. By camera makers -they can't affort to wait till some organization (CIPA in this case) will agree with their needs. Standard organisations simply don't act fast enough. The result we have, is MakerNote.
2, By Adobe -they not only have the power for doing that, they are forced to react quick (on camera developement). And at last, they know what data is needed (e.g.TimeZone). The result we have, is XMP (and DNG).
Ok, as programmer, I'm not a big fan of XMP.. I prefer things ordered by numbers.. somehow difficult to parse XMP data, though...

Greetings,
Bogdan

[Phil Harvey]

Hi Bogdan,

XMP really is not suited to embedded processors (as in cameras), but it seems that some Canon cameras are starting to produce JPEG's containing XMP.  I wish it weren't necessary because XMP is a real pain (creating XMP is easy, but reading or editing existing XMP is a realy pain).

- Phil

[BogdanH]

..but it seems that some Canon cameras are starting to produce JPEG's containing XMP...

That would explain something I've discovered recently. Canon DPP (you know, raw converter) shows some of metadata written into raw files afterwards (by using Exiftool). The funny thing is, those tags (mostly related to location and photographer's data) are named as "IPTC" inside DPP. But gues what: if data is stored as "real" IPTC data, DPP ignores that data entirely -DPP shows those values only if they're stored inside XMP. This behaviour somehow confirms what you're saying.

Just out of curiosity... can you name Canon camera which allready does write XMP data into image file?

I wish it weren't necessary because XMP is a real pain (creating XMP is easy, but reading or editing existing XMP is a realy pain).

Agree -especially if it's comming from you

Bogdan

[Phil Harvey]

The funny thing is, those tags (mostly related to location and photographer's data) are named as "IPTC" inside DPP. But gues what: if data is stored as "real" IPTC data, DPP ignores that data entirely -DPP shows those values only if they're stored inside XMP.

What ExifTool calls IPTC is the old IPTC standard.  But there is a new IPTC schema for XMP which is probably what DPP is referring to. 

Just out of curiosity... can you name Canon camera which allready does write XMP data into image file?

Basically all of the new PowerShot models I think.  You can check the Canon images in my metadata repository to see the exact models.  So far, only the Rating tag is being written.

- Phil