APS-C cameras potentially creating harsh bokeh?

[Spotmatic]

Most of us know the famous swirly bokeh of for instance the Helios-40 lens. The "cat-eyes" are caused by optical vignetting: with a wide open lens specular highlights in OOF areas are not round anymore.

Now, I just viewed this picture from a fellow Pentaxian. This was made with a Pentax 50/1.2 lens on a Pentax K10D (APS-C sensor):



As can be seen the brightest stars are rectangular instead of round. According to the maker it's a problem of this combination of camera and lens. A K10D has a small mirror box (smaller than that of a FF camera) and the 50/1.2 has a very big rear element. What happens now is that the pixels in the corners cannot see the whole back element, causing vignetting.

As can be seen here, the mirror box fits inside the bayonet mount completely:


The back element of the 50/1.2:


Knowing this, isn't it also possible that the bokeh of the given combination is affected, making the bokeh more harsh? And what does this mean for the speed of the mentioned f/1.2 lens as less light is able to reach the sensor? Wouldn't this defy the point of using a f/1.2 lens (at least on APS-C)?

I'm interested in your thoughts on this subject. I'm not an optics designer but I want to understand the whole concept and the possible drawbacks for bokeh lovers

[poilu]

APS-C is just a crop of the digital FF, bokeh should be the same
vignetting and bad border is certainly affected on FF
Now on film, bokeh is completely different; more smoother and nicer
the square stars are probably more a bayer demosaic algorithm problem

[Spotmatic]

[martinsmith99]

[Spotmatic]

[Dco]

Interesting, I've never seen such a thing when I used the 50mm f1.2...
I'm not so sure about the results. Isn't the photo focused at infinity? In that case, OOF areas shouldn't exist, I believe that the stars should be in focus...
Another thing, that photo is a long exposure (30 sec.) so it's normal the stars appear a bit blurry because of their movements.
I'd like to see a 100% crop of one of the biggest stars, and also another sample, with a faster shutter speed, to compare...
One last question. Was it a raw o jpeg image?

[no-X]

[Spotmatic]

Jiri,

Photographing stars is a quite demanding application. So lens abberations and things like vignetting will show up very quickly. Regarding the vignetting of the mirror box the following link (provided to me by the maker of the star photo in my first post) is a quite interesting read. This example (in the link) is on a Pentax LX so it's understandable that on APS-C the vignetting is even more severe.

http://plg.komkon.org/k50_12/wo1.html

So my question still remains: will the bokeh on APS-C be affected too? I'm inclined to think it will.

Edit: the maker also wanted to add the following. Astrophotography is indeed quite demanding for the quality of a lens. Some stars are much brighter than others. The brighter ones only need a second to be exposed properly on the sensor/film; the less bright ones need an exposure of many seconds/minutes but this causes the brightest stars to be overexposed.
If a good lens is able to focus 99% of the star's light in one dot, and the other 1% in a larger disc, then a 1000-fold overexposure makes the larger disc to show up. Normal photography is of course less demanding.

Note from Spotmatic: the above explanation is correct for objects in focus. But what about OOF objects, that's what interests me?

[no-X]

Yes, vignetting can be present, but it will never change shape of in focus objects. Some forms of optical vignetiting can change shape of OOF objects - most known exaple are ovoid OOF lights, but it will never affect the shape of in focus objects.

Imagine, how would it affect usage of P6 lenses on M4/3 cameras. Or old large format lenses on APS-C cameras, where the crop factor isn't 1.5x, but 4x or more. But these combinations don't embody these (or similar) artifacts, even with such significant differences.

[Nesster]

Perhaps rather than the size of the mirror box - after all, the rear of the lens fits in there - it's a matter of the angle of light hitting the individual sensor elements. My understanding is that a digital sensor does not gracefully accept light except at specific angles... and now that I think of it, perhaps there's some veiling flare possible from light that hits at a too low angle?

[greg]

I believe there was a thread earlier regarding the use of a choke - similar to a fixed aperture - placed between the rear of the lens and the mirror to prevent reflection problems. It cuts down on the size of the image circle hitting the inside of the camera body and was rectangular in shape. I seem to remember it was from a Russian based forum.

In the above picture it shows some really regular shaped squares. Are they the same in other images from the same camera? If so, they could simply be dead (frozen) pixels - a common thing. Take a picture with the body cover/lens cap in place, this will make the dead pixels stand out, or at least it was the method used with the old Fuji 6800z. My copy of that camera had over 50 dead pixels and was considered a low number at the time. Actually, it isn't a 'dead' pixel, some are colored, it depended in which state the pixel was frozen.

[naplam]

I'd say the mirror box is vignetting; the rear element is so big there isn't a clear path from the extreme borders of the lens to the sensor.

[sichko]

[Spotmatic]

[LucisPictor]

I don't think that bokeh highlighta appear rectangular because of the smaller mirror box. At least I can't confirm this experience.
I also have a lens with a huge rear lens, the Nikkor-S.C 1.2/55, and it does not cause rectangular highlights in my 40D.

[Spotmatic]

So this possible problem is purely academical?

[poilu]

[Ballu]

[Minolfan]

If the rectangulair highlights only appear in this kind of pictures, why bother?