Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Foveon Sensor for RED ?

Daniel Browning said:
Incidentally, the "look" that Foven has (aliasing) is very fragile in post processing. As soon as you downsample the image, say, to half the size, any well-written resampling filter will remove the aliasing. As I said earlier, that's why people find and use resampling filters that add aliasing (even with Bayer cameras).
Click to expand...

Well written downsample filters cannot remove any aliassing already in the image. That's the problem with aliasing - once it's in you're stuck with it until you blur the image so much that you practically loose the image to get rid of the aliasing.

Getting the "right" downsample filter is always a trade off between ringing, aliasing and blurring....

Graeme
 
Graeme Nattress said:
Well written downsample filters cannot remove any aliassing already in the image. That's the problem with aliasing - once it's in you're stuck with it until you blur the image so much that you practically loose the image to get rid of the aliasing.
Click to expand...

Thanks for the correction; I need to remember that.

Imagine if you had a test pattern consisting of black squares on a white background spaced in perfect frequency with the AA-less sensor array. The image would be all black if the alignment happened to be exactly right, and no amount of resampling would fix that.
 
Now, chroma aliasing can be dealt with if the luma is relatively alias free. But allowing luma aliasing in to the vast extent that Sigma do, ignoring the good "advice" of Foveon themselves who state that their chip still needs an OLPF....

Graeme
 
Daniel Browning said:
Imagine if you had a test pattern consisting of black squares on a white background spaced in perfect frequency with the AA-less sensor array. The image would be all black if the alignment happened to be exactly right, and no amount of resampling would fix that.
Click to expand...

Sampling theory is based upon band-limited signals. The black/white test signal is not band-limited. (Graeme, thanks again, for reminding me this a long time ago. :pinch: )
 
I think the three of us are all saying the same thing. This worries me. Are not conjunctions signs? :-)

Graeme
 
Hi Daniel,

But, that also brings us to a little realized point. Suppose you are given a signal that is very wide band. Then, the nearest-neighbor interpolation, which is much derided, and treated like a pariah, under some situations turns out to be better than, say sinc interpolator, for some part of the signal, but not necessarily all of it.
 
As I've said before, there is not one best downsample filter for all purposes and occasions.

Graeme
 
A visual alignment of the planets....
 
You must log in or register to reply here.
Back
Top