Please consider registering
guest

Log In Register

Register | Lost password?
Advanced Search:

— Forum Scope —



— Match —



— Forum Options —




Wildcard usage:
*  matches any number of characters    %  matches exactly one character

Minimum search word length is 4 characters - maximum search word length is 84 characters

Topic RSS
Anti[aliasing filters
September 15, 2012
7:35 am
Razz16mm
Member
Forum Posts: 482
Member Since:
March 27, 2012
Offline

IMO optical anti-aliasing is absolutely essential to high quality film like digital imaging.
I happened across this unique new solution that might be worth exploring, if not for the first run of D16's, maybe as a retrofit and possible upgrade to future production runs.

http://www.mosaicengineering.c...../nbaa.html

September 16, 2012
3:29 am
joerubinstein
joerubinstein

Camera Owner
Forum Posts: 715
Member Since:
February 28, 2012
Offline
2
0

Very very interesting! I will definitely give them a call next week to talk about what they are doing. The sample images are very puzzling. looking just at the figure 2 it looks like the image on the right has quite a bit more resolving power.

Thanks very much for bringing this to our attention!

Best, Joe

September 16, 2012
6:15 am
Razz16mm
Member
Forum Posts: 482
Member Since:
March 27, 2012
Offline
3
0

When aliasing causes chroma moire it pollutes lower frequency details well below the range of the detail that produced it. It actually reduces image resolution IMO.
Nikon posted samples comparing their D800 and D800E, the E version has no OLPF filter. The filtered version looks a bit softer at first glance as expected. But if you magnify down to the pixel level, the resolution of real image detail is virtually the same. The detail borders are just smooth instead of jaggy. Aliasing introduces false detail that gives a superficial appearance of a sharper image, but it is illusion. On fine detail patterns like fabric weave that cause moire, the filtered version resolved higher clean detail compared to the masking caused by the chroma moire.

I was floored by the Mosaic samples too. Seems like they have found a solution that has all the benefits with less detail softening compared to conventional birefringent filters.

September 18, 2012
3:47 pm
joerubinstein
joerubinstein

Camera Owner
Forum Posts: 715
Member Since:
February 28, 2012
Offline
4
0

I put a call into them today, hopefully I hear from them tomorrow :)

September 20, 2012
9:11 pm
analoggab

Camera Owner

Kickstarter Backer
Forum Posts: 168
Member Since:
April 7, 2012
Offline

If you find that option is better than the one you currently had with the D16, will you have time to switch?

Didn't you guys have a good OLPF filter solution in place?

September 21, 2012
1:42 am
joerubinstein
joerubinstein

Camera Owner
Forum Posts: 715
Member Since:
February 28, 2012
Offline
6
0

We have an OLPF option, but it wasn't in our sensor prototype so we haven't gotten to test it yet. This is still a place we are looking to stand out.

I talked to Dave at Mosaic today and it seems like we could release the camera with a removable OLPF / IR, and if a partnership with Mosaic works out, the original OLPF could be removed by the user and a new one put in. There is a software element to it too, so it would need to be integrated on several levels, and would make the footage shot with it not compatible with other transcoding software. The process is really interesting, Dave explained it like this; basically they add very faint predictable noise that they can remove in post, and this allows them to differentiate between detail and morie more easily.

I haven't gotten the full scoop, but the process sounds really interesting and the results are impressive!

I don't know if we'll make it standard on cameras ever, and almost definitely not in this first 100, but it may be a very nice option for people if they want to commit to using our software for all of their transcoding.

What do you guys think?

Thank you Razz for pointing us in this directions, either way it's very interesting!

September 21, 2012
2:31 am
Thyl
Member
Forum Posts: 271
Member Since:
June 22, 2012
Offline
7
0

Someone here on the forum pointed to an Arri analysis (titled 4K+ systems theorey basics for motion picture imaging) that showed that S35 film contains an equivalent of at least 4K . More interesting here, however, was the finding that actually, since moire patterns are not present with film, even recording with a 4K digital camera will not be equivalent to S35 in this regard (at least that's what I took from the paper, but then, I am no expert here). Of course, this has to do with the irregular (and changing from frame to frame) nature of film grain.

So, if these guys could really provide a true 2K resolution with moirees removed, that would be better than a conventional, resolution degrading anti aliasing filter. I presume that this "noise" is added via the filter? And it would result in a "bogus" CinemaDNG file that needs special "cleaning" before it could be used? Would that mean that no changes to the camera firmware would be required? Or if, why not also putting the cleaning step into the camera? Is that only possible after de-bayering?

What would that option cost? In hardware and software?

September 21, 2012
6:26 am
Razz16mm
Member
Forum Posts: 482
Member Since:
March 27, 2012
Offline

Sounds very interesting indeed. The results speak for themselves. Making the OLPF filter user removable/interchangeable might be the best all around solution, but personally I would not shoot without one, even with the minor loss in resolution.
Developing a software switch that could be turned on or off for use with the advanced filter option might be the best all around solution. That could be added down the road a bit. No fixed pattern imager, even 3-chip solutions, can optically resolve 100% of its pixel resolution without unacceptable artifacts. The ARRI Alexa for example effectively resolves 1.6k horizontal from a 2.9k pixel sample. But the Digital Bolex can potentially exceed the practical resolution of S16mm film, around 800 lines vertical.

http://mosaicengineering.com/i.....before.jpg

http://mosaicengineering.com/i.....aafter.jpg

September 21, 2012
2:38 pm
wado1942
Member
Forum Posts: 339
Member Since:
March 19, 2012
Offline
9
0

I'll point out that #1, they don't show the same shot with a traditional OLPF to compare and #2, there's still alias distortion in the image. As I've said previously, there's no substitute for a proper OLPF. The idea of merely allowing alias distortion and using specialized software to try and find it to remove later will complicate the works while most likely giving inferior results under certain conditions not shown in their little comparison.

I don't know of any digital cameras, 3-chip or otherwise, that resolve 100% of their pixel resolution. Those that theoretically do (but in practice still don't), lack OLPFs and you can easily see the alias distortion. I've learned to short my focus when those situations arise, which keeps you well below full resolution any way. An OLPF that's properly paired with its image sensor will reduce clarity very little and allow virtually no aliasing. There's nothing more annoying to me than cheap video, of which aliasing is a sign.

I think it may be a good idea to use a traditional OLPF and perhaps make this other idea an option down the line.