Diffraction

[David Kilpatrick]

See this:

http://forums.dpreview.com/forums/read. ... e=34310470

I have had a full day (well, three hours allowing for travel and lunch!) specialist one to one briefing with Canon for reviewing the 1D MkIV.

Part of this included discussion, and info, on the size of the pixel WITHIN the grid. The pixel pitch may be X, but if there's a load of wasted space (not gapless) the actual pitch per sensel can be MUCH smaller. Gapless microlens arrays make the pixel pitch pretty much equal to the stated figure. Non-gapless microlenses make it smaller, and small sensel wells under the microlenses reduce it further.

The actual real size of each receptor could be half that of the 'pixel pitch' calculation or stated figure.

Depending on the sensor design, there could be a half stop or even a stop difference in the diffraction limit aperture for apparently similar resolutions. As a guide, those cameras which achieve exceptional high ISO are likely to be using larger actual sensel areas and gapless microlenses. They will generally be less prone to diffraction loss than earlier designs.

David

[Greg Beetham]

Ke? David what constitutes a 'load of wasted space' I'm not following.....
Greg

[David Kilpatrick]

Imagine the sensor is a checkboard. Normal pixels (sensels) are the area you would see a chess pawn, sitting on each square. The rest of the square is wiring and silicon substrate. The new Canon sensels are the size of draughts pieces, and cover more of the square. Any part of the checkerboard grid not covered is the wasted space.

David

[Greg Beetham]

Ah so! catching maybe three times as many photons as a regular sensel at a given ambient without raising the sensitivity, very clever.
Greg

[alphaomega]

So does that explanation potentially mean that the new Sony 12/14Mp sensors in A500/550 have larger sensels than the 12Mp sensor in the A700? What has this discussion got to do with "Diffraction"?
If the sensel size determines relative higher ISO performance, why do CMOS sensors then typically produce better high ISO performance compared with CCD? Does a CMOS sensor provide more space on the board for sensels or is it inherent in the CMOS type sensor quality wise rather than individual sensel size?

[David Kilpatrick]

I don't know - but Sony claim reduced gap sensels on the A900 sensor, I imagine they have carried the same through to the A500/550. CMOS is not inherently less noisy or more sensitive than CCD, but the way the sensors are built can dissipate heat faster - which is why all the new video DSLRs are CMOS, not a single attempt to make video on a CCD sensor. The circuits may be more efficient, require lower power, etc, giving more scope for efficient A to D processing - including on-sensor.

David

[Greg Beetham]

I sort of assumed it was to do with the diffraction limit of a reasonably good lens, (as well), the thinking is, as you pack ever more sensels onto a chip eventually you reach the stage that the sensel size is going to be as small or even smaller than the diffracion spot of a good lens.
Assuming that is correct of course, and some info I've read suggest that it is, if the space between sensels is utilized by making them larger then you can ostensibly keep (or increase) the number of sensels and still remain at a size that's equal to or larger than the diffraction spot. I don't know what the limit is exactly for FF, but one site suggested that for APS-C the limit (for the regular layout, I think it was) was around 18MP, beyond that there is no advantage in increasing the number of sensels, (or so they say).
Greg

[alphaomega]

Interesting point David makes above regarding the possible reason why CMOS sensors do better at high ISO

CMOS is not inherently less noisy or more sensitive than CCD, but the way the sensors are built can dissipate heat faster

So maybe there is something in the point made in my DSLR thread "Freeze your sensor for top high ISO results".

I was just thinking after posting the other day that accomplished photographers such as David or perhaps other members may have taken high ISO photographs of the same spot at different times of year (warmth/frost) using the same camera and setting. It would be interesting to know if they could spot a difference in ISO performance.

[harvey]

I sort of assumed it was to do with the diffraction limit of a reasonably good lens, (as well), the thinking is, as you pack ever more sensels onto a chip eventually you reach the stage that the sensel size is going to be as small or even smaller than the diffracion spot of a good lens.
Assuming that is correct of course, and some info I've read suggest that it is,...

You don't seem convinced.

I don't see anything new here from film with diffraction limiting the usefulness of small f-stops related to how good the resolution of the film was.

Maybe someone will invent an amazing lens (with negative refractive index components) that can beat the diffraction limiit.

Harvey

[Greg Beetham]

I think the atmosphere is the ultimate limiting factor, it has been accused of being a very messy substance to try to receive a point source through optically, (speaking telescopes that is), but photographically speaking it get's very convoluted, with airy disk size, (diffraction), circle of confusion, f-stop employed, sensel size...here is an interesting article with an interractive table that you can set.
Greg

http://www.cambridgeincolour.com/tutori ... graphy.htm

[harvey]

I think the atmosphere is the ultimate limiting factor, it has been accused of being a very messy substance to try to receive a point source through optically, (speaking telescopes that is),

It would limit our photo opportunities to only take photos on the top of mountains near those telescopes!

...but photographically speaking it get's very convoluted, with airy disk size, (diffraction), circle of confusion, f-stop employed, sensel size...here is an interesting article with an interractive table that you can set.
Greg

http://www.cambridgeincolour.com/tutori ... graphy.htm

Yes, you are right. I'm also aware of that site. The other very useful and relevant article is
this one

Harvey

[Greg Beetham]

Point source transmission through an optical system is the ultimate test for any optical assembly, telescope or camera lens, (diffraction limited), but there is still another problem for lens type optical systems...achromatism.
I agree both those articles are very good, although in view of David's news, an update might be in order to incorporate the latest sensor types.
Greg