Fast lenses & microlenses

ZoneV · Expire: 2011-12-02

Mark Dubovoy wrote on Luminous Landscap an interessting finding of DXOmark:
http://www.luminous-landscape.com/essays/an_open_letter_to_the_major_camera_manufacturers.shtml

Fast lenses are not fully supported from the camera sensors. Light rays that enter the sensor at wider angels are not fully absorbed from the sensistive areas.
That is not realy new - most who has worked with microlenses know that their quantum efficency depends on the light incident angle. But new is the big difference between the DSLR cameras - and that cameras seems to boost the ISO speed to camouflage this sensitifity loose.

Realy tragic seems their assumption, that the depth of field could be larger with the newer sensors / smaller.

enzodm · Joined: 11 Sep 2010 Posts: 350 Location: Italy

Interesting, although the open letter seems to blame CMOS sensors even if in the chart also CCD sensors are represented.

ZoneV · Expire: 2011-12-02

Yes, I think the article is not very accurate.
I suppose this effect belongs to the microlenses - but I found that not in the document.

As far as I know, the marginal rays are not complete lost, but less effective in terms of quantum efficency of the sensor.
Take a look here on page 18 as an example:
http://www.kodak.com/global/plugins/acrobat/en/business/ISS/datasheet/fullframe/KAF-8300LongSpec.pdf

Interesting is the difference between horizontal and vertical angle.

At the moment I think, that the light damping of the marginal rays could get a better smother bokeh Smile

But I am glad to have the old 5D.

The ISO boost is tested and verified.

Anu · Joined: 14 Apr 2009 Posts: 879

LL is a site where technical facts aren't always too accurate. - even DPR might be more solid Wink

This is yet another sample of such an article. For example he seems a bit clueless about why medium format cameras use CCDs - they are available, CMOS-sensors are not (CMOS-version would be an expensive stitchjob) - light hitting the sensor at an angle is actually a smaller problem with medium format due to the flange focal distance.

cooltouch · Joined: 15 Jan 2009 Posts: 9096 Location: Houston, Texas

I would want to read a white paper or two from the camera manufacturers on this same subject before I arrived at a conclusion. It seems to me that LL is leaping to a conclusion just a tad too quickly -- before all the facts are in so to speak.

Seems to me also that if the camera makers were really "gaming the system" wrt ISO increases, a substantial number of users would have detected the additional noise before now. My Canon DSLR is one of those that were measured on the charts, and it being the cheezy entry-level DSLR that it is, I can tell you with all certainty that it is easy for me to detect the added noise whenever I move the ISO from 100. It just doesn't handle higher ISOs that well. So, if this were really going on, I woud
for sure be seeing an increase in noise when I'm shooting with my f/1.2 and f/1.4 lenses wide open -- but I see nothing of the sort, in fact.

I have pics I can share, and let you be the judge regarding noise, far as that goes.

Anu · Joined: 14 Apr 2009 Posts: 879

There is discussion on this at the LL forums, so I don't need to reinvent the wheel here. There were one or two posters who pretty much managed to make the article look rather silly and even contradict it's own evidence Wink

Of course it is an exercise in the readers data mining capability to find the posts that follow logic and evidence, so I will leave that to you Wink

(Summary: just disregard the LL open letter - it is paranoide "they are cheating us" crap.)

ZoneV · Expire: 2011-12-02

There is a simple test - I think it should work. Don't know all automatic in a camera:
Take a fast Canon EF lens, for example Canon EF 50/1.2, take a photo (iris full open) of - as an example - white wall. Full manual without AF. Illumination constant - not old flurosecent light (or take realy long exposure time)
Now release the lens mount button, unscrew the lens 20-30° - the electronics didn´t relaize which lens is on the cameraq now. Mak a second photo with the same parameters.

Some users in a German forum did this for me - and the normal image was a bit brighter.
Better only look in the center, in case the camera makes some vignetting correction.

I think the theory of this problem is true. I do not know how different leses are designed regarding telecenticity on the image side. With telecentricity this is no problem at all. Without it is.

ZoneV · Expire: 2011-12-02

Anu · Joined: 14 Apr 2009 Posts: 879

ludoo · Expire: 2011-12-05

What I find interesting in the published data is that the 5d (old) sensor is the one where light loss is minimal. It might be totally unrelated, but the only thing I really liked about my 5D was its sensor: it has a certain quality which I cannot really describe.

I tried explaining it just yesterday to my boss (a Canon crop user) and the best thing I came out with is that the 5D images are more "luminous", less flat especially in contrasty/dark situations than all other cameras I've had. In comparison, my K-x has much better colours but images look more 2d and flatter, kind of like the difference between a beautiful watercolour (Kx) and a shiny oil painting (5D).

bob955i · Joined: 15 Apr 2007 Posts: 2495

Anu · Joined: 14 Apr 2009 Posts: 879

bob955i · Joined: 15 Apr 2007 Posts: 2495

Anu · Joined: 14 Apr 2009 Posts: 879

bob955i · Joined: 15 Apr 2007 Posts: 2495

You are being pedantic and straying perilously close to trolling.

Kindly refrain from the latter or I will lock this thread.

LucisPictor · Expire: 2013-12-03

visualopsins · Expire: 2025-04-11

I agree that 5D rendering seems very similar to film. I also agree there are significant differences between sensor and film, the linearity of sensor, the non-linearity of film, sensor color metrics and film emulsion response to color, etc., which make the statement wildly inaccurate.

No doubt 5D1 is a great camera! I think it is the first digital camera to provide results similar to (nearly as good as) film.

I remember reading a few years ago ads from a camera manufacturer showing micro-lens modifications that minimize light loss due to light arriving at an angle.

The problem occurs when using lenses with large rear elements -- light from the circumference of the rear element hits the micro-lens array at a steep angle; some of that light gets lost before reaching the sensor.