A lens can be too sharp, it looks like this...

Started 8 months ago | Discussions thread
abera Regular Member • Posts: 103
Re: That's a far cry from "too sharp".

Great Bustard wrote:

Erik Kaffehr wrote:


A lens can be too sharp, it looks like this... On another thread I shared an image where the lens was sharp enough to cause serious issues with aliasing.

Aliasing is not caused by the lens, it is caused by the sensor not being able to correctly sampling the image projected by the lens.

Now, let's look at some details:

The very colorful tree tops here are caused by aliasing.

Pretty smooth surfaces can show aliasing. The diagonal bands on the upper window are also caused by aliasing of the blends behind the window.

So, what can we do about it?
With the gear at hand, there are three possible options:

  • Use a small aperture, so diffraction reduces sharpness.
  • Blur the color channels, the 'moiré' reduction in Lightroom and other programs does just that.
  • Defocus or use camera shake.

Neither solution is really great.

Moiré reduction smears out colors, as shown below:

Notice the difference in color on the windows frame.

The central region is sharp (not "too sharp", but sharp). However, let's look at the edges:

Kind of a disappointing showing given that they were well within the DOF at what is likely the lens' sharpest aperture.

I used to have East-German Carl Zeiss Jena 50mm f/1.8 Pancolar - the 8 bladed zebra version with loads of thorium dioxide in a couple of elements which demonstrated the need for AA filters corner-to-corner wide open on 24 MP FF sensor (with single directional weak AA filter). That's with a lens from the 60's.

In any case, the issue that you have displayed is why an AA filter is a good thing, despite the current call for no AA filter. If you want more resolution, you don't remove the AA filter, you get a sharper lens and/or a sensor with more pixels.

There are technical solutions. The best one would be having more megapixels, like much more megapixels.


Another solution would be to keep using AA-filters with high resolution sensors.


Multishot is possible, but many things would move in such a picture.


The camera used here was the Sony A7rII, it has 42 MP, slightly less than the Nikon Z7.

For the technically minded, I would also recommendthis article.

Pentax had a very interesting solution: use the IBIS system to create a pseudo AA-filter. I don't know if their implementation allows for variable amounts of blur, nor do I know how well it actually works in lieu of an AA filter. But it does sound like a good idea.

I seem to recall the system had some limitations (after all you can only move the sensor so fast), and also the AA effect is less optimal than with is achieved with birefringent material like quartz (which I believe is used for image sensors, though could be wrong).

The best idea, of course, is to have a "proper" AA filter on a sensor with tons of pixels. This is, in my opinion, one of the best arguments for sensors with huge pixel counts even if one isn't doing heavy cropping.

The best idea is to have a sensor which samples at a rate where diffraction is large enough to prevent aliasing - on full frame with color filter array it might be more than a billion pixels for wide apertures.

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow