9 replies to this topic

[Glasseye Snapper]

In the dSLR/housing forum Mike Veitch complaned about manual WB setting on the Canon 30D.

Being a Canon fan I sadly have to add another strike against Canon that I recently came across and that has relevance for the scientific use of UWphotography. The latest TTL electronics, including the one implemented by the Canon 20D and most other "recent" cameras, get focus distance information from the lens to estimate the proper strobe power. There is a special tag in the EXIF header to write this distance information into, and many cameras do, but Canon DOESN'T, not even in the 30D and other more recent ones

I complained, and based on Googling many others have, but the powers that be have decided that it is not a priority or whatever their reason is to hide that information. With the distance and focal length you can calculate the magnification and I could write a program to include an optional size bar in the image or allow the user to click on two points and request the real-world length between those two points. It could of course also tell you if you are at 1x or 0.78x magnification. Pretty useful in my opinion, especially if you want to use your camera as a scientific data collection instrument. Alas, no such luck with Canon. Has anyone heard about a way around this or has anyone used this information with another brand of camera?

Bart

[Alex_Mustard]

There are often third party EXIF readers, such as te ones under File Info in photoshop that can access much more of the EXIF than the manufacturer's defaults. Maybe you can find the info there?

Alex

[Paul Kay]

Try downsizing the file then opening it as a text document! Somewhere in there you might find the distance information if it written in to the EXIF data. It may simply be that Canon don't use the existing protocols which are after all, not written in stone. And it may be that 'distance' is too vague a term - ie is this from the image plane, lens PP, etc.?

BUT and it is a very big BUT, if you are dealing with underwater images, any recorded information will not represent actuality as the distance your lens is focussing at will not be the same as the actual (physical) distance due to refraction.

[Alex_Mustard]

BUT and it is a very big BUT, if you are dealing with underwater images, any recorded information will not represent actuality as the distance your lens is focussing at will not be the same as the actual (physical) distance due to refraction.

Although if you are using a flat port there is a simple linear relationship between the two.

Alex

[Paul Kay]

Although if you are using a flat port there is a simple linear relationship between the two.

Alex

Depends on the lens - if you are using a macro lens with internal focus this should be true (both the current Canon's have), but if the lens extends substantially as it is focussed (such as some Sigmas) things might not be so straightforward as you get closer.

[Alex_Mustard]

I see. The Nikon 60mm and old 105mm (non-VR) both extend during focusing. :-(

Alex

[Glasseye Snapper]

Thanks for your thoughts!

I have already downloaded a number of EXIF readers including a very impressive one that really has reverse engineered all there is to extract (I'm not at home so can't tell you the name of the free program). Basically, the problem is that the information is simply not there. I also tried the RAW files as I thought it was more likely to have vendor-specific details but again no luck. On the web several people confirm that the information isn't there and I finally asked Canon directly, mostly to let them know that they should make the information available, and they replied that the information isn't there. No reason given, but my message would be forwarded internally. I'm not too hopeful for a firmware update soon as others have contacted Canon about this before, but you never know what happens if they get sufficient requests or messages start popping up everywhere that Nikon is better than Canon because ...
(actually if memory serves me well I believe the 10D or some older model did include the distance information, so they do know how to do it)

Internal focusing lenses are actually more of a problem than the extending lenses. The latter keep their focal length constant and just change the distance from the sensor. With given sensor to subject distance (this is what is in the EXIF of Nikon and others) and known focal length it is simple to calculate magnification.
In contrast, internal focusing lenses don't change the distance from the sensor much, instead they reduce the focal length (in a sense they blur the concept of prime and zoom lenses). The Canon 100mm becomes a ~75mm at closest focus. The 180mm becomes ~120mm. This makes the lens more compact, with no externally moving parts, and it compensates for loss of light/aperture as you focus closer. However, if you thought you spent US$1800 for the Canon 180mm believing you really got a 180mm lens for your macro work then unfortunately it isn't so

Anyway, all of the lens and flat port effects (dome would be much trickier) can be modeled in the calculations with reasonable precision, it is the focus distance that is missing. Of course you would have to shoot with relatively shallow DOF to get the most accurate position but I think it would be a great tool to find out how big/small our underwater subjects really are.

Bart

[bvanant]

Thanks for your thoughts!

I have already downloaded a number of EXIF readers including a very impressive one that really has reverse engineered all there is to extract (I'm not at home so can't tell you the name of the free program). Basically, the problem is that the information is simply not there. I also tried the RAW files as I thought it was more likely to have vendor-specific details but again no luck. On the web several people confirm that the information isn't there and I finally asked Canon directly, mostly to let them know that they should make the information available, and they replied that the information isn't there. No reason given, but my message would be forwarded internally. I'm not too hopeful for a firmware update soon as others have contacted Canon about this before, but you never know what happens if they get sufficient requests or messages start popping up everywhere that Nikon is better than Canon because ...
(actually if memory serves me well I believe the 10D or some older model did include the distance information, so they do know how to do it)

Internal focusing lenses are actually more of a problem than the extending lenses. The latter keep their focal length constant and just change the distance from the sensor. With given sensor to subject distance (this is what is in the EXIF of Nikon and others) and known focal length it is simple to calculate magnification.
In contrast, internal focusing lenses don't change the distance from the sensor much, instead they reduce the focal length (in a sense they blur the concept of prime and zoom lenses). The Canon 100mm becomes a ~75mm at closest focus. The 180mm becomes ~120mm. This makes the lens more compact, with no externally moving parts, and it compensates for loss of light/aperture as you focus closer. However, if you thought you spent US$1800 for the Canon 180mm believing you really got a 180mm lens for your macro work then unfortunately it isn't so

Anyway, all of the lens and flat port effects (dome would be much trickier) can be modeled in the calculations with reasonable precision, it is the focus distance that is missing. Of course you would have to shoot with relatively shallow DOF to get the most accurate position but I think it would be a great tool to find out how big/small our underwater subjects really are.

Bart

Try e-mailing Chuck Westfall at Canon. He used to be quite active on Rob Gailbraith's fora and might be able to tell you how to get a workaround. If I remember (I am at the lab, not at home) there is distance data that is displayed on the Canon speedlights, so if you house one, you might be able to get the data from one. Also, if you are using one of the ttl converters from Matthias, you can ask him how to get the distance data
Bill

[Jolly]

Yes, it will be much trickier, not to say probably impossible with a dome glass. First of all you might miss the dome glass data already (exact curvature radius, thickness, material’s exact refraction index, etc.) and moreover the bend field of the subject a dome glass creates.
I have calculated some uw focus distance examples:
http://wetpixel.com/...ll/dome-theory/
But it’ll be a lot of work and not really precise with all those rough values. It was just meant as an example to give some idea of the relation. You should also consider measuring the particular water you dive in (different refraction index?). About the flat port you should find some way to include the distance between lens front and port glass. Let’s say 1cm with brand A and 4 cm with brand B where the refraction takes place. It’s quiet similar to this: 100mm 1:1 macro lens with +3 dioptre, what reproduction ratio? It’s not possible to say as long as you don’t include some lens data. I would suggest too many variables for calculating exact sizes of subjects for macro work. I don’t think it’s impossible. But I think you have to deal with many imprecise values and variables if you want to get some true scientific size values.

Just went thru 10D shots and I have distance information in EXIF data (to me they don't really look like precise values)except with some lenses that do not deliver distance information. Surprisingly, third party lenses like Sigma 18-50 f2.8 and Sigma 10-20 provide distance data to the camera in my shots. You have just convinced me to keep at least one 10D body as caliper ruler

I agree with Bill, I think there is some difference between the camera using distance information for E-TTL II metering and writing them into EXIF data.

Julian

[Glasseye Snapper]

Thanks Bill & Julian,

I'll try to get hold of Chuck Westfall, nothing beats being able to talk directly to an insider. Also good to know the 10D indeed does include the distance information. It seems they are intentionally hiding the distance information, perhaps it is so imprecise that they'd rather not show it. Canon seems to be rather conservative that way. However, I expect they have encoders on the lens motor so they should know exactly where the lens is positioned and thus what the focus distance is. I don't think decoding the TTL signal is helpful for me, even if the information is in there, because I need an easy solution to get the data stored together with the image. E.g. my PhD doesn't depend on it, it is just a hobby project.

If this doesn't work out I can always take a zoom lens, set it to manual focus and leave it at a fixed focus distance and use the zoom to fill the frame with a fish. The zoom's focal lens will be reported in EXIF so I can callibrate lens magnification underwater with a ruler as function of the zoom setting (there is actually an NASA/Smithsonian paper on doing this). This may also work better behind a dome.

I'll let you know if I get any further with this.

Bart