Canon 17mm TSE Unusual Optical Characteristics

[Ghibby]

Hi all,

A few months ago I noticed some odd optical traits in my 17mm TSE lens.  These are actually rather positive ones too. That is that the lens produces 16 point sun stars from point light sources at certain apertures.  Conventional thinking would say this is impossible, the lens has an 8 bladed iris so there can only be 8 point sun stars. It is only iris's with an odd number of blades that produce double the number of spikes on a sun star, eg 9 bladed apertures will create 18 point stars, 7 will create 14 and so on.

My finding shows that the 17mm TSE Lens creates nice 16 point stars between f4.5 and f8.0, by the time you hit f9.0 the expected 8 point sun star is rendered with just some traces remaining of the 16 point star.  I went through my archives and have observed this occurring consistently with the lens.  To prove it to myself over the Christmas holidays I photographed my Christmas tree with its bright lights at a series of apertures.  The results support the findings above. See images, at f5.6, 8.0, 9 and 16

I am confident that I have also found the reason for this phenomenon. The blades of the iris are rounded but not in the conventional sense. Canon has used a complex curve made of 2 discreet curves in a bid (I assume) to keep the iris more circular over a greater range of settings.  The transitions between theses curves on each blade is just sharp enough to create the extra spike per blade on the sun stars for some apertures. By the time the lens is stopped to f9 and smaller apertures the transition section of the blade is no longer on the periphery of the iris and so the extra sun star spikes disappear.

To me this is great news, the lens is at its performance peak (un-shifted) from f5.6 to f8.0 with regard to micro contrast and minimal diffraction softening on the EOS 5Ds and this is exactly the range that the beautiful 16 point sun stars are created.  A bonus for night time architectural photography!  To get nice sharp corners when shifted heavily, (say 8mm and more) you do need f11 and occasionally f16 but with really nicely set up tilts most images can successfully be made in the f8 to f9 range even with big shifts. 

I would be really interested to see if anyone else out there using the 17 TSE has come across this trait.  I have not analysed the 24mm Mk2 TSE thoroughly for this characteristic but I think there may well be a trace of it visible at some settings.  It is not as clearly rendered or obvious as with the 17mm lens though. 

Ben

[marc aurel]

Hi Ben,

yes, I have noticed the unconventional behaviour of the TS-E 17mm too. And I am sure you are right about the unusal form of the iris as the reason for this.
The sun stars of lenses can be quite different, and they do change with different apertures. Last year I tested the lenses I use for architecture. I placed a flashlight in my corridor and took a series of photos at different apertures. Not the perfect setup because the flashlight is not a perfect punctual light source (and there's a slight reflection on the floor). But it is similar to what you face when you photograph artificial light sources at night. Click on the thumbnails for full size.

As you can see from the screenshots the 9-bladed 24-70mm f2.8 L II is quite different from all the others. Canon is quite proud of it. But I prefer the look of the the TS-E 17 and even more of the TS-E 24. They look more natural to me. The zoom is so extreme with its endless thin blade-like beams. I once had a customer who was quite upset about this because it dominated the image too much.

The images taken wide open are much closer to what the human eye perceives (because our iris is round - like the iris of a lens wide open). Sadly image quality and depth of field is not sufficient wide open (at least for most purposes).

Regards -
Marc

[Ghibby]

Hi Marc,

Your analysis is very useful, shows with great clarity what the sunstars are actually doing.  Interesting that on stars with more than one spike per blade that alternating spikes are either hazed or quite clear.  I can only assume this is caused by the way the spike is created, ie mid blade or blade junction. 

Personally I prefer the sunstars from the 17mm to the 24mm, the ones from the 24-70 are amazing though, for me easily the best. I think its a photographer thing, I have had similar comments on a few images from clients when the sunstars are very dominant.  The stars from the Sigma 35 and 50mm Art lenses appear to have similar definition to the Canon 24-70 II, I like the rendering from these too. 

Interesting stuff indeed.

Regards
Ben

[Shiftworker]

That's not the only strange characteristic of this lens... I've found it has a curved focus plane which gets more pronounced the more you stop the lens down which means that the corners and edges can actual get worse as you stop down. The best way to use this lens I have found is to focus about midway between center and edge ( in stopped down mode) and don't stop down past F8. It's not in the same league as the 24mm TS-EII but it's the best 17mm shift lens out there ie it's the only 17mm shift lens out there....