Thanks for letting me know I wasn’t alone in this experience. You do good work.

Here is a theory:

Apart from the lens type the lens reports to the camera its focal length range, the currently set focal length and the maximal aperture.

While using IDs of existing Canon lenses (simply because Canon would not allocate their own) third party lenses may report their actual focal length range and aperture.

Using the above data the camera firmware can distinguish the third party lens from the Canon lens it “mimics”(this is what ExifTool does I believe). Once it has done this it is logical to not apply illumination correction in order to avoid the described problem.

The discrepancies may be due to the database in the firmware of different cameras being different (more or less complete, more or less wrong). I can not imagine Canon putting too much effort into this – it is a great opportunity to say “Use only Canon lenses, we do not guarantee results with anything else”.

Here is an example of the camera firmware logic that I am suggesting (built from your data above):

“A lens that identifies itself as Canon 85 f/1.2 L but reports a focal range 17-50 and f/2.8 must be a “Sigma 17-50 f/2.8 OS”

Lens 3rd. party …

With Canon 5DIII Peripheral illumination turned on.

Sigma 24-60mm f2.8 EX DG
exifTool reports …
Lens Type … Canon 28-70mm f2.8L or Sigma or Tamron lens
Lens ID …….. Sigma 24-70mm f2.8 EX
Peripheral illumination state off, value 0.

Sigma 17-35mm f2.8-4.0 EX DG HSM (later of these models)
exifTool reports …
Lens type … Canon EF 28-80mm f2.8-4L or Sigma lens
Lens ID ……. 100% correct ID of this lens
Peripheral illumination state off, value 0.

Sigma 70-200mm f2.8 APO EX HSM (film era version)
exifTool reports …
Lens type … Canon 28-80mm f2.8-4L or Sigma lens
Lens ID ……. Sigma APO 50-150 f2.8 [II] EX DG HSM + 1.4x
Peripheral illumination state off, value 0.

Thanks for all the hard work. I rarely use third party lenses with my Canon body so this may not pertain much to me. But there was one other thing on the PIC that I had encountered when I did some testing a few weekends ago. I upgraded by 5D Classic to a MKII and found some odd behavior with PIC, focusing and the self-timer function combined together. I did this testing on a 5D and then the Mk II. In case you find yourself without a thing to explore or look into, I would welcome if you wanted to look into the issue that I extensively blogged about here: http://www.rohangupte.com/blog/files/upgrading-canon-5d-to-canon-5d-mark-ii.php

Thanks for all that you do – I don’t say this often to you…

Regards,
Rohan

I’m not sure I understand the question, but in each case we tried several copies of each lens on a couple of copies of the camera. Every Tamron 70-300 or 17-50 identify themselves as 28-70s. On some bodies they get automatic peripheral illumination correction (every copy of the 70-300 placed on any copy of the 5D II, for example). Which is why we don’t think lens ID has anything to do with peripheral illumination correction.

Roger

That makes perfect sense. As a rental house, though, we aren’t able to do that so having the data available will help our renters who get a camera and third-party lens know what to expect. Or really, at least know they need to check before shooting 1,000 jpgs and finding out their all not quite right.

Thanks for the input — I think every third-party lens owner should be doing exactly what you describe.

Roger