Actually, the Olympus is oversampling but still averaging over its big pixels. Imagine an image convoluted with one "large" pixel, and then sampled densely. That convolution acts as a kind of an AA filter.
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
why buying Canon 5DS instead of 5DS R
- Thread starter Stollen1234
- Start date
What you think are sharp apertures are still being softened.
Soft+Sharp is still better than soft+soft, so you see an improvement, but it isn't as much as you would get from sharp+sharp.
That is precisely what I told you in the part of the post you didn't quote.
I don't need to "imagine" that, I've done it, measured it, patented it, published it, sold it... As I said, it is just microscan, something we pioneered 30 years ago and it certainly eliminates luminance aliasing. Olympus are also "macroscanning", hence the 8 step sequence, to eliminate chroma aliasing, not something I was concerned about with monochrome sensors 30 years ago.
Last edited:
I know, that's the point of the AA filter, and this is exactly where it is needed. But I stull use them, while you are advocating avoiding sharp apertures with AA-less camera.
Sharp+sharp is worse, called aliasing.
Not so: sharp + undersampling = aliasing, not sharp + sharp. There are many ways to avoid undersampling and softening is only one - particularly poor way of doing it.
In the context we are discussing it, it is so. If we are allowed to fantasize, a zillion mp sensor would do better, or a Pentax type of switchable AA filter with the current densities is a better solution, at least on paper.
Not at all. We have discussed several alternatives, in several threads. There are also external AA filters now available which claim to be blur free...
http://www.mosaicengineering.com/products/nbaa.html
So, sharp+sharp ≠ aliasing.
Yeah, sure. Can they be possibly be vaguer than that?
Sharp+sharp = aliasing. It is a theorem.
I don't know. How much vaguer would you like? Why don't you ask them?
Thats your theorem, not the sampling theorem.
As detailed here, there is always a stopped down aperture which gives almost identical performance without an AA filter as the sharpest lens with an AA filter. You aren't forced to project images onto the focal plane that the sensor can't unambiguously resolve. You have control over how sharp the image is by use of the lens settings and that is better than just accepting a factory level of softness that you can only make a little or a lot worse.
Well, that is a rebuttal of yours.
So you are saying, make one of the "sharp" soft, and then sharp + sharp=sharp is true?
In other words, when you think that f/4 is the best aperture, use f/1.2 or f/11, or misfocus, because your camera is too sharp? That would be a very creative use of MA, for example - tune it so that it you camera would frontfocus, well except when you shoot wide open.
BTW, I have RAW files posted by a fellow member, A7R+50L wide open. There is strong moire on one of them, and visible aliasing just slightly behind the DOF, where there are many high frequencies left, due to diffraction effects.
My 70-200/2.8L IS II is diffraction-limited at f/3.2. I'd say that's a good bit faster than f/11, and I know there are primes that are diffraction-limited at f-stops faster than that.
Jeff Fenske
Leading Member
- Messages
- 951
- Reaction score
- 223
Didn't Nikon completely remove the two filters of the D800e, replacing them with nothing in the D810, by redesigning the distances? And the result is no loss of sharpness and probably less light falloff too, especially off center.
Correct me if I'm wrong, but I've never heard Nikon had to add something to replace the AA filters they took out.
It wouldn't surprise me if Canon hasn't already started development on the 5Ds R's replacement, which won't have any AA filters, just like the Nikon D810.
The 5Ds R may soon be a dinosaur.
This is my understanding as well. I wasn't aware that AA filters are there to correct SA. Isn't this corrected with the lenses?
I have not researched what Nikon actually did but in principle, you want to put there glass that imitates what happens to a ray having the same (average in some sense, I guess) index of refraction as the two AA pieces removed. The lenses are designed for a filter stack of certain thickness there, see the LR article.
Last edited:
I can see why they might do that.
Jeff Fenske
Leading Member
- Messages
- 951
- Reaction score
- 223
This test shows why the 5Ds R/D800e method is inferior, and why Nikon completely removed the AA filters in the D810. Nikon also says the D810 is sharper. And this is why the 5Ds R may soon be a dinosaur, when Canon replaces the 5Ds R with a D810-like 5Ds RII, which they may already be developing.
– –
Loading…
photographylife.com
Is Nikon D810 Sharper than D800E?
"As you may already know, the difference between the Nikon D800E and D800 is their filter stacks – the D800E has the same size stack as the D800, but its third filter reverses the effect of the first one, essentially cancelling out the effect of the optical low pass filter [which is what the 5Ds R does – Jeff]. …
"In comparison to the above, Nikon completely removed the OLPF filter on the D810. Contrary to what some people think, although there is no low pass filter on the D810, it does not mean that there is no filter stack at all – the D810 still has a filter to cut off UV and IR. The big question that comes up, is whether the omission of the low pass filter actually results in increased sharpness. Shortly put, is the D810 sharper than the D800E? To answer this question, I ran numerous lab tests….
"A total of approximately 30 different tests were performed. Here are my findings:
- In the center of the frame, the Nikon D810 mostly performed slightly better than the D800E. However, this difference, as can be seen below is very minor and amounts to less than 3% difference – not something one would ever be able to visually notice.
- Mid-frame performance varied by approximately 5%, with the D810 continuously showing advantage over the D800E.
- Corner performance varied greatly, somewhere between 8% and 10% in improved sharpness on behalf of the D810.
We are going in circles. For somebody who dislikes AA aliasing, the D800e is better, and the D800 is even better. If you like aliasing, it is the other way.
EDIT: I just noticed this: "I mounted the lens on a very sturdy tripod setup, acquired precise focus using Live View, then simply changed camera bodies without touching the focus ring. The results were quite interesting." Cannot believe that he would do that. Not that it changes anything in my opinion.
Last edited:
Jeff Fenske
Leading Member
- Messages
- 951
- Reaction score
- 223
My emphasis seems to be different than yours. Perhaps you are right in theorizing that removing the aliasing in the 5Ds R (and Nikon D800e) with the first AA filter and then putting it back again with the second, cancellation AA filter does remove some of the aliasing, but it's probably not much. It's mainly two filters that don't do anything but decrease sharpness and light transmission.
My point is that completely removing the AA filters is the right way to do AAless when sharpness is the goal, which Nikon has done with the D810, and Canon is likely to also do, unless the AAless route is proven to be undesirable, which may happen, because sharpening can often achieve similar results; though, AAless is the current trend.
Putting two filters in the light path that bend light one way and then bend it back again is bound to affect image quality, and is an inferior design that was apparently implemented by Canon to be able to build both 5Ds models at the lowest cost. They probably didn't know early enough in their design process that Nikon was going to come out with the completely AAless D810, which would render the D800e obsolete.
Which is what the 5Ds RII will do to the 5Ds R, I'm guessing.
So as long as Canon put the 2 AA filters into both 5Ds models, they might as well do something useful, which the 5Ds does, by removing aliasing artifacts (that may often appear in the 5Ds R) and moiré (which will occur less often).
The 5Ds may give a more natural looking result, which can also be sharpened to look more like what the 5Ds R will give.
The AA filters in the 5Ds are likely to be weaker than in the 5D3 (because video is no longer the priority, and this is what Nikon did in the D800, and because of what Canon has hinted at), so the AA filters in the 5Ds are not going to affect sharpness compared to what Canon shooters are used to.
He then did the tests which resulted in the numbers that I just shared. 8-10% reduction in sharpness at the edges is significant — from filters in the light path that mostly don't do anything but cancel each other!
And Nikon publicly says the D810 is sharper.
Learning from his first mistakes and telling us about it shows his integrity in "a total of approximately 30 different tests were performed." It's to his credit that he admitted his mistake, which shows he can be trusted, and that he he just wanted to know the truth, himself.
If they did that then they will certainly have increased spherical aberration on the D810 compared to the D800 - its the laws of physics.
"No loss of sharpness" is very definite and somewhat over-optimistic. There may be less loss of sharpness from the spherical aberration than the AA filter had, but it certainly won't be none. You can get an idea of how much loss of sharpness will result from a post I submitted yesterday.
Loading…
www.dpreview.com
Although a hypothetical example, the conditions aren't too different from a real Canon case, and it is explained in adequate detail for you to convert to actual conditions in the D800/810 comparison.
Would they tell you if they did? After all, Nikon don't tell you what the thickness of the sensor window or the AA filter is in any of their cameras - we have to find that out from the guys who tear them apart and measure such things. So if Nikon replaced an optically active element with a completely passive one, as required to achieve "no loss of sharpness", that completely passive element could well translate to "nothing" in marketing speak.
Well that's certainly one way of doing it, but the point is that this technique is entirely flexible: you only introduce the amount of softness that the image actually needs, rather than be faced with a "fixed penalty" unsharpness from the moment you fit your sharpest of lenses onto the body.
Not every subject aliases and, within days of using your new 5DsR you will have a much better feel of when it, together with the lenses you use, are likely to encounter such problems and need to be opened up, stopped down or simply minor focus adjustments to eliminate aliasing with less loss of sharpness than the AA filter imposes. Just another example of learning your tools. ;-)
--
Its RKM
Last edited:
Keyboard shortcuts
- f
- Forum