fredmiranda.com Login
This topic is in the 2015 archive. You can read it and like its posts here. It is read only.

Archive 2015Moire? 5Dsr.

  
 
Sneakyracer
Offline
Details
p.3 #1Moire? 5Dsr.
dhphoto wrote:
Sneakyracer wrote
gdanmitchell wroteMy take on this is that there is not yet universal agreement as to whether the advantages/disadvantages of AA filtering at high sensor resolution is, on balance, an improvement or not — given that there are both potential advantages and disadvantages, and that the impact of these pluses and minuses is at least partly affected by what you shoot.

Before a company makes a commitment to go one way (no AA filtering) or the other (AA filtering) consistently, the AA-filter-cancelation approach allows them to make the two versions with the greatest consistency among the design and manufacture of the two bodies.

It looks like Nikon must have decided that the non-AA version was mostly preferred and good enough to simply not need AA filtering. Sony must have come to the same conclusion a bit sooner. Canon seems to be on the fence, which isn't completely surprising as this is their first high-MP sensor camera that does away (virtually) with AA filtering.

I think that both the potential resolution advantages of eliminating AA filtering in high resolution sensors and the potential aliasing disadvantages will turn out to be much less than we might think. If so, at least with high MP cameras, I suspect that Canon will also eventually simply do away with the AA filter system entirely — just like Nikon did in the D810, which seems to be well-received from this perspective.

Dan

Jeff Fenske wroteInteresting that Sony didn't include an AA filter at all in the new a7R II, just like what Nikon did with the D810, while Canon did Nikon's D800E canceling method.

DPR writes this about their Studio Scene results:

http://www.dpreview.com/articles/7945517371/opinion-did-sony-just-do-the-impossible?

No AA filter

For one, there's no anti-aliasing filter at all, rather than the two birefringent filters in the 'anti-aliasing canceling' setup we saw in the D800E and, now, the 5DS R. Although the improvement may only be subtle, we saw slight improvements in micro-contrast when shooting our studio scene with a D810 vs. a D800E with the same exact lens (samples to be published shortly).
Glad Canon gives one the choice to do away with the AA filter right now. IMHO the New Canon has such a high pixel density (more than the Nikons and Sonys obviously) that an AA filter is not really needed. For Video it is good to have but for stills, no way IMHO.
Except that Canon haven't done away with the filter, they've added yet another one to the 5DSR to cancel the AA filter out


Yes, to keep the cover glass thickness on the sensor the same.

http://www.lensrentals.com/blog/2014/06/sensor-stack-thickness-when-does-it-matter
Jun 25, 2015 at 02:06 AM
gdanmitchell
Offline
Details
p.3 #2Moire? 5Dsr.
dhphoto wrote:
Except that Canon haven't done away with the filter, they've added yet another one to the 5DSR to cancel the AA filter out


That is getting closer to the truth, though it implies that there are more layers of filtering in the "R" model than in the non-R or other cameras with AA filters.

The normal AA filter has two layers that work together to provide the anti-aliasing effect. The "R" model also has two layers, but they are oriented in such a way that one cancels the effect of the other.

So, technically speaking, Canon has not "added a layer," but instead has rotated the orientation of one of the layers by 90 degrees.

To be somewhat accurate I refer to three sorts of cameras:

1. The camera has no AA filter – e.g. D810, Fujifilm X-trans, etc
2. The camera has AA filtering — e.g. the 5Ds and most other DSLRs
3. The camera has not AA filtering — e.g. the 5Ds R and the older D 800E

Dan
Jun 25, 2015 at 06:56 AM
dhphoto
Offline
Details
p.3 #3Moire? 5Dsr.
gdanmitchell wrote:
dhphoto wroteExcept that Canon haven't done away with the filter, they've added yet another one to the 5DSR to cancel the AA filter out
That is getting closer to the truth, though it implies that there are more layers of filtering in the "R" model than in the non-R or other cameras with AA filters.

The normal AA filter has two layers that work together to provide the anti-aliasing effect. The "R" model also has two layers, but they are oriented in such a way that one cancels the effect of the other.

So, technically speaking, Canon has not "added a layer," but instead has rotated the orientation of one of the layers by 90 degrees.

To be somewhat accurate I refer to three sorts of
...Show more →

Hairs suitably split GDan

The point is, to say the 5DSR has no AA filter is wrong
Jun 25, 2015 at 07:02 AM
gdanmitchell
Offline
Details
p.3 #4Moire? 5Dsr.
dhphoto wrote:
Hairs suitably split GDan

The point is, to say the 5DSR has no AA filter is wrong


On both points we can agree! :-)

Dan

(The reason the distinction between "filter removed" and the cancellation method is significant has an effect on camera design. I think we are unlikely to see a company offer two versions of a camera where one has an AA filter and the other has removed the physical filter, since that requires different physical configuration of the internals of the camera. The AA filter cancellation method used by Canon and formerly used by Nikon on the D800 and D800E resolves that issue and allow the company to offer the option more economically... but it does make one wonder. Note that today Nikon offers only one model, the D810, and it has no filter. In the end, I think that a lot of these things are pretty inconsequential when it comes to the photographic result...)

Edited on Jun 25, 2015 at 07:15 AM · View previous versions
Jun 25, 2015 at 07:11 AM
alundeb
Offline
Details
p.3 #5Moire? 5Dsr.
Great, the difference between the 5DS and the 5DS R is said to be impossible to see in print, but we must beware not to say that a self cancelling AA filter (effectively no filter) is the same as no filter
Jun 25, 2015 at 07:14 AM
dmcphoto
Offline
Details
p.3 #6Moire? 5Dsr.
If you batch process lots of photos and can't or don't spend time to individually optimize each one, then the 5DS is for you. If you make a living photographing screen doors or other very moire-prone items like fabric, then the 5DS is for you. If you photograph landscapes or other items that aren't usually prone to moire and go through your RAW files to select the best ones for continued processing and then spend considerable time optimizing each one, the 5DSR is for you.

Because of its lower resolution the Nikon D800E and D810 models are slightly more prone to moire than the 5DSR. Nikon and their customers were so worried about moire that they stopped making a D8XX model with AA filters. Moire will definitely happen sometimes, and if you individually process each image then getting rid of it is easy. Lightroom and Adobe Camera RAW have tools specifically designed to deal with moire and they work very well. Moire is not a big problem unless you need to turn out lots of images quickly and can't, or don't want to, spend the time processing them individually. FWIW, you'll find moire in some photos taken with any camera, it's just not as severe and doesn't happen as often. You'll virtually never see moire at smaller apertures because diffraction effectively acts as an AA filter. How small the aperture needs to be depends on the DLA of your particular sensor, but I think f/11 would certainly do it on a 5DSR.

Jun 25, 2015 at 08:15 AM
Monito
Offline
Details
p.3 #7Moire? 5Dsr.
Don't neglect the green aliasing that shows up in blonde hair. Moiré tools don't get rid of that.
Jun 25, 2015 at 08:22 AM
dhphoto
Offline
Details
p.3 #8Moire? 5Dsr.
Monito wrote:
Don't neglect the green aliasing that shows up in blonde hair. Moiré tools don't get rid of that.


Agreed.

As has already been said the LR moire tool (I haven't used a different one) does usually work but not always, it depends on a lot of factors
Jun 25, 2015 at 08:36 AM
Ulff
Offline
Details
p.3 #9Moire? 5Dsr.
dmcphoto wrote:
If you batch process lots of photos and can't or don't spend time to individually optimize each one, then the 5DS is for you. If you make a living photographing screen doors or other very moire-prone items like fabric, then the 5DS is for you. If you photograph landscapes or other items that aren't usually prone to moire and go through your RAW files to select the best ones for continued processing and then spend considerable time optimizing each one, the 5DSR is for you.

Because of its lower resolution the Nikon D800E and D810 models are slightly more prone
...Show more →
Thanks a lot, Dean, for this summary. I'm debating S vs. SR, too, and haven't made a decision yet. I usually don't do batch processing of images, instead I work a lot of hours on a few selected images. I understand, that it is in most cases easy to get rid of moire effects. What I don't know yet: is it possible to achieve the same level of sharpness and resolution with the S (vs. the SR) when good selective sharpening (selective because this should prevent from additional noise in areas that don't need to be sharpened) techniques are being used? Or is it simply impossible to get the same resolution and sharpness with the S irrespective of post processing techniques?

Jun 25, 2015 at 09:52 AM
dhphoto
Offline
Details
p.3 #10Moire? 5Dsr.
Ulff wrote:
dmcphoto wroteIf you batch process lots of photos and can't or don't spend time to individually optimize each one, then the 5DS is for you. If you make a living photographing screen doors or other very moire-prone items like fabric, then the 5DS is for you. If you photograph landscapes or other items that aren't usually prone to moire and go through your RAW files to select the best ones for continued processing and then spend considerable time optimizing each one, the 5DSR is for you.

Because of its lower resolution the Nikon D800E and D810 models are slightly more prone to moire than the 5DSR. Nikon and their customers were so worried about moire that they stopped making a D8XX model with AA filters. Moire will definitely happen sometimes, and if you individually process each image then getting rid of it is easy. Lightroom and Adobe Camera RAW have tools specifically designed to deal with moire and they work very well. Moire is not a big problem unless you need to turn out lots of images quickly and can't, or don't want to, spend the time processing them individually. FWIW, you'll find moire in some photos taken with any camera, it's just not as severe and doesn't happen as often. You'll virtually never see moire at smaller apertures because diffraction effectively acts as an AA filter. How small the aperture needs to be depends on the DLA of your particular sensor, but I think f/11 would certainly do it on a 5DSR.
Thanks a lot, Dean, for this summary. I'm debating S vs. SR, too, and haven't made a decision yet. I usually don't do batch processing of images, instead I work a lot of hours on a few selected images. I understand, that it is in most cases easy to get rid of moire effects. What I don't know yet: is it possible to achieve the same level of sharpness and resolution with the S (vs. the SR) when good selective sharpening (selective because this should prevent from additional noise in areas that don't need to be sharpened) techniques are being
...Show more →

This is what I was talking about yesterday.

Depending on what you are going to use the files for, I'd say it doesn't matter because if you are printing them the differences won't show up because any differences between the 5Ds and 5DsR will be evened out by the print media, slight changes in inkjet shadow detail etc, if you are only looking at 100% crops then potentially they (just) might but of course with the R you do risk the possibility of moire when you least expect it
Jun 25, 2015 at 10:00 AM
dmcphoto
Offline
Details
p.3 #11Moire? 5Dsr.
Ulff wrote:
Thanks a lot, Dean, for this summary. I'm debating S vs. SR, too, and haven't made a decision yet. I usually don't do batch processing of images, instead I work a lot of hours on a few selected images. I understand, that it is in most cases easy to get rid of moire effects. What I don't know yet: is it possible to achieve the same level of sharpness and resolution with the S (vs. the SR) when good selective sharpening (selective because this should prevent from additional noise in areas that don't need to be sharpened) techniques are being
...Show more →

In theory it would be impossible to get exactly the same resolution if the photos were shot with an excellent lens at an aperture wider than the DLA, which is f/6.7. Deconvolution sharpening of the S photos would get them the closest to the R photos. Whether one could see the difference depends on factors like print size and eyesight. As with anything, if you magnify it enough any difference will become visible.

OTOH, using a 5DS and 5DSR with a lens aperture smaller than the DLA should produce images with no significant difference at all, and that will also eliminate (or depending on exact aperture greatly reduce) moire.

I wouldn't lose any sleep over this because the differences won't be dramatic, and at sizes like 16x24 inches they'd be invisible.
Jun 25, 2015 at 10:35 AM
alundeb
Offline
Details
Jun 25, 2015 at 10:51 AM
dmcphoto
Offline
Details
p.3 #13Moire? 5Dsr.
alundeb wrote:
TDP has posted resolution tests of both the 5DS and 5DS R at various apertures:

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=458&Camera=980&Sample=0&FLI=0&API=6&LensComp=458&CameraComp=979&SampleComp=0&FLIComp=0&APIComp=6


You can see a difference between the cameras at wider apertures that's gone by f/11, at least to my eyes. It looks like the maximum difference is at f/4 and f/5.6. That's pretty much as expected since f/11 is well beyond the camera's DLA of f/6.7, and lens aberrations take a toll at the widest apertures.

TDP is a fantastic resource!
Jun 25, 2015 at 11:23 AM
alundeb
Offline
Details
p.3 #14Moire? 5Dsr.
dmcphoto wrote:
alundeb wroteTDP has posted resolution tests of both the 5DS and 5DS R at various apertures:

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=458&Camera=980&Sample=0&FLI=0&API=6&LensComp=458&CameraComp=979&SampleComp=0&FLIComp=0&APIComp=6
You can see a difference between the cameras at wider apertures that's gone by f/11, at least to my eyes. It looks like the maximum difference is at f/4 and f/5.6. That's pretty much as expected since f/11 is well beyond the camera's DLA of f/6.7, and lens aberrations take a toll at the widest apertures.

TDP is a fantastic resource!


We can agree about that fantastic resource!

To my eyes, the difference between the S and the S R at f/11 is just as much as at f/5.6 There is additional blur at f/11, but the relative difference, in amount of blur is the same.

The 5DS is not diffraction limited at f/6.7. The resoution limit because of diffraction ( Rayleigh criterion) is about f/13. At f/6.7 there is just some blur, more blur than at f/5.6 and less than at f/8.

See here and the following discussion:
https://www.fredmiranda.com/forum/topic/1371725/2#13071869
Jun 25, 2015 at 11:30 AM
dmcphoto
Offline
Details
p.3 #15Moire? 5Dsr.
alundeb wrote:
dmcphoto wrote
alundeb wroteTDP has posted resolution tests of both the 5DS and 5DS R at various apertures:

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=458&Camera=980&Sample=0&FLI=0&API=6&LensComp=458&CameraComp=979&SampleComp=0&FLIComp=0&APIComp=6
You can see a difference between the cameras at wider apertures that's gone by f/11, at least to my eyes. It looks like the maximum difference is at f/4 and f/5.6. That's pretty much as expected since f/11 is well beyond the camera's DLA of f/6.7, and lens aberrations take a toll at the widest apertures.

TDP is a fantastic resource!
We can agree about that fantastic resource!

To my eyes, the difference between the S and the S R at f/11 is just as much as at f/5.6 There is additional blur at f/11, but the relative difference, in amount of blur is the same.

The 5DS is not diffraction limited at f/6.7. The resoution limit because of diffraction ( Rayleigh criterion) is about f/13. At f/6.7 there is just some blur, more blur than at f/5.6 and less than at f/8.

See here and the following discussion:
https://www.fredmiranda.com/forum/topic/1371725/2#13071869


Thanks for the link! I was going by the DLA posted in the 5DS review on TDP at http://www.the-digital-picture.com/Reviews/Canon-EOS-5Ds.aspx
Jun 25, 2015 at 11:42 AM
dmcphoto
Offline
Details
Jun 25, 2015 at 02:36 PM
Ulff
Offline
Details
p.3 #17Moire? 5Dsr.
Thanks a lot for the answers, very helpful! In the meantime I extensively compared the raw-samples that dpreview provides from both cameras. I tried to make look both files as similar as possible in post-processing. As mentioned here before, the sample image has a lot of moiré.
Moiré-reduction for the SR worked remarkably well, although for this specific image it was very time consuming to correct all occurrences of moiré. In areas where moiré was most pronounced in the SR, there was also some moiré in the S.
Sharpness of the S was also remarkably easy to add in order to look like the SR. After sharpness was applied carefully, there was almost no difference between the cameras anymore. Sharpening added some noise as a side effect, but for 95% of the image this wasn't noticeably more than the R had.
In summary, I came to the conclusion, that the difference between both cameras is a lot less important for me than I initially thought. As someone who likes to make the whole sharpening procedure only at the end of a long processing (and who is a bit too lazy to correct every moiré), maybe I'm better suited with the S, given the additionally lower price...
Jun 25, 2015 at 03:58 PM
alundeb
Offline
Details
p.3 #18Moire? 5Dsr.
dmcphoto wrote:
http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm


That is also a good resource, but regarding the so-called diffraction limit, they make one big mistake and some simplifications.

1) The diffraction blur does end at the borders of the Airy disc. It extends infintely in a wave pattern, and affects low frequency contrast as Toothwalker points out here:

https://www.fredmiranda.com/forum/topic/1371725/2#13072143

2) The diameter of the airy disc is defined by the occurence of the first minimum in the airy pattern. Regarding resolution, there is nothing special that happens when that diameter equals the size of a pixel. It does not mean that the blur ends inside one pixel, just think of different alignment, and also the further extending Airy pattern. Also, the Airy disc is gradual, so you continue to resolve even when the airy disc extends more than one pixel.

3) There is a limit to resolution because of diffraction. That is when two airy discs are so close together that it is no longer possible to separate the peaks of those discs from each other. That happens soon after the Rayleigh criterion:

http://www.normankoren.com/Tutorials/MTF1A.html

4) Bhe big mistake CiC does, is that they both illustrate and mention the Rayleigh criterion, but in their calculators they use the diameter of the Airy disc. Using the diameter of the airy disc in their calcluators may feel intuitively right, but rgearding a diffraction "limit", it is totally meaningless.

5) Bayer sensors samples green at 70% of the pixel pitch and red and blue at only 50% of the pixel pitch. This means that to resolve all detail for all colors, we need to double the pixel density.
Jun 26, 2015 at 12:20 AM
Jeff Fenske
Offline
Details
p.3 #19Moire? 5Dsr.
dmcphoto wrote:
alundeb wroteTDP has posted resolution tests of both the 5DS and 5DS R at various apertures:

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=458&Camera=980&Sample=0&FLI=0&API=6&LensComp=458&CameraComp=979&SampleComp=0&FLIComp=0&APIComp=6
You can see a difference between the cameras at wider apertures that's gone by f/11, at least to my eyes. It looks like the maximum difference is at f/4 and f/5.6. That's pretty much as expected since f/11 is well beyond the camera's DLA of f/6.7, and lens aberrations take a toll at the widest apertures.

TDP is a fantastic resource!


I also agree that TDP is a *fantastic* resource. But it could be even more fantastic if he allowed comments, which could have refuted the Cambridge in Colour myth that TDP spreads, that smaller pixels cause diffraction sooner.

The diffraction we see is from the lens.

These are the results of a slower, Canon 200mm f/2.8 lens, which shows the lens is in diffraction at f/8. And f/11 is much worse: http://www.photozone.de/canon_eos_ff/580-canon_200_28_ff?start=1
Jun 26, 2015 at 01:36 AM
Monito
Offline
Details
p.3 #20Moire? 5Dsr.
Small pixels don't cause diffraction "sooner". They enable you to see it "sooner".
Jun 26, 2015 at 03:08 AM
FM ForumsCanon ForumJoin Upload & Sell
  
 

You are not logged inLog inRegister