jray
Offline
• • •
Upload & Sell: Off
|
 p.2 #20 · Megapixels -> lenses limiting factor? | |
Regardless of lens quality, there's another fly in the ointment which is why I think the current pixel density of the D60, 10D and 300D, are getting close to the point where we will start to see diminishing returns. IMO, 20-24MP or so on a full frame 35mm format sensor should run us up close to that unmovable wall, physics. Keep in mind a full frame 35mm sensor with the same pixel density as a 10D is already at 16MP.
Even the best of lenses will have their optic quality overshadowed at smaller apertures by diffraction. This is bound to physics, and there is no way out. Since the amount of diffraction is a fixed angular figure based upon wavelength, images will have an increasing percentage of their image circle comprised of diffracted light as apertures become smaller. This reduces image sharpness and ultimately limits the maximum practical pixel density of a given sensor design.
As digital sensor pixel densities increase, the number of usable working aperture values will decrease until that nasty little critter Airy's Disc (Sir George Airy) bites us on the butt (below). Couple this with reduced DOF ('depth of focus' at the sensor end) at larger aperture values coupled with AF error (my rant below), and we're getting close to the point of limited returns in regards to the detail that can be captured by a 35mm sensor, regardless of the lens optics.
I'll stay away from a bunch of math here since; 1) Although I can deal with it when need be, I don't really like to, and 2) It's the quickest way to have a thread become tedious and boring. I'll just list a few links to sites with pertinent data:
Airy pattern/Rayleigh Criterion: http://molecularexpressions.com/primer/java/imageformation/rayleighdisks/
Airy's disc: http://hyperphysics.phy-astr.gsu.edu/hbase/phyopt/cirapp.html#c1
The Rayleigh Criterion: http://hyperphysics.phy-astr.gsu.edu/hbase/phyopt/raylei.html#c1
Main Diffraction Menu: http://hyperphysics.phy-astr.gsu.edu/hbase/phyopt/diffracon.html#c1
Light & Vision Menu: http://hyperphysics.phy-astr.gsu.edu/hbase/ligcon.html#c1
My AF rant:
First I want to give my own 'expanded' definition of a 'blur circle'. I think of it as the aggregate physical size of any virtual image point that is increased in size by 'various' factors such as diffraction, DOF, or AF error.
In addition to diffraction limiting, the accuracy of the auto-focus system should be factored into the decision to couple a higher density sensor to a given camera design. The higher the sensor pixel density, the more accurate the AF system needs to be in order to insure the overall blur circle is about half the size or smaller than the area of an individual photosite. If we consult Nyquist, it should actually be less than half the size to insure more than 2x sample frequency, but this is the world of analog where photosites are square (more area) and blur circles are round, so 2x should be enough 
The D60, 10D, and now the 300D all have pixels densities of ~18,355 pixels per square millimeter, or ~135.5 per lineal millimeter, and they all have AF systems that do not properly support such a pixel density IMO. Although the 10D and 300D AF systems are more sensitive and faster than those on the D60 and D30, their accuracy and repeatability just don't cut it as far as I'm concerned, especially at larger apertures. Although this is only my opinion, I base it upon a number of images I've taken with both a D30 and 10D where the 10D captured no more detail than a manually focused D30 shot. Such D30 images when upsampled to 3072 x 2048 for comparison contained every bit as much image detail, and sometimes more if the 10D's AF was off by much. This tells me the 10D AF system has far too much error for its sensor's pixel density. Although AF error will always have an effect on the overall image sharpness, it should never result in captured image detail that is the same, or less, than that of a sensor of significantly lower MP IMO.
The focusing system on Canon cameras comes in two varieties; High-Precision, and Normal-Precision. As per Chuck Westfall, Director/Technical Information Dept. - Canon U.S.A.:
"A "normal-precision" AF sensor produces focusing accuracy within the depth of focus for the lens' maximum aperture. A high-precision" AF sensor produces focusing accuracy within 1/3 the depth of focus for the lens' maximum aperture."
While 'Normal-Precision' may have been acceptable for film (not IMO), it's already having a direct effect upon the captured image detail of the 10D and 300D as far as I'm concerned. To be honest, I've never agreed with classic DOF calculations for critical focus and have always used figures for the next largest aperture when possible as my maximum DOF criteria. Increased sensor pixel densities will only exacerbate the issue. Minimal math follows. Boo... Hiss...
Assuming the 10D/300D/D60 sensor size of 22.7mm x 15.1mm relates to the effective area imaging area, then the pitch between individual photosites is ~.007mm. If the effective imaging area is actually less, then the pitch between photosites is even smaller.
At f/11, the blur circle caused by diffraction (Airy's Disc) is almost equal in size to the pixel pitch, but the blur circle caused by the 10D's 'acceptable' AF error is even larger at times. This shouldn't be. When AF error on a given camera produces more detrimental effects to the image than diffraction, it's time to hit the drawing board. Canon will need to update their AF system before offering pro-sumer 8MP DSLRs. Canon's Normal-Precision specs just wont cut it for higher densities. I won't buy an 8MP DSLR that only offers more image detail when manually focused. I already have one in the 10D, why do I need another one? End of AF rant....
Between the effects of DOF and AF error on one end, and diffraction limiting on the other, we're getting close to the end of the MP war for the 35mm format IMO. Now it's time to concentrate on greater dynamic range, larger color gamut, lower noise, etc. These are the things that push digital far past what film ever offered, and will issue in a new era in photography for all of us.
|
