OLPF-mod/fish-net/back-filter for Moire?
- Thread starter USLatin
- Start date
shaocaholica
Well-known member
Not yet. I still don't have the camera and I don't want to make any delays by having the shop shoot something since they're not familiar with the video controls anyway.
shaocaholica
Well-known member
Man, can't wait. Did you shoot something before you sent it so that you could shoot it again that has moire/aliasing?
Didn't have time. Its ok though, the people who will be testing it will have access to unmodified 7Ds and T2is.
shaocaholica
Well-known member
You can't just do that. The sensors in P&S cameras are tiny. The OLPF is attached up against the sensor and must be the same size. A OLPF from a P&S would be too small to cover the entire sensor in a 7D or T2i. The only viable OLPF transplant is from another camera with the same sensor size. You could even take a OLPF that is much larger and have it cut down but I'm not sure how easy that is.
shaocaholica
Well-known member
Thats pretty much it.
Angry Leprechaun
Well-known member
Can't wait to see the tests, but I was wondering if this would help with moire and aliasing.
http://www.bhphotovideo.com/c/product/457714-REG/Tiffen_W77DDFX1_77mm_Digital_Diffusion_F_X.html
Obviously the difference probably wouldn't be substantial, but even if it helps a little bit I wouldn't mind getting one.
http://www.bhphotovideo.com/c/product/457714-REG/Tiffen_W77DDFX1_77mm_Digital_Diffusion_F_X.html
Obviously the difference probably wouldn't be substantial, but even if it helps a little bit I wouldn't mind getting one.
Rick Burnett
Section Moderator
I think the issue is people need a solution between the lens and the sensor, not before the lens. I don't know what this filter is you've specified, I did try to find out more info, but from the Caprock 2.0 I own, it doesn't work on all my lens very well.
sorry my fault, the P&S filters are too tiny, but if the OLPF can be put in the back of the lens without affecting the performance ( http://www.optics-online.com/lpf.asp ) , a 14x14mm or even a 12x12mm filter should be enough to cover the image.
Lets hope your D30 OLPF works fine.
shaocaholica
Well-known member
I'm pretty sure that the OLPF has to be right up against the sensor, not anywhere in the optical path. Although I'm not sure what kind of effect it would have at different locations in the optical path. Still, its a lot less stuff to worry about if its just installed permanently in the camera vs a removable filter. Some lenses have really really large rear elements. Look at a 200/2.8 or 300/2.8 or even a 50/1.4.
Barry_Green
Moderator
It probably would, actually. But the question is whether it would perform equally at all focal lengths. That's the thing with the Caprocks -- with certain focal lengths combined with certain Caprock filters, the combination is magical. But use that exact same Caprock with a different focal length and the combination is tragic.
I would expect that a diffusion filter would help, but would it help equally across the board? Or would it be more applicable at tele lengths and less applicable at wides? I don't know... hmmm... this is a job for someone who lives next to a big camera rental place (yes, L.A., I'm lookin' at YOU) to try it out...
The simple fact of the matter is that anything you do to minimize what's sharply in-focus, will help to minimize the aliasing.
But that is still in the optical path.
So I've read conflicting definitions of an OLPF. By name, Optical Low Pass Filter, it would imply that it would cut UV and higher frequencies. Couple this with the IR filter that is usually present and you really have a band-pass arrangement.
However, I just read some website that claims that these OLPF are simply a screen that blurs a picture a couple of pixels. That certainly does not fit the name of OLPF.
Either that or my EE education and former robotic vision and spectral studies are all wrong.
It seems that the problem is really higher frequency harmonics being aliased(a different definition of aliasing) down into the visible spectral range, right?
EDIT: good info here: https://www.avxcorp.com/docs/catalogs/opticaldev.pdf
Last edited:
Ian-T
Veteran
Hmmm….. If I read you right you must mean this is Kinda High On our Long term priority Interests. <<..if you can make any sense out of that statement. :cheesy:
Barry_Green
Moderator
It doesn't affect the frequency of light that gets passed. It affects the sharpness of detail that gets passed. In other words, an OLPF is a blur filter.
Whether the name is "correct" or not, the function that the filter performs is blurring.
shaocaholica
Well-known member
"Low pass" is a little bit confusing. There are other things that are measured in frequency other than photons such as spatial detail. What we're dealing with is the latter. What the OLPF does is only allow detail below a certain threshold to pass. The threshold can be thought of as a spatial frequency. Hence the "low pass" descriptor.
From what I've been told and read, light passing through the filter doesn't get blurred, instead it gets broken up into "shapes" that are ~1/2 the wavelength of the threashold. There is no detail within the light shape and such no detail below the wavelength may pass. Think of it as a kind of microlens arrary thats "defocused".
shaocaholica
Well-known member
Update:
Camera is on its way back. Arriving tomorrow.
Camera is on its way back. Arriving tomorrow.
Then how come datasheets like the one I linked to claim these are UV blocking filters and make no mention of blurring?
That's because it's not "blurring", it's limiting spectal folding. I'll try to explain easily.
In theory, you have a CMOS sensor with RGB sensitive arrays which give you 3 images in 3 discrete spectral ranges. With the OLPF as both a IR and UV limiting device, you are technically creating a band-pass filter which trims the bandwidth of your signal into a usable range.
In all e-mag(electro-magnetic) spectra you have frequency folding which in layman's terms means that a frequency will have duplicate images of itself at 1/2 and 2x it's center frequency. Since IR is lower in frequency, the lower image at 1/2 the frequency of the center will be cut off. The upper image at 2x the center frequency will be cut off by the UV(low pass) filter.
That's good and all but you will have overlap at the extreme low and high ends of your usable spectrum from the images above and below. It's that overlap that causes your moire. Tightening up your bandpass cutoff frequencies will trim the overlap and effectively stop your moire but you are cutting your bandwidth down(as you already know) and limiting your true resolution.
(stolen from wikipedia)
notice the where the blue(spectra of interest) and green(nyquist images) overlap on the first plot. Since your IR filter is going to be much closer to the ideal cutoff point, the lower image doesn't overlap. Since your OLPF is designed for much higher bandwidth(full sensor bandwidth) but the video routine doesn't use full bandwidth, you get overlapping images. Moving your OLPF cutoff frequency will kill that overlap, as in the second plot.
http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem#Aliasing
If that were the case, you would LOSE that frequency and thus you would lose that specific color. Detail is not a function of frequency, it is a function of photons being resolved in a pattern. If you were to somehow reduce the wavelength(frequency) you would shift color. Does not compute.
That's because it's not "blurring", it's limiting spectal folding. I'll try to explain easily.
In theory, you have a CMOS sensor with RGB sensitive arrays which give you 3 images in 3 discrete spectral ranges. With the OLPF as both a IR and UV limiting device, you are technically creating a band-pass filter which trims the bandwidth of your signal into a usable range.
In all e-mag(electro-magnetic) spectra you have frequency folding which in layman's terms means that a frequency will have duplicate images of itself at 1/2 and 2x it's center frequency. Since IR is lower in frequency, the lower image at 1/2 the frequency of the center will be cut off. The upper image at 2x the center frequency will be cut off by the UV(low pass) filter.
That's good and all but you will have overlap at the extreme low and high ends of your usable spectrum from the images above and below. It's that overlap that causes your moire. Tightening up your bandpass cutoff frequencies will trim the overlap and effectively stop your moire but you are cutting your bandwidth down(as you already know) and limiting your true resolution.
(stolen from wikipedia)
notice the where the blue(spectra of interest) and green(nyquist images) overlap on the first plot. Since your IR filter is going to be much closer to the ideal cutoff point, the lower image doesn't overlap. Since your OLPF is designed for much higher bandwidth(full sensor bandwidth) but the video routine doesn't use full bandwidth, you get overlapping images. Moving your OLPF cutoff frequency will kill that overlap, as in the second plot.
http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem#Aliasing
If that were the case, you would LOSE that frequency and thus you would lose that specific color. Detail is not a function of frequency, it is a function of photons being resolved in a pattern. If you were to somehow reduce the wavelength(frequency) you would shift color. Does not compute.
shaocaholica
Well-known member
I think you've dug a bit too deep. The IR blocking properties of OLPFs are secondary. Its accomplished via a dielectric coating which has nothing to do with the primary purpose of limiting resolution. The resolution limiting is accomplished via optical "dispersion" via the crystalline structure of the OLPF.
http://en.wikipedia.org/wiki/Anti-aliasing_filter
Just be mindful that theres a lot going on in the OLPF stack. Its not all accomplished via one science if you will.
http://en.wikipedia.org/wiki/Anti-aliasing_filter
Just be mindful that theres a lot going on in the OLPF stack. Its not all accomplished via one science if you will.