5Ds vs 5DsR - the true resolution difference?

[nemophoto]

FINALLY, Canon brings out a higher res camera. Only been waiting three or four years. While I'm a pretty diehard 1D user (owned every camera in the series and currently the 1Dx), and have never really taken to the 5D3, this new camera will fit the bill for some of the work I do. (Personally, I don't care if Sony may theoretically have a greater DR, the DR of Canon's cameras is more than sufficient for my needs. Plus, after a day using a friend's Sony A7r, I came away hating the camera and the EVF).

So, my question is (and it might be answered by Nikon users who have used both the D800 and D800E), is there really much of a difference in real resolution? As primarily a fashion shooter, moire has always been a concern and issue. My 1Ds3 showed very little, but I think had a fairly strong AA filter. My 1Dx shows a bit more (especially in blue jeans and certain other types of fabrics), because it has a weaker AA filter to tweak the res upwards. (My own evaluation, nothing I ever read.) At 50MP, the camera will have far more res than anything I've had to date, so it's really a curiosity question.

[Bart_van_der_Wolf]

So, my question is (and it might be answered by Nikon users who have used both the D800 and D800E), is there really much of a difference in real resolution? As primarily a fashion shooter, moire has always been a concern and issue. My 1Ds3 showed very little, but I think had a fairly strong AA filter. My 1Dx shows a bit more (especially in blue jeans and certain other types of fabrics), because it has a weaker AA filter to tweak the res upwards. (My own evaluation, nothing I ever read.) At 50MP, the camera will have far more res than anything I've had to date, so it's really a curiosity question.

Hi,

I think you'll love the image quality of the 5DS (not the 5DS R if used at wide to medium apertures). Higher resolution will significantly reduce the chance of moiré, especially with a good Raw converter like Capture One (hopefully it will support the 5DS).

I can generate moiré on my 1Ds3 when I use a good lens and perfect focus at optimal apertures (f/4-f5.6 have often the highest resolution until f/7.1 at the narrower end when diffraction take over), but the 5DS should do significantly better.

The denser sampling of the 5DS' 4.14 micron pitch sensels will start to become visually affected by diffraction from f/4.5 and narrower, so around the optimum of the lens. Looks like a perfect match and predictable increase of diffraction smoothing (and aliasing reduction as one goes narrower). Resolution will become totally diffraction limited from f/14 and narrower, virtually (in theory only blue wavelengths could but they contribute little to luminance resolution) no aliasing can be generated due to diffraction cutting resolution at the sensor's Nyquist frequency, not even on the 5DS R version.

Cheers,
Bart

[Phil Indeblanc]

I would agree with Bart, and many positive things about the 5Ds. One thing I can't agree with is coming to a conclusion about a camera and its system within one day(or even a week if you're not prepared to relearn things). That is just not possible. 10 years ago I was forced into Canon from Nikon, and I was really fighting with the menu system and placement of a number of things, and with time I just had to slow down and accept it being a different approach to acheiving the same thing. As long as you are open minded and don't think that it is either your way(the way you are accustomed to to ), or the highway, you will discover somethings that are a bit more logical, and somethings that are not.

Since these menus do so much they reach over areas they can do well for a certain number of things. As soon as they do something else, they will struggle, and it simply goes both ways. Often this adapts to shooting style and shooting subject/situation you find yourself in more often than not.  I say this with very little experience in the Sony menu, yet working with a number of systems, each has its approach, and you have to just stop and read to relearn things that you wish you didn't have to.

[John Koerner]

Even though I wish they had 4K for UHD video, they have a pretty nice commercial explaining the internal body/vibration-dampening improvements and overall concept:

http://youtu.be/mzHv-Ej_l8w

[LKaven]

Even though I wish they had 4K for UHD video, they have a pretty nice commercial explaining the internal body/vibration-dampening improvements and overall concept:

I have to like Samsung's approach to 4K, which involves a full readout of the 28MP sensor for each video frame.  This kind of bandwidth and throughput is what is necessary, along with fabrication processes that reduce thermal buildup.  Supersampling makes a beautiful video frame.  The next step involves delivering at least 10 bits and at least 4:2:2 in a raw format.  But I don't think we're going to see this kind of throughput from a Canon sensor with this many pixels on it for the time being. 

[torger]

In fact, I think that there has never been a case for dropping the AA filter. The sole reason is that razor cut pixel peep has a special attraction to us photographers that triggers our want-feelings and it sells well. When it comes to real image quality having an AA filter is always better, unless you shoot at very small apertures (like f/14) where the AA filter won't make any contribution.

You gain extremely little in resolution, say 50 megapixel without AA resolves like a 53 megapixel camera with AA (Bart can probably say something more scientific about that), and do you really need that tiny bit extra in resolution and at the same time increase the risk for various aliasing artifacts?

When an image is published in some form it nearly always involves some scaling up or scaling down, and there can also be image transformations in post-processing like rotations and distortion correction. An aliased input won't do you any good in those cases.

That said the aliasing on the 5Ds R is probably not going to be that bad in practice due to the very high pixel count, so I think most will choose that anyway because people can't resist that pixel peep joy

[dwswager]

In fact, I think that there has never been a case for dropping the AA filter. The sole reason is that razor cut pixel peep has a special attraction to us photographers that triggers our want-feelings and it sells well. When it comes to real image quality having an AA filter is always better, unless you shoot at very small apertures (like f/14) where the AA filter won't make any contribution.

You gain extremely little in resolution, say 50 megapixel without AA resolves like a 53 megapixel camera with AA (Bart can probably say something more scientific about that), and do you really need that tiny bit extra in resolution and at the same time increase the risk for various aliasing artifacts?

When an image is published in some form it nearly always involves some scaling up or scaling down, and there can also be image transformations in post-processing like rotations and distortion correction. An aliased input won't do you any good in those cases.

That said the aliasing on the 5Ds R is probably not going to be that bad in practice due to the very high pixel count, so I think most will choose that anyway because people can't resist that pixel peep joy

What people seem to forget is that the math describing these things all tend to be approximations with assumptions implied.  Once you get in the field where the assumptions may not apply the whole thing can fly out the window.  Nothing is ever as cut and dried as the lab measurements predict they will be.

The 5DsR apparently is similar to the D800e in that it has the OLPF and then filters the effect back out.  I shoot the D810 and the D7100 and neither has an OLPF in the filter stack.  I don't shoot fabrics and other other tightly patterned things that tend to produce moire and have not seen significant artifacts from either, though the D810 is somewhat better than the D7100.  I would wait till it gets into the hands of folks before purchasing any new technology.

[Bart_van_der_Wolf]

What people seem to forget is that the math describing these things all tend to be approximations with assumptions implied.

Hi,

I disagree, when assumptions are based on pure physics.

Things like aliasing are well understood by those versed in in DSP. The only variable is the quality of the optics, which can work like a very mild OLPF if focused well, and make aliasing impossible when defocused or when used with narrow apertures.

I've done an analysis (but not published all details) based on the known specifications (EXIFs also offer info). They show very interesting effects from the combination of sensel pitch and already measured lens performance at various apertures. Based on that, I expect a serious boost in image quality, despite the hardly improved DR. The quality gains are mostly related to an improved MTF.

Things like the separation specifics of the Bayer CFA are unknown, and therefore indeed speculation.

The fact that Canon have specifically addressed things like mirror vibration by reengineering the mechanism from spring to gear based, and vibration timeout setting intervals, looks promising. Practical tests will have to tell how effective it is.

Cheers,
Bart

[ErikKaffehr]

Hi,

I am shooting a 6.8 micron MFD on Hasselblad V. I do see a lot of aliasing issues on everyday subjects. It seems that the problem goes away at f/16, while at f/11 it is still very present. This is moiré on sea water surface:



Best regards
Erik

Hi,

I disagree, when assumptions are based on pure physics.

Things like aliasing are well understood by those versed in in DSP. The only variable is the quality of the optics, which can work like a very mild OLPF if focused well, and make aliasing impossible when defocused or when used with narrow apertures.

I've done an analysis (but not published all details) based on the known specifications (EXIFs also offer info). They show very interesting effects from the combination of sensel pitch and already measured lens performance at various apertures. Based on that, I expect a serious boost in image quality, despite the hardly improved DR. The quality gains are mostly related to an improved MTF.

Things like the separation specifics of the Bayer CFA are unknown, and therefore indeed speculation.

The fact that Canon have specifically addressed things like mirror vibration by reengineering the mechanism from spring to gear based, and vibration timeout setting intervals, looks promising. Practical tests will have to tell how effective it is.

Cheers,
Bart

[dwswager]

Hi,

I disagree, when assumptions are based on pure physics.

Things like aliasing are well understood by those versed in in DSP. The only variable is the quality of the optics, which can work like a very mild OLPF if focused well, and make aliasing impossible when defocused or when used with narrow apertures.

I've done an analysis (but not published all details) based on the known specifications (EXIFs also offer info). They show very interesting effects from the combination of sensel pitch and already measured lens performance at various apertures. Based on that, I expect a serious boost in image quality, despite the hardly improved DR. The quality gains are mostly related to an improved MTF.

Things like the separation specifics of the Bayer CFA are unknown, and therefore indeed speculation.

The fact that Canon have specifically addressed things like mirror vibration by reengineering the mechanism from spring to gear based, and vibration timeout setting intervals, looks promising. Practical tests will have to tell how effective it is.

Cheers,
Bart

We expect to see some gains from the increased MP and other refinements.  My point is, that while we can predict some of these, we won't actually know until empirical testing data comes in.  And as far as having 50MP goes, that is helpful if you are limited to a 22MP 5DmkII, but not compared to a 50MP A7RII or 50MP DX10 which will in all likelihood also carry DR and ISO performance gains over a 5DmkIII.

As to my actual comment that you quote, I ask:  "What is usually the 1st assumption when dealing with lenses in the field of optics?"  Hint:  One has never been manufactured!

[nemophoto]

Thanks for all the different perspectives. Thanks especially to Erik for posting the image from the Hassy. That's exactly the kind of thing I want avoid when I shoot. Back in the day when I shot with the original 1D, I frequently jumped through hoops to try to get rid of moire in garments. Back then, clients were even less forgiving because I was trying to transition them from film! My question was posed as curiosity, but for all practicality, I'd go with the 5Ds. I can't believe that in real world circumstances, the resolution of the 5DsR would be that significant over the "plain vanilla" 5Ds.

[dwswager]

Thanks for all the different perspectives. Thanks especially to Erik for posting the image from the Hassy. That's exactly the kind of thing I want avoid when I shoot. Back in the day when I shot with the original 1D, I frequently jumped through hoops to try to get rid of moire in garments. Back then, clients were even less forgiving because I was trying to transition them from film! My question was posed as curiosity, but for all practicality, I'd go with the 5Ds. I can't believe that in real world circumstances, the resolution of the 5DsR would be that significant over the "plain vanilla" 5Ds.

In any event, I would wait until the results are in from both cameras.  But I shoot the D810 (has no OLPF in the stack, not just filtered back out like the D800e or 5DsR) and have not experienced the type of issues Erik's Hassy photo shows with similar subjects.  Obviously, Nikon learned something from the D800 pair that allowed them to release a unified body without the OLPF.   

[ErikKaffehr]

Hi,

Just to say, it is not very common. But it occurs often enough to be a nuisance. Small pixels are beneficial in suppressing moiré.

This article covers a lot of it: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/78-aliasing-and-supersampling-why-small-pixels-are-good

Moiré is not the only colour artefact, see the attached shots:

• 6.8 microns, no OP, processed in Capture 1 V8
• 3.8 microns, OLP, processed in Lightroom 5
• 6.8 microns, no OLP, processed in Lightroom 5

This shot is essentially a test shot, I have noticed that the 6.8 micron camera was aliasing on the subject, and went back to make different test shots. Different cameras, using same focal length from the same shooting position.

I would say that small pixels go a long way to reduce aliasing. The second image is much softer, but contains better detail, check the readability of the text on the signs. The softness is a combination of lens MTF (a decent quality tele zoom at 100 mm at f/8) and probably also an OLP-filter.

Also note that all the thumbnails are aliased, but image #2 doesn't show aliasing at actual pixels. Downscaling is an operation introducing aliasing. Correct downscaling often includes blurring the image slightly before downscaling and applying some sharpening after final size has been reached.

Best regards
Erik

Thanks for all the different perspectives. Thanks especially to Erik for posting the image from the Hassy. That's exactly the kind of thing I want avoid when I shoot. Back in the day when I shot with the original 1D, I frequently jumped through hoops to try to get rid of moire in garments. Back then, clients were even less forgiving because I was trying to transition them from film! My question was posed as curiosity, but for all practicality, I'd go with the 5Ds. I can't believe that in real world circumstances, the resolution of the 5DsR would be that significant over the "plain vanilla" 5Ds.

[torger]

Good demonstration Erik. It also shows that Lightroom is not the best at hiding aliasing artifacts, which I've also seen.

Using RawTherapee I've noted that the default demosaicer "Amaze" is great at extracting the most from AA-filtered shots, but renders quite some false colors when there is no AA filter. However when switching to the "DCB" demosaicer you get performance close to Capture One and Phocus in that regard, but there is no demosaicer than can completely solve the problem of a broken signal...

I've attached an enlarged 100% crop of 6um pixels from my H4D-50 shot at f/11, converted by Phocus. This is a section of chaotic tree branches, so you won't find any moire, but false colors are very much present. Most of the magenta stuff is due to aliasing. I think some confuse it with chromatic aberrations, and when it's light desaturated magenta towards blue it's not that apparent. It would be notably worse in a Lightroom conversion.

Today I shoot mostly at f/16 to kill most false colors with diffraction, and I also gain some depth of field. I would probably get slightly better image quality though if I could shoot at f/11 with an AA filter, as the AA filter is more effective at removing aliasing than diffraction is.

[John Koerner]

I have to like Samsung's approach to 4K, which involves a full readout of the 28MP sensor for each video frame.  This kind of bandwidth and throughput is what is necessary, along with fabrication processes that reduce thermal buildup.  Supersampling makes a beautiful video frame.  The next step involves delivering at least 10 bits and at least 4:2:2 in a raw format.  But I don't think we're going to see this kind of throughput from a Canon sensor with this many pixels on it for the time being. 

I agree. Unfortunately.

[nemophoto]

Thank Erik. Interesting sample of moire, etc. Why didn't you create the same size sample of the smaller micron sensor as an example compared to the others?

For those pining away for 4K video (and I'm not one of them), this isn't the camera to feature it. There was a great interview on DPReview with Chuck Westfell from Canon. In it he was very clear that, essentially, the "logistics" (my word) of pushing that much data through a 50MP is not practical. I think, for the time being, those truly interested in video, would be better served using the 5D3 or even the 7D2. Or of course, pick up the 1Dc. I think Canon figured they should through video in with the camera, but probably felt this was a camera to serve still shooters (like me) who have clambered for higher res first, and the occasional videographer second.

I'm not quite sure why everyone is trashing the camera before any true samples have been test, whether or not is uses the 7D2 chip fabrication or not. If you don't like the results, switch to Sony or Nikon. For me, I believe as much as I'd prefer a 1D-series body, it will do well for my needs. Until Sony actually introduces something, it's still vaporware. For a few short minutes, here and there, I've contemplated jumping ship, but I'm honestly too heavily invested in Canon gear at this point. Had Sony put more effort into something like the A99, that would have been far more appealing. I personally hate the A7 series cameras. Using one for a day made me feel that way. I just found it annoying to use. But that's me. I'm a traditionalist and prefer and optical viewfinder. Mirrorless cameras are fine if I want to plunk around, take shots while hiking, etc. But give me the two ton brick for my commercial work. I know it won't fail me. Just my 2-cents.

[ErikKaffehr]

Hi,

I didn't have the time...

Here is a comparison, both samples from processed in Lightroom.

BTW, one reason not to downscale the 3.8 micron image is that original image actually shows the true resolution of that image.

Best regards
Erik

Thank Erik. Interesting sample of moire, etc. Why didn't you create the same size sample of the smaller micron sensor as an example compared to the others?

[nemophoto]

Erik, very interesting to see. Thanks for sharing! The comments made by some that smaller photosites are less likely to cause moire might be the reason the original 1D had such issues. I believe the photosites were something 8.3 microns.

[Phil Indeblanc]

Maybe the science says otherwise, and some tests use particular lens selects that show issues of moire without AA filters, but with the select gear I have and use....Importantly with subjects I work with, I have to say that the sensors I use without an AA create a noticeably sharper, more 3D raw file vs those with an AA.

The difference is something I see, and I cannot see post sharpening be the better alternate, nor be equal to getting the original lines (or as good) in the image.
Maybe its the macro subjects, but the difference is clear in my tests(DSLR, full, crop, DSLR without AA, MFdb without AA).

Having said this I have respect for a number of posters here, and i'd like to see how I can better understand this.
There are a couple users here that have "scientifically" tested a few things, and I'd like to put theirs and my findings to see whats going on....Before I no longer have the gear/chance to do so.  Since in macro work the pixel peeping is not only a self driven desire, as the work has to be perfect and demands it. Editing has to be clean with a solid chunky file.

I'm open to testing, if you are. I can think of Erik and Bart....  and there are a few others not coming to mind at the moment...but we can setup some parameters to work under/?

[ErikKaffehr]

Hi,

The AA-filter reduces MTF, so that is expected.

The Canon 5Ds has higher resolution (in lp/mm) than any full format or MFD camera, so it may be a good candidate to use without AA-filter, although theoretical studies indicate that pixel pitches around 2.5 microns may be needed for correct rendition with very high quality lenses at near optimal apertures.

OLP filtered or not, the OLP filter is always designed for the pixel size, so the effect of the OLP filter is reduced withe decreasing pixel size.

What apertures do you use? On my P45+ at 6.8 microns I see a lot of issues when shooting at f/11, they go away at f/16. From that I guess that the Canon 5Ds would be aliasing free at f/9 or so.

Best regards
Ertik

Maybe the science says otherwise, and some tests use particular lens selects that show issues of moire without AA filters, but with the select gear I have and use....Importantly with subjects I work with, I have to say that the sensors I use without an AA create a noticeably sharper, more 3D raw file vs those with an AA.

The difference is something I see, and I cannot see post sharpening be the better alternate, nor be equal to getting the original lines (or as good) in the image.
Maybe its the macro subjects, but the difference is clear in my tests(DSLR, full, crop, DSLR without AA, MFdb without AA).

Having said this I have respect for a number of posters here, and i'd like to see how I can better understand this.
There are a couple users here that have "scientifically" tested a few things, and I'd like to put theirs and my findings to see whats going on....Before I no longer have the gear/chance to do so.  Since in macro work the pixel peeping is not only a self driven desire, as the work has to be perfect and demands it. Editing has to be clean with a solid chunky file.

I'm open to testing, if you are. I can think of Erik and Bart....  and there are a few others not coming to mind at the moment...but we can setup some parameters to work under/?