F stop Metadata

[isaac]

Advertisement (gone after registration)

Hello Dave!

what tools are you using to suss EXIF data?

I am using a 'homewritten' tool.

...hmmm, my EXIF reading gives a decimal number at your entry 4 (0x0310)

How'd you derive the lens ID?

A double shift right (division by 4) gives the offset (binary representation of the lens code) in the lens table of the M8 firmware (M8-????.upd)

Again, I get a number resembling the Kelvin temp for your item 9 (0x0321), even when using preset, or manual WB settings... do you get words, or a number?

There is a non-zero value only if Auto-WB is selected, which is an approach to the color temperature in Kelvin. If there is a value of zero, than item 4 (0x310) has the 'normal' LightSource (TIFF/EXIF/DNG) meaning. So far my observations...

My findings are that your item 10 (0x0325) is a boolean for whether IR/UV is active, and the camera CAN determine the lens+correction algo, and...

Thank, this is a useable hint, I will go deeper into.

...your item 12 (0x0331) also seems a boolean, which in my earliest DNGs(12 Nov 06) was always "0", yet now is always "1", even if I disable lens detection.

From what I have 'researched', items 10 and above have changed in the history of firmware. But, thanks, I will also have a deeper look into this.

I have only one coded lens (2.8/28 ASPH) and another which is not coded, but also recognized (old 2.8/90).

Your item 15 (0x0340) would appear to be the equivalent to the "ImageUniqueID"

Yes, of course. I have to add this in my software, but it is so obvious...

While I suspect there might be some data in the file to determine the aperture, I haven't been able to find it.

In my former posting, there were ALL bytes of MakerNote mentioned. There is no more (hidden) information in DNG.

The most interesting to me, lately, is stuffing the value of 0x0321 into the EXIF "UserComment" field so I have the Kelvin temp in the TIFF or JPEG result of my scripts+"WhiteBalance" to record how WB was determined and truely "AsShot".

Very good idea, but in this case, you have to recreate the whole DNG file because all tag related data is between the thumbnails data and the raw data (or you alter only the pointers and put the new information at the end of the file).

The only thing, which I have not already detected, is the 'tint' value (green/magenta shift), maybe it is the magic value of item 0x0320, who knows?

And there are also three (alsways constant) values in items 0x311,0x312 and 0x313 of unknown meaning to me. They are equal (c4, cc, d4) in a lot of DNG I have from third party, but I have one (of unknown source) where this values are b8, c0 and c8. Maybe sensor specific calibration?

[DaveSee]

Hello Dave!

 

I am using a 'homewritten' tool.

 

Hiho!

 

I'm using Phil Harvey's `exiftool` program, and looking at the DNG and JPEG data only.

A double shift right (division by 4) gives the offset (binary representation of the lens code) in the lens table of the M8 firmware (M8-????.upd)

 

Super! This is the magic I sought... I've a DNG with this field's value of 113... and yes, it was a 28.25 used to make it... and in another, 147... the 75/2... Danke!

There is a non-zero value only if Auto-WB is selected, which is an approach to the color temperature in Kelvin. If there is a value of zero, than item 4 (0x310) has the 'normal' LightSource (TIFF/EXIF/DNG) meaning. So far my observations...

 

For 0x0321 I'm getting what would seem to be the Kelvin WB value, even in Manual mode.

From what I have 'researched', items 10 and above have changed in the history of firmware. But, thanks, I will also have a deeper look into this.

I have only one coded lens (2.8/28 ASPH) and another which is not coded, but also recognized (old 2.8/90).

....

 

Very good idea(regarding field 0x0321 and WhiteBalance written to "UserComment"), but in this case, you have to recreate the whole DNG file because all tag related data is between the thumbnails data and the raw data (or you alter only the pointers and put the new information at the end of the file).

Actually, I grok the "WhiteBalance" and field "Unknown:0x0321" from the DNG /before/ I run the data through de-mosaicing, gamma and TIFF or JPEG creation so when writing the TIFF or JPEG, I write additional EXIF, XMP and some IPTC into the output image files. I leave the DNG alone! First rule, do not alter the source file! I also re-write the "Orientation" and image dimensions meta-data because Adobe's products are easily broken if given a file with orientation other than "Normal".

The only thing, which I have not already detected, is the 'tint' value (green/magenta shift), maybe it is the magic value of item 0x0320, who knows?

And there are also three (alsways constant) values in items 0x311,0x312 and 0x313 of unknown meaning to me. They are equal (c4, cc, d4) in a lot of DNG I have from third party, but I have one (of unknown source) where this values are b8, c0 and c8. Maybe sensor specific calibration?

Yes, over the near year's worth of DNGs, these fields have had a constant--and so too 0x0304, for me, no flash--that changes with change in firmware only.

 

I've some coded, some not, and shot some "A" mode with AWB, some "normal" mode with Manual K settings... fields 0x033{0,2,3} have always read "0" for me despite the code in my face

 

rgds,

Dave

[sreidvt]

Sean,

 

I checked your M8 Updates article and there is no table. In fact you state that the aperture information is not recorded in the EXIF file. Was your memory slipping or did you actually have the data somewhere else?

 

I'm not sure what table you're thinking of. Which data? The info. is in that article in a section called "The M8's External Metering Cell". Obviously, we've learned that the aperture estimate can be found by a knowledgeable person but it does not appear in the normal EXIF data and was not intended to. Might you be mixing me up with Scott?

 

Cheers,

 

Sean