Looking into Makernotes of Minolta RD-3000

[dct]

Hi all

I don't know if there are any other people out there, interested in such an old thing like an RD-3000  
Let us see...

Attached see an image of the Minolta Dîmage RD-3000. More informations at
http://www.dpreview.com/reviews/specs/Konica_Minolta/minolta_rd3000.asp

First of all a big thanks to Phil, for your work on exiftool. I hope that some of my informations may also be of interest for any update on exiftool, around makernotes.

Then let me  explain the problem regarding the makernotes of the RD-3000 and about my first findings exploring these exif data. I will show here my complete investigation and thoughts. Maybe this helps to find any wrong assumptions I made or pushes someone other in the right direction to find the required solutions.

============================================================

1) Looking for the makernote

There seems to be many myths around the makernotes for this camera. Even the very interesting pages of "geve" (http://gvsoft.homedns.org/exif/makernote-identifier.html#Minolta6) don't match the reality for the RD-3000. In short: There is no information on the web.

Why did I look for this makernote? I was wondering if the old RD-3000 did record any lens information (Vectis lenses) or details on exposure and similar things like in other cameras. And so I started over...

Empiric scrutiny of a meta data dump (exiftool -htmlDump) brings us to the starting point:

The tag Exif:MakerNotes is there in JPEG and TIFF files coming from the RD-3000.

Code Select Expand

Always same content is: 92 7c 00 07 00 00 01 f4  00 00 06 0c

Usage:
92 7c 00 07 37500 (Makernotes Tag ID)
00 00 01 f4 500 (Lenght in Bytes)
00 00 06 0c Starting address of MakerNote

============================================================

2) Wrong makernotes address in JPEG

Here we have the first big firmware problem of the RD-3000:

For best quality images (TIFF), the address 060c is right. But in the Exif:MakerNotes tag of the JPEG files, you have always exactly the same starting address of 060c, which is wrong, pointing somewhere into the JPEG stripes.

The same makernotes as in TIFF are there, but always (as far I could test) after the Exif:FNumber value. I think mostly at 01ce.

Here is my first question: Is there any way to rebuild the right makernotes address for JPEG files, without using some special hex editor tools? I tried with some makernote switches of exiftool, but I didn't get it.

So keep this problem in mind, if you try to interprete makernotes from this camera, using JPEG files.

============================================================

3) RD-3000 makernotes format

Looking deeper at the right address, there is a header, like the makernotes of many other brands, which can be used as a strong identifier for this cam, because I didn't see it in any other makernotes:

Code Select Expand

Fixed content for TIFF is: 4d 30 32 33 53 54 52 54 00 00 00 82 00 00
Fixed content for JPEG is: 4d 30 32 33 53 54 52 54 00 00 00 00 00 92

Usage: 4d 30 32 33 53 54 52 54... "M023STRT" (Makernotes ASCII Start ID)

At the end of the 500 Bytes there is always an ending delimiter for this makernote:

Code Select Expand

Fixed content is: 00 00 00 00 45 4e 44 20
This is true for both JPEG and TIFF images.

Usage: ...00 00 00 00 45 4e 44 20 "END " (Makernotes ASCII End ID)

============================================================

So far my first results. In a few weeks I will have the spare time to check further for the content of these 500 Bytes, testing with different - but similar - images.

Does this make sense? Comments and help are always welcome.
Thank you

Valentino

Modified 2010-01-03: added info to RD-3000

[Phil Harvey]

Hi Valentino,

This is very interesting.  I will try to answer your questions.

Is there any way to rebuild the right makernotes address for JPEG files, without using some special hex editor tools? I tried with some makernote switches of exiftool, but I didn't get it.

ExifTool will only fix the offsets of entries within a makernotes IFD, not offsets within the ExifIFD.  An error in the ExifIFD is a major error from ExifTool's point of view.

Any fixing of the makernote offset in the ExifIFD would have to be done by specialized code.

I would be interested to hear if you discover any useful information in these maker notes.  In the JPEG image it looks like a 10-byte header followed by about 36 big-endian 4-byte integers (which may contain camera settings), followed by some other stuff.  I don't have a TIFF sample -- if you could send one to me it would be great, my mail is philharvey66 at gmail.com - thanks.

- Phil

[dct]

Hi Phil

Just sent an email with two examples (TIFF/JPEG) with two different fix focus lenses, hence four files.
See the readme.txt for my test scenario settings.

Kind regards,
Valentino

Modified 2011-01-04: added test scenario - sent to Phil - inline for discussion tracking

Test scenario:

Test situation, described with similar tags of 0x927c, using MakerNoteMinolta, as of
http://gvsoft.homedns.org/exif/makernote-minolta-tag0x0001.html and
http://www.exiftool.org/TagNames/Minolta.html#CameraSettings

Values here are absolute camera values of the RD-3000 during test scenes, just for information,
not yet verified against the image content or the formulas and tables of GVsoft or Exiftool.

File identifier for the 2x2 test sets:

DHA017.JPG and DHA0018.TIF
Lens ::= V 50 mm Macro

DHA019.JPG and DHA0020.TIF
Lens ::= V 17 mm Wide

Camera settings are nearly the same for all four files. Unknown tags are omitted.
N/A ::= function not available on RD-3000.

Common camera settings:

01 ExposureMode ::=       Manual
02 FlashMode ::=      Wireless
03 WhiteBalance ::=      Flash
04 MinoltaImageSize ::=      1984x1360
05 MinoltaQuality ::=       Super Fine (for TIFF) / Fine (for JPEG)
06 DriveMode ::=      Single
07 MeteringMode ::=      Multi-segment
08 ISO ::=         200
09 ExposureTime ::=       1/60 (for Macro Lenses) / 1/45 (for Wide Lenses)    
10 FNumber ::=         6.7    
11 MacroMode ::=      N/A
12 DigitalZoom ::=      N/A
13 ExposureCompensation ::=   0    
14 BracketStep ::=       N/A
16 IntervalLength ::=       N/A
17 IntervalNumber ::=      N/A
18 FocalLength ::=      50 mm (for Macro Lenses) / 17 mm (for Wide Lenses)
19 FocusDistance ::=      ~1000 mm
20 FlashFired ::=      Yes

23 MaxAperture ::=      6.7 (of the camera)

31 Saturation ::=      N/A
32 Contrast ::=         N/A
33 Sharpness ::=      N/A
34 SubjectProgram ::=      N/A
35 FlashExposureComp ::=   0
36 ISOSetting ::=      200 (not the same entry as #8?)

41 ColorFilter ::=      N/A
42 BWFilter ::=         N/A    
43 InternalFlash ::=      N/A

48 FocusMode ::=      MF
49 FocusArea ::=      normal

Additional camera information, for all the four files:

RD-3000 was powered with batteries
Wireless Remote Flash Controller mounted on hot shoe and set to control channel #1
Flash 3500xi set apart as wireless slave on channel #1

[Phil Harvey]

Thanks Valentino,

I'll take a look at them tomorrow morning.

- Phil

[Phil Harvey]

I looked at the files but haven't done a detailed analysis, but I think with a large enough sample set there may be some useful information that can be extracted.

It is funny how bad the Minolta programmers were.  The focal length is written as 17/17 for the 17mm lens, and 50/50 for the 50mm lens, which both evaluate to 1.0 mm!  Doh.  Quirks like this in the standard EXIF are very annoying because I don't like to add manufacturer-specific patches to the standard EXIF code.  (But adding specific patches for maker notes tags is expected of course.)

- Phil

[dct]

I looked at the files but haven't done a detailed analysis, but I think with a large enough sample set there may be some useful information that can be extracted.

For shure! As stated before, I will try an exposure series (manual bracketing) with minimal configuration and setting changes to give a larger base for in-depth investigation.

It is funny how bad the Minolta programmers were.  The focal length is written as 17/17 for the 17mm lens, and 50/50 for the 50mm lens, which both evaluate to 1.0 mm!  Doh.

Yes, this is the really annoying downside. On the other hand, I figured out that the RD-3000 firmware writes always (example)

Code Select Expand

Exif:FocalLength (0x920a) :  920a 0005 0000 0001 0000 01b6

Usage:
920a      37500 (FocalLenght Tag ID)
0005      rational64u (Type of coding)
0000 0001 1 (number of entries)
0000 01b6 (offset)

At 0x01b6 we find:

Code Select Expand

FocalLenght Value 00 00  00 50 00 00 00 f0

Usage in RD-3000 is:
00 00 00 50 80 (minimum focal lenght of used lens in mm)
00 00 00 f0 240 (maximum focal lenght of used lens in mm)

Which is part of the new Exif:LensSpecification tag (0xa432 = 42034) in exif 2.23! 

Let us see, what other funny things we will find out next...

Valentino