Canon Exif FocusDistance Upper/Lower

[apap]

I'm using a 40D and Lightroom and Dxo for RAW conversion. DxO need focus distance for lens correction but if you use a 40D the input has to be manual on the other hand if you have a 400D and a supported lens DxO will read it automatically.
Also the Canon DPP has a lens correction function that I think is based on the maker note items FocusDistanceUpper and Lower.
I have roughly investigated the behaviour for actual focus distance in relation to the FDU & FDL values by taking a number of shots with a EF-S 10-22 @2mm F5,6 the result can be seen in the attached picture.
The FDL follows the actual focus distance reasonably well from the start but makes jumps at around 1,3 to 1,5 and 1,8 to 2 m actual focus distance. The move - jump -stay constant -jump -stay constant behavior matches the movements of the focus distance slider in DPP lenscorrection so I'm convinced this is what is used.
Changing F-stop will move the position of the jumps but the values for FDU & FDL will be the same in the "stay constant" region.
Why have Canon designed it like this ??
Anyone that knows/or can guess the definitions for the FDU & FDL ?

BR/ Stefan

[Bob Atkins]

I've no idea what Canon are doing here, but one guess might be that aberration corrections change significantly for close focus, but once you get past a certain distance, aberrations stay pretty constant, so at those distances the EXIf data just reports some "long distance" data.

Why they don't put the actual focus distance in the EXIF data and then just use their software to decide what level of aberration correction is needed I don't know. They seem to have quantized diatances into a number of steps, presumably each one of which corresponds to a preset aberration correction function which is included in DPP.

I think some older bodies such as the D30 (not 30D) and D60 actually record focus distance in the EXIF data. Why Canon stopped doing this is a total mystery to me. I've seen reports that when asked, Canon more or less say "because that's the way it is" and offer no more reasonable explanation.


Here is a copy of a reply from Chuck Westfall of Canon USA to a similar question two years ago:

1) Most but not all current Canon EF lenses report distance data to the camera. The lens specification charts we've published since 2004 usually indicate which lenses support distance data. However, these are not comprehensive lists, because they typically do not contain information about discontinued lenses. I would be happy to provide a current listing, but I'm not sure if this forum provides a good way to publish it. Perhaps Asher K. or Michael T. could comment.

2) To the best of my knowledge, none of the current EOS digital SLRs records distance information in Exif data, even with lenses that support it. Most likely this is because our current software applications like DPP, RIT, etc., don't use it.

3) I understand that some 3rd-party software applications like DxO do use this information. Several users have requested that distance information should be restored to the Exif data generated by our cameras, in order to support these 3rd-party applications. We have forwarded these requests to Canon Inc. via our monthly reports, but so far there is no indication that these requests will be granted. If anything changes on this subject, I will let you know.

Obviously in the last two years DPP has started using distance info, but Canon still don't seem to be keen on putting actual distances into the EXIF data. Whether this is or may be related to patent issues I don't know.

[apap]

Thanks for the long answer !
When I first saw the curve I thought the second jump might be related to the hyperfocal distance. The hyperfocal distance if I may believe the DOFMaster calculator is 4,53 at 22mm/F5,6 so the first jump is at around 30% and the second at around 40%.  Maybe the intention is to divide into close, mid and far focus only revealing distance values for the close focus area. Please have a look at my data below:
0,3      0,5    0,75    1        1,3      1,5      1,8        2             2,5          3            4           5              6
0,35   0,58   1,03   1,39   2,24   6,48   6,48   655,35   655,35   655,35   655,35   655,35   655,35
0,29   0,49   0,83   1,03   1,39   2,24   2,24      6,48       6,48      6,48      6,48      6,48      6,48


/Stefan   

[Bob Atkins]

Seems resonable.

If you look at lens aberrations as a function of focus distance, they're pretty constant in the far focus range and only increase for near focus. There's something called the conjugate ratio which is the ratio of the focal length to the focus distance. From about 1:20 and upwards, lens performance is pretty constant. So for a 100mm lens, that would be from around 6ft to infinity. Only when you start focusing closer than about 1:20 do aberrations increase in some lenses - unless you have a lens with floating elements which move around to correct the close focusing aberrations.

I doubt it's directly tied to hyperfocal distance, other than by the fact that the hyperfocal distance tends to correspond to fairly close focus for short and normal focal length lenses. There's really no direct theoretical connection between HFD and the level of aberrations.

Perhaps Canon just have correction functions for near, mid and far distances (probably defined by conjugate ratios), because that's all that's really needed, plus they only need three sets of correction data for each lens rather than large data tables or attempts at parametric fitting equations over the whole distance range. I suspect their data gives a "ballpark" correction. I've found that I can improve the correction slighly by making manual adjustments in DPP. If I had 2000 images to correct though, I'd just go with the Canon defaults!

I'd still like to see Canon put actual focus distance data in the EXIF info though. Unless there is some sort of patent issue, I can't see any reason for them not to. They have the distance info from most lenses, so why not write it out?