Do some lenses cause more aliasing than others?

Samuel H said:
Today, that statement that "all DSLRs have aliasing/moire" is no more true than "all video cameras have aliasing/moire".
Specifically, the D5200 doesn't seem to have more aliasing/moire than, say, an EX1: if you're adamant to make it appear, it will show up, but it's going to be a feat.
Click to expand...
Most interesting. I've googled but can't find any video of a zone plate or even a res chart on the D5200. Can you point me towards one?
 
J Davis said:
every single digital camera (still or video) out there, alexa, red, f5 can moire and show aliasing
Click to expand...
Yes, that's true. And every single animal out there poops. But an elephant does it a tremendous amount more than a mouse does. And line-skipping sensor sensor technology delivers much more aliasing than a properly-engineered video sensor does.

There are two major contributors to the aliasing issue, there's the way the sensor is designed (proper sampling or line-skipping) and there's the effect of the OLPF. The BMC shows horrific aliasing, even though it has a properly-sampling sensor on it, because they didn't put ANY OLPF on it at all. It looks almost as bad in terms of aliasing as a line-skipping DSLR does. The degree of tuning of the OLPF does affect the presence of aliasing and of the rainbow aliasing effects.

A line-skipping DSLR doesn't have an OLPF in place, because such a filter would ruin the stills performance.

If Nikon has truly mastered the problem, that would be fascinating to see. But the only way to actually know, is to see a zone plate or at least a resolution chart. That will tell you what the camera is actually doing, which is the only way to know. Eyeballing "real world" shots doesn't tell you anything, so I'd be looking for actual video zone plates. Alan Roberts did some tests on the D4 and D800, finding the D4 surprisingly acceptable in 1080, but the D800 was "worrying". So maybe the D5200 is indeed clean... would like to see it proven out though.
 
what I wrote ->

Moire on the dslr's is due to downsampling as you know but moire on the the c100, c300, red, alexa and any digital sensor can happen like you are describing. You need to understand the cause and then you can be better armed to combat it. As Noel says, sharpening makes it worse so kill that setting in camera.

the material has a fine weave which forms a grid, the sensor has an array of pixels which forms a grid, moire is caused by the interpretation of one grid by the other when the ratio of one grid to the other goes beyond the nyquist limit. changing the distance between the subject and the camera (*) forces a change in the ratio but this may not be an option if you are not allowed to reframe. Another trick is to shoot just slightly off focus, just enough for the moire to disappear, then sharpen in post. On set, 1:1 monitoring is key as you have discovered.

I know some post production tricks that will get rid of moire on your footage, you can read them here
http://jdmax.com/how-to-fix-chroma-moire-and-chromatic-aberration-in-post-for-canon-dslr/

* or zoom.
Click to expand...
 
Barry_Green said:
Most interesting. I've googled but can't find any video of a zone plate or even a res chart on the D5200. Can you point me towards one?
Click to expand...

No, I haven't seen those either. I can point to some macgregor tests that are terribly clean:
https://dl.dropbox.com/u/13362047/5200vs5dm3.01.mp4
https://dl.dropbox.com/u/13362047/5200vs5dmk.02.mp4
Depending on how good a monitor you use for viewing that, you may see some aliasing in the strip of light below the light-blue car, but that's about it. It also has much better gradients and much less FPN than the 5D3, and similar (if not better) DR.

I may be renting/borrowing a D5200 in a couple of weeks, to make a non-beta version of Flaat for Nikon. I have a resolution test chart, I'll shoot it if I get the camera.

For the record: my bet is that the Toshiba sensor in the D5200 does in-sensor per-channel binning with a full-sensor readout, instead of line-skipping. Not all sensors can do that, but some can (one example I'm certain about: the cmosis CMV12000).
 
Last edited:
Barry_Green said:
T...But when the DSLR does a lousy job of pixel-skipping and scaling, then how can you even begin to compare that against a designed-for-the-purpose dedicated cinema camera like a C300?
Click to expand...

That's the point, the pixel skipping is a compromise on making the DSLR camera 'work' for stills, where much higher resolution is the current standard, and getting 1920x1080 out of the sensor. Get rid of the need for 'high resolution stills', I don't see why the AF100(*) needs to be 2-3x the price of the GH-x, other than some sort of 'pro' market attribute. In the car world this is sort of equivalent to how Land Rovers in the US are expensive encrusted automobiles with lots of nobs and 'comforts', were as in various 3rd world nations one can by a Land Rover with just an engine, manual transmission, and sufficient body as required. (Same is true for many US 'luxury' imports...).

The reason I've been given for 'why the US price is so high'... is, 'so as to not dilute the market appearance of premium'.

So, put a small piece of high quality glass on the DSLR sensor, make a bit of improvement for 'binning' and not 'skipping', dump the audio capture capability (although at the economies of scale, that's probably only about a 10 cent part...).. and at the same price as the DSLR have a 'purpose built' cheap solution.

*Since I can't find a definitive statement from Panasonic about the internals of the AF100 (or the GH-x for that matter...) what I intuit from others who have guessed at the inner workings, the AF100 may 'bin' samples to achieve the downsizing. But the others seem to think that it skips to subsample the higher resolution sensor.

In ether case, and for all 'digital sampled system' there will always be aliasing issues when the frequency components exceed 1/2 the spatial sample rate, aka pixel resolution, if the the filter does not cut off at 1/2 the max sample rate.

DSP-wise one can create a low pass filter, or 'dither' fast edges, as does film grain in the almost 'continuous' case of photographic film.
 
Last edited:
You must log in or register to reply here.
Back
Top