What filter to reduce aliasing?

Thanks for the update Infurno.

Complete guess, but I wonder if these filters can be stacked (cumulative effect) ?
Can you try some wide shots with the 2.0 ?

I only have have a single 4.0 and looks like your results with a 2.0 are leading to the same conclusion, ie : hard to predict exclusion of these horrible artifacts (way to many factors involved ... focal length, aperture, relative size of what is causing the problem in the shot, etc).

I have been experimenting in post by applying gaussian blur (or mediam) the top layer and then doing a layers 'color' blend ... helps in some cases (for moire) but aliasing is hard to fix

For those asking for 'before / after' grabs ... I did intend to post some test results from the few 4.0 test, but concluded that it was not worth it as the 'after' was just as bad ! I'm sure these filters have a 'sweet spot' but are way too weak for wideangle work on a 5Dii ...

The 60D is rumored to have a 15Mp sensor and Digic V. That would be about 12Mp in 16x9 format. It is also rumored to support 24p and 30p.

if Canon includes manual controls, and increased the speed to process the full sensor in video, that may fix the problem. If they did, a Digic V 5DM3 or 6D may follow shortly with the problem solved, leaving lots of 5DM2 bodies for sale on ebay.

No doubt that within give or take 12 months the problem is "solved". But I will keep my 5D for stills.

I will shoot more tests today with the 2.0 and post anything interesting. There are some telephone lines outside my office so I will try and get a sample of that. Grass may be another good one to try, and I have not had a chance to mow

Just spoke to a Caprock rep about how the filter works and how to best use it. The focal length and aperture will affect the performance. They also have 1.0 and 1.4 versions, I may grab the 1.0 to see what it will do.

They will take returns, but I wish I could just get a few samples to experiment with without forking over a couple hundred.

Well, it doesn't seem to affect too much the resolution and sharpness, after all. Or am I wrong?

It does blur everything a little, but you have to zoom in to notice.

Yet, about your sample. I don't see substantial differences concerning sharpness (same applies to resolution, I guess) between both with and without filter.

Moreover, you're quoting contradictory conclusions. If you need «to zoom in to notice» how can it show «some slight visible blur»? OK, you introduced the word 'slight', will it be enough for some or a lot of applications?

I will give you this NR example posted in these same boards a few days ago:



http://alainrobert.biz/avant_apres2.jpg

For the shooter himself, the results of the filter @post are a «little bit soft», too me just perfect in order to use it without doubts. OK, granted this sample, of course. No further tests added.

That's why I'd like to read additional impressions on the subject. As well, more posts with comparative pictures. Especially, on the wide end.

More than charts, real world situations that's what counts. If it was for the first charts published @reduser.net, I wouldn't have bought the Canon, found what I've seen from real world footage and I'd still be constrained to deal with the workarounds for handling the 20 pounds of a RED ONE setup, everytime I need.

Thanks once again.

NICE!!! This might get me the softness I want and a lot of the noise is gone and pixels of lost color information.

I will shoot more tests today with the 2.0 and post anything interesting. There are some telephone lines outside my office so I will try and get a sample of that. Grass may be another good one to try, and I have not had a chance to mow

I've never seen aliasing on grass. (But try befor mowing! ) Straight or slightly curved thin and non organic lines are the best/worst for the test IMO. Try with cars, roofs, buildings in general and so on. The telephone lines you intend to shoot should work fine. Some posts before you wrote you were printing some pattern for testing. Seems to be a good idea, especially because it makes testing easy and repeatable. Did you succeed with that?

Here are some resolution charts:
http://files.getdropbox.com/u/305574/aliasing/Chart.tif
http://files.getdropbox.com/u/305574/al ... ck-2.0.tif

Looking over what I shot today, will probably post a few more crops.

Here is a sample of some power lines:


The pattern on the lines you see is two twisted together. This is actually a pretty good result. Most of the aliasing and moire is gone.

Here are some resolution charts:
http://files.getdropbox.com/u/305574/aliasing/Chart.tif
http://files.getdropbox.com/u/305574/al ... ck-2.0.tif

Looking over what I shot today, will probably post a few more crops.

Apart the introduction of vignetting, I don't see considerable less sharpness or resolution or, at least, it does seem so to me. Unless, a little bit less aliasing, even though, most probably to be seen on real world context as you could post before.

I would try the Caprock 1.0 filter with no fear.

If you'll confirm no losses as far as sharpness/resolution concerns (well, then, it would remain the 1.4 to test it, right?), it eventually is a must have 'cause you can get better results in the end.

PS -- Now that you beat me with your findings from real world (we won't shoot charts in real life, correct? ), aside the softness introduced but you're giving us a crop and again you won't crop your footage, I say, you will have an important task in the hands: what's the best compromise in terms of moiré/aliasing killer and sharpness/resolution? Caprock 2.0, 1.4 or 1.0 filters?

Thanks for your care. I'm sure me and others won't forget you and your tests when we'll be using the ultimate outcome based on the best Caprock filter model solution brought by you.

Yeah, kind of wish I ordered 1.0 today. Would have had it by Friday. I am certain it would add a lot of blur, but worth a shot anyway.

Maybe I can get faster shipping.