Lens data requested

As originally posted by me here, and modified somewhat:

-----

Given the wish from a lot of us that Dalifer, ExifTool, RAWhide, and other programs (*cough* Lightroom, but I'm dreaming in Technicolor) supported lens identification info fully, we as the users need to be proactive on accumulating this information. If we want the data, we as Minolta mount users have to get it together.

Of course to make this feasible we need the raw data first.

I ended up using ExifTool today with all my (many) lenses to add the info to the Dyxum makernotes database. Unfortunately, the database seems to have several inconsistencies, and so for all my observations, I have noted them in comments with date so people actually know where it came from, and from who. If you can access the makernotes database and can confirm a value, please post saying so!

I figure I'll repeat what info we actually need to make identifying a lens from exif data work:

To review:
There are only 3 pieces of lens fingerprints info provided by the exif data for certain:

1. Lens ID. Unfortunately, this is not a standalone identifier. Some Sigma lenses for example use the same lens ID cause Sigma reverse engineers everything and doesn't pay licensing fees.
Additional bad thing that may happen: could a given lens have more than one lens ID based on production date, etc?
Best case scenario: Only one lens has the ID. Happy day. We can ID the lens.
Worst case scenario: Several lenses have the same ID. The good news here is that we do know a bunch of lenses the lens is not!

2. Focal length. This makes identifying primes pretty easy (naturally), and does help tell us in what focal length range the lens will fall (a shot taken with a 70-300 lens will not have an focal length of 30mm recorded, right?)
Quirky exceptions: I believe the beercan records and exif focal length of 75mm when the lens is set to 70mm. Also, doesn't a Sigma 15mm Fish record as 16mm in exif?
Best case scenario: prime
Worst case scenario: 18-200, 28-300 and other superzooms which in essence don't tell us anything either.

3. Maximum aperture. The lens records what the largest aperture available is. Of course this info is functionally useless without the focal length, but we have that... this info is kinda needed in conjunction with the focal length. We need to create tables for each lens stating what the max aperture at each focal length is. For constant aperture lenses, this is easy. For variable aperture lenses, we will need to record this data somewhere (this data is not currently recorded online anywhere so far as I know, though it has been discussed in the past).


As for identical Minolta/Sony rebadges, and when we just can't ID the lens for certain: So this is where we're forced to give up. All we can do here is present a user a list of what lenses it might be, and have the user choose.
Best case scenario: the user actually only has one lens that matches the available options. Since in the vast majority of cases, this data is getting observed on the photographer's computer, he/she should know what lens he/she actually has. Give them the job of ID'ing the lens in their collection. Flag that lens somehow to know that any future photos with matching criteria are that lens, though provide the user a way to edit the flagged lens collection (people's lineups change with time). In future iterations, just check the flagged lenses a user says are in their lineup to minimize the necessary amount of processing time.

Worst case scenario: more than one lens. User will have to flag both, and manually identify each shot that could be from any of more than one lens. If user's brain works (most users offload shots to computers relatively soon after taking them, they should remember the lens used).

I'll leave the issue of storing data (xml, csv, etc) as discussed by persons more capable than I, but I will say this: Software that accesses such a database really should be able to select it as an external file, and not integrate it into source code since the software and the lens ID data will not get updated at the same time as said software in all likelihood.

Store the database as a user downloadable file from dyxum (is there any other logical choice of where to keep such a database?). Actually this is already the case, though the info dyxum has is incomplete at best... but it's still by far the most complete database available.

Oh, and of course as the unique identifier for each lens (since the makerfile lens id is not mutually exclusive?): the dyxum lens database id, naturally.

Edited by hotwire - 26 November 2007 at 05:13

http://www.dyxum.com/dforum/forum_posts.asp?TID=6371&PID=50077#50077

dalibor wrote: Hello, let my add my two cents to this very interesting thread. I have started to collect Lens IDs two years ago when I have bought my first 7D. The purpose was to display the name of the lens in my Dalifer software, which is the utility for decoding the MakerNote contents. Dalifer provides you with the most detailed Exif/MakerNote information from all software. And the lite version is free. See http://www.dalibor.cz/minolta/dalifer.htm The version available for download is rather old, but I can provide you with a beta version which does have all known LensIDs built-in. I am very glad that someone is interested in this stuff as well, so we can compare our results. I can see three different groups of LensIDs: 0-127   = new Minolta lenses 128-141 = Cosina/Voigtlander lenses >25501 = old Minolta + new Sony lenses >45711 = Vivitar lenses In my opinion it makes no sense to study the LensIDs of Sigma, Tokina and Tamron lenses. Those seem to be randomly chosen and they overlap with Minoltas and with each other as well. I think that Cosina/Voigtlander and Vivitar have paid license fees to Minolta and they got its ID range assigned. The others probably didn't want to pay and they use the LensIDs as they want without any logic. But this leads to a conclusion that the camera firmware either does believe the LensID when fine-tuning the ADI and AntiShake and therefore not work proerly with non-Minolta lenses. Or, as the IDs are not unique, the LensID is not used at all. What do you think about it? I would like to point out one interesting LensID, which is 25858. This is the only ID I know so far which does not follow the xxxx1 pattern. This is Minolta 35-105 f3.5-4.5 (new). I thought that the last digit could be a revision number of the lens, but I have not seen the photo with ID 25851 and the other old/new/RS lenses do not use this pattern anyway. The Michael Hohner's list is a very good piece of work. Just do not expect that every lens listed there has its own ID. Some variants of the lens Michael lists use the same ID (for example 50/2.8 Macro = 50/2.8 Macro RS). And some of the lenses might have two different IDs (for example 85/1.4 RS G and 35-105/3.5-4.5 (Macro) and 35-80/4-5.6 II). But definitely Michael's list is the best refference table and I would suggest to use his naming convention for the lenses. I was not aware that there is a quest for a LensID database in minolta-forum.de as I do not visit this forum regularly as I do not feel comfortable reading in German, but I am very glad that it is there. I would welcome if we could start such thread here at Dyxum in English and use Matthias as a inter-forum-ambassador. :-) BTW here is my wish-list for a pictures taken with particular lens, which I would need to make my database complete. 20/2.8 (RS = with rubber focusing grip) 85/1.4 G (D) Limited (longer version) 200/4 Macro APO G 300/2.8 APO G (D) SSM 600/4 HS-APO G 24-50/4 (RS = with rubber focusing grip) 28-80/4-5.6 xi (Macro) 28-85/3.5-4.5 (RS = with rubber focusing grip) 35-80/4-5.6 (II = 49mm thread) 35-105/3.5-4.5 (Macro) 70-210/4.5-5.6 (II = without focus hold button) 100-300/4.5-5.6 APO (the older version with smooth focusing ring) any lens with/without teleconverter Minolta 1.4x APO (D) any lens with/without teleconverter Minolta 1.4x APO original any lens with/without teleconverter Minolta 2x APO original any Cosina lens any Vivitar lens any Voightlander lens any Opteka lens Regards, Dalibor

Edited by hotwire - 24 November 2007 at 23:14

http://www.dyxum.com/dforum/forum_posts.asp?TID=6371&PID=50215#50215

my-spot wrote: In the software I'm writing I provide all of the lens codes I have found in a .csv file that can be easily edited by the end user using excel or a text editor. I don't make any claims about its accuracy but the idea is that most people will not personally own more then one lens with the same ID. If the program displays the wrong lens for a given ID then the user can edit the list so the correct lens is used for a given ID or add the value. Although the code isn't finished yet, I will make it further filter based on FL and aperture which is also supplied in the list. A copy of my csv file can be found here...

Edited by hotwire - 28 November 2007 at 05:04

http://www.dyxum.com/dforum/forum_posts.asp?TID=6371&PID=52029#52029

Kiklop wrote: Welcome to dyxum Phil; I really apreciate all your sharings !! I guess you noticed mention of xml file provided by Sony?

Edited by hotwire - 28 November 2007 at 05:06

http://www.dyxum.com/dforum/forum_posts.asp?TID=6371&PID=52678#52678

my-spot wrote: This has been a most interesting thread. But we all have separate lists, I have my lens list, Kiklop has the Dyxum List, there is the Sony list, Dalibor has one, Phil has a list, and there are a few others as well. All of us have programs written or in the works that use a LensID list. I propose that we all come up with a common format (.csv or .xml ??) that can be maintained here (Dyxum) that we all can use in our programs. This would assure we all have a consistent and complete list. We would have to make sure all of the fields (LensID, Brand, Name, Min FL, Max FL, Min Ap, Max Ap, etc...) we need are included, etc, etc, etc... Dalibor, Kiklop, Phil, others... What do you think?

Edited by hotwire - 28 November 2007 at 05:05

Current Lens ID List on Dyxum:

http://www.dyxum.com/dAdmin/lenses/MakerNoteList_Public.asp?stro=lens

And to edit data for that page:

http://www.dyxum.com/dAdmin/lenses/EditLensMakerNote_Public.asp?IDManofacturer=16

Edited by hotwire - 24 November 2007 at 23:00

I have spent part of today going through my lenses and recording the focal lengths they provide info on. Surprisingly, the lenses don't necessarily have the same intervals of focal length recording. I need some more time on this (tomorrow hopefully), where I will get the max apertures at each respective focal length (for the variable aperture zoom lenses), but I figured I should post the values I have observed with my lenses:

Sig 12-24: 12, 13, 15, 16, 17, 18, 20, 22, 24mm
KM 17-35 D: 17, 18, 20, 22, 24, 26, 28, 30, 35
KM 28-75/2.8: 28, 30, 35, 40, 45, 50 55, 60, 70, 75
Sony 70-200 G: 70, 75, 80, 85, 90, 100, 105, 110, 120, 130, 135, 140, 150, 160, 180, 200
Min 100-300 APO: 100, 120, 135, 140, 150, 160, 180, 200, 210, 230, 250, 280, 300               
Tamron 24-135: 24, 28, 35, 40, 45, 50, 55, 60, 70, 75, 80, 85, 90, 100, 110, 135
Sony 18-70: 18, 20, 22, 24, 26, 28, 30, 35, 40, 45, 50, 55, 60, 70     
Min 28-135: 28, 30, 35, 40, 45, 50, 55, 60, 70, 75, 85, 90, 100, 110, 120, 135
Min 70-210/4: 75, 90, 100, 105, 120, 130, 135, 150, 160, 180, 200, 210

What this means in layman's terms: depending on the lens you mount, you may set the lens for a particular focal length, but your exif will not necessarily list that focal length. It will only list the nearest interval of focal length.

As a clear rule, the interval of the exif recorded focal length is proportional to the focal length.

Generally, from 0-20mm (approx), it records in 1mm intervals, with some exceptions of 2mm lengths (though notably my sample here is very limited, and I don't have anything less than 12mm)

From 20-30, it appears to all be in 2mm intervals, with one exception (no 26mm or 30mm on the my Tamron 24-135).

From 30mm and up things get interesting. Some intervals of 5mm, some of 10, and some 30mm intervals, but very little in consistency. Seems to all depend on the lens.

If anyone wants to discuss this project with me, please pm me, or even send me a message on msn and we can maybe talk real-time. I still need to figure out the most effective way of logging and presenting this information, and would love some feedback on that.