Steen Bay
Veteran Member
Is it just the S3, or is it all 'compacts' that doesn't have a AA-filter? How is that possible without creating a lot of problems with moiré / aliasing?
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Are smaller pixels bad for diffraction? NO!!!!!!!!!
- Thread starter Lee Jay
- Start date
The S3's lens is a 6-72mm. I used it at the long end. I used a 70-200 on the SLRs at the wide end.
--
Lee Jay
(see profile for equipment)
All that I know of (I suspect the micro 4/3 and Sigma DP's are the exception).
The lenses act as the AA filter by not quite being sharp enough. The sensors generally lack an IR filter too, with a lens element serving that purpose.
--
Lee Jay
(see profile for equipment)
That would be tough for me since I don't have sensors of the same size with different pixel density. Plus, sensor size doesn't matter for this.
Maybe I'll do it again by simulation with all shots from the same camera. That would be easy.
--
Lee Jay
(see profile for equipment)
rhlpetrus
Forum Pro
It only depends on aperture used and how much you magnify image for view. If you use same magnification for viewing image (assuming both sensor have enough resolution for that), results should be same regarding diffraction, just like for oof points (used for DoF calculations).
--
Renato.
http://www.flickr.com/photos/rhlpedrosa/
OnExposure member
http://www.onexposure.net/
Good shooting and good luck
(after Ed Murrow)
--
Renato.
http://www.flickr.com/photos/rhlpedrosa/
OnExposure member
http://www.onexposure.net/
Good shooting and good luck
(after Ed Murrow)
But see, that's not quite true, even for DoF calculations. The assumption with DOF calculations is that all the blur is due to OOF, and none is due to pixelation (sampling), i.e. CoC > > pixel size. While this is often a good assumption (CoC=0.030mm for 5DII producing an 8x10 print for normal viewing distance while pixels are 0.0064), it's obviously never perfect since CoC is never infinity and pixel size is never zero.
The same is true here. Smaller pixels actually reduce the total blur even when diffraction-limited because the total blur is a combination of that due to diffraction and that due to sampling. Reduce the sampling blur and you get less total blur even when diffraction-blur dominates. That's my point - smaller pixels make blur smaller (not bigger, as some would imply) even when diffraction-limited with the larger pixels.
--
Lee Jay
(see profile for equipment)
Peace On Earth
Leading Member
I am dizzier the more I read about this ... just shoot, not happy shoot again :d
All you need to know is, getting a 7D won't hurt your performance due to diffraction compared with other 1.6-crop cameras.
--
Lee Jay
(see profile for equipment)
SayersWeb
Senior Member
I've tracked your posts for many years and have great trust in what you say.
So the question I am left with is... what about the 7D will hurt the performance?
Vivid1
Forum Enthusiast
Wellll - No AA is different than some AA isn't it
Actually it is. Even at small f-stops you still have a very thin pin sharp area even though the COC is smaller on either side of the focus plane. DOF is calculated for specific print size and When you enlarge very much like here, the relative COC enlarges as well causing even slight focus errors to become apparent. You need to be sure what you are seeing is diffraction effects and not focus errors.
Problem is if you upsize or downsize you get software induced characteristics: Bicubic smooth will give different results from bicubic for example which might show up at these enlargements.
Only if you don't enlarge or crop a lot more (when you've used small fstops) which many people believe the extra pixels will allow them to.
I agree that the sensor size does not directly affect the amount of diffraction on the image plane, but the design of lenses used on smaller sensors might.
Given that the f-number denotes the amount of light per unit area that hits the image plane, does it not follow that an EF-S lens of the same FL and f-number will have a smaller aperture diameter compared to an EF lens, due to the reduced image circle?
Therefore is it not possible to have 2 lenses of the same FL and f-number with different amounts of diffraction at the image plane?
Jeff Kingston
Forum Enthusiast
Actually you convinced me. There is indeed more detail that comes through with the smaller pixels, diffraction notwithstanding. I guess it shows that several elements come into play in determining the final result.
No. Every point on the sensor plane, within the image circle, receives light from every point within the physical aperture stop. The size of the image circle has no effect on how large the aperture opening needs to be in order to let in a certain intensity of light.
(Image from Wikipedia)
No, although differences in lens design (aperture shape) and mechanical variation due to tolerances can lead to small differences in the real world. But the f-stop is by far the most important factor, for a given wavelength of light.
Boby Six Killers
Member
This is more accurate ...
interesting..
and they all look blurry in the first place.
and F11 is at limits for 1.6 crop imo
and they all look blurry in the first place.
and F11 is at limits for 1.6 crop imo
When f-stops are high (for DOF or whatever) some people seem to get afraid of using cameras with small pixels. This is because they approach their "diffraction-limit" (at the pixel level) at lower (faster) f-stops.
THIS IS WRONG!!!! DON'T BE AFRAID OF SMALL PIXELS!!!!
Smaller pixels have a detail advantage at every f-stop .
The test below is with all images at the same final size, all at the same f-stop (except the last one - see next paragraph), all at the same focal length, and the smaller pixels clearly record more detail even though they are diffraction-limited. Yes, the additional detail is not nearly what the over 3-times smaller pixels would record if they weren't diffraction-limited, but they get more detail nevertheless.
Of course, if you want the most detail possible, use the smallest pixels at faster f-stops, as the last image shows. However, this isn't always possible due to DOF or slow-shutter-speed needs (use ND filters to avoid that last issue).
Can we now please stop saying crazy stuff like "the 7D can't be used past f8 while I can use the 5D all the way to f16" or other nonsense like that? Doubt it, but here's hoping!
P.S. Ignore noise in this test. ISOs were all at base, not equivalent ISOs.
--
Lee Jay
(see profile for equipment)
let's see - diminishing returns for lens. diminshing returns for enough light. etc..
On a full frame I'd go f16 or f19 for deep depth of field outside in the light. On a 1.6 crop, the most I'd go is F11 for deep depth of field. After that the sharpness declines for many lenses and one gets dimishing returns trying to go deeper on a crop.
I agree with your analysis that defraction shouldn't be the concern. The big concern becomes lenses and light
If you need more DOF than you can get with f11, why not go further? Sure, you'll lose a bit of sharpness on the in-focus sections, but not nearly as much as you'd lose on the OOF sections at f11. Besides, diffraction-limited DOF is format independent, so you can't do better on a single exposure no matter what you do anyway. This is especially applicable to macros.
--
Lee Jay
(see profile for equipment)
because..check f11 vs f16 on a crop camera here. MTF's go way down. F11 on a crop is already as deep as F 19 on a FF
http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=458&Camera=474&Sample=0&FLIComp=0&APIComp=7&LensComp=458&CameraComp=474&SampleComp=0&FLI=0&API=6
John Sheehy
Forum Pro
Larger pixels are one of the sources of irresolution. Most people fail to notice that, because they view images with less and larger pixels at smaller magnifications.
--
John
Keyboard shortcuts
- f
- Forum