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 :)

Canon 7D looks quite Nice

Thanks Jacek. I don't see how the 20micron grain size has any effect on DOF though - it could effect image sharpness, but not DOF, just as pixel size doesn't effect DOF in a sensor, just overall image sharpness. Now that can have an effect on the perception of DOF, but not to the magnitude that we see here, and because the visible grain structure is very fine to the eye, much finer than the display LCD on the back of the camera that has no problem adequately demonstrating very different DOF to that seen on the OVF.

I get what you're saying about the areal image though. That would indeed effect things.

Also what I'd read was that the ground glasses in modern DSLRS are not really a ground glass, but have a surface of micro-lenses and these lenses have an effective aperture greater than that of a fast prime, hence the issues we see. What do you think on that?

Graeme
 
The amount of information available on this forum is amazing.
I have 25+ years dealing with ADC, DAC, anti-aliasing filtering, reconstruction smoothing filters, over-sampling, etc for electrical signals. I have also been taking stills and video for just as long. I have not had to deal with aliasing problems for optics. I expect I hadn't identified the visual noise/distortion in my digital stills and digital video as aliasing problems.
My question is, can the aliasing distortion be dealt with by putting an optical filter on the lens that softens the resolution?
I ask this to improve my understanding.
If an external filter can be used to control the aliasing this of course won't improve the resolution, but it seems to me that the softer results would be better than the aliasing distortion. This external filter would be multiplied effectlvely by the focal length of the lens such that it would need to be "tuned" for each focal length. This would be a really bad solution for a zoom lens because it would have to be tuned for the shortest focal length, which would make the longest focal length terribly soft.
I look forward to a Red solution that addresses these aliasing issues by designing the sensor with the appropriately matched Optical Low-Pass Filter to get the maximum resolution with the lower distortion.

Best Regards,
Michael D. Floyd
 
Thanks for asking Michael!

Most cameras have optical low pass filters between the lens and the sensor to ensure that we don't get nasty aliasing on our images. Problem is though, that a physical OLPF is a very slow filter, and hence to get any decent cut at Nyquist, you also must be effecting details that in an ideal world you'd not want to be diminished. So there's a trend to use a less powerful OLPF than you'd otherwise use, but in reality, this doesn't cause too many problems....

...Until vDSLRs came along... As you know, downsampling is a sampling process, which means it needs a low pass filter before decimation. These vDSLRs appear to lack either adequate or any downsampling filter at all, leading to visible aliasing artifacts even on real-world material. It's doubly obvious on motion footage because aliases travel in the opposite direction to the motion of the object that is aliasing.

So yes, Canons have adequate optical filtering built in, but it's only adequate for stills mode. For video mode, the filtering is utterly inadequate due to the lack of proper downsampling filters in the electronics.

Graeme
 
Thanks Graeme.

I read into this that a filter put on the lens for video use, would have to soften the image an extreme amount to attenuate the sharpness enough to eliminate the aliasing distortion. I expect this is due to the limit on the optical low-pass being single order, and the inability to make multi-ordered optical filters?
This would explain why "final" resolution of the 3k Scarlet is about 2.5k. The DSP processing is performing the precision multi-ordered low-pass filtering. In effect, there is a mild over-sampling process. I have again learned something from this forum.
Regards,
Michael
 
Yes, if you wanted to eliminate aliasing artifacts on a 5D2 in video mode, you'd need an extra external filter. It would have to filter predominantly in the vertical orientation too, as due to the way that camera works, aliasing is much much worse vertically than horizontally, and horizontally the image is already soft.

High order filters, in software, require negative coefficients in the kernel. Lack of negative photons makes this rather hard to achieve with optical filters :-)

That said, sharp filters ring, so they're not always what you want either.

Final measured resolution of a 3k Scarlet will be less than 3k due to the optical filtering. That does indeed mean that there is a slight over-sampling, which is the correct approach. With the 4k RED One, we measure around 3.2k. This means almost complete lack of any aliasing artifacts and a natural looking image. See RED One zone plate: http://www.cinematography.net/natress-red-res.html

Graeme
 
Those elusive negative photons.|:)
Making sharp filters that don't ring for electrical signals, is the motivation for using over-sampling and making the sharp filters with DSP. There seems to be a similar strategy for video.
Michael
 
hi mr. nattress,
which external filter do you can suggest for reducing aliasing's problem in canon 5d?
thanks
ivan
 
Ivan - On a 5D2 you'd have to replace or add to the existing OLPF, and it would have to be about 3 times as strong as the existing one vertically. I doubt that would be an off-the-shelf part. It would utterly screw up still imaging though.

Michael - low pass filters don't get sharper than a gaussian before they begin to ring. A good OLPF will give you a mostly band limited signal, and from then on you can use any DSP filter to downsample that you want. That gives the most flexibility. Yup, it's like oversampling. However, with images I tend to find that not downsampling and projecting at native resolution looks better (more natural) than downsampling to say 2k and projecting 2k. Just like 192khz/24bit audio sounds better as such rather than as downsampled to 44.1khz /16 for CD, even if the d to a oversamples on the output.

Graeme
 
I don't think the effect has anything to do with polarization.

Graeme
 
Here's a link to some 7D at IBC images http://bit.ly/wVApr
regardless of aliasing, codec, skipped lines etc the images I've seen so far look quite nice. It looks like this cam is set to be popular among the indie film set.
 
If someone wanted to get some nice lenses to use on the 7D that could later be taken off and used on the Scarlet S35, think some Zeiss Compact Prime ZF's with a converter to the Canon mount would be a nice idea, or would some EF lenses and a Birger mount (for use on the scarlet) just be the best idea?
 
Wesley Scoggins said:
If someone wanted to get some nice lenses to use on the 7D that could later be taken off and used on the Scarlet S35, think some Zeiss Compact Prime ZF's with a converter to the Canon mount would be a nice idea, or would some EF lenses and a Birger mount just be the best idea?
Click to expand...

I've been wondering about the same. The Bierger route sounds like a lot of work and cost that you might not get back.

Definitely the Zeiss ZF's if you can get a set of those. I'd make sure to get the 28mm and the 50mm f/2 due to CA on the f/1.4, unless you are getting the 100mm which covers you on macro so maybe go ahead and deal with the f/1.4 50mm.
 
Well I am thinking of starting out with three, a wide-angle like the 28, a 50, and the 100, (you kind of read my mind). Which should generally meet about 95%+ of the shooting circumstances i'd run into.

But I am thinking get the 7D now, start getting used to a DSLR-style video camera, then when the S35 comes out, snatch it up and I can have my A and B cam ready, with lenses for both, and some experience with the same type of camera.
 
Wesley Scoggins said:
Well I am thinking of starting out with three, a wide-angle like the 28, a 50, and the 100, (you kind of read my mind). Which should generally meet about 95%+ of the shooting circumstances i'd run into.

But I am thinking get the 7D now, start getting used to a DSLR-style video camera, then when the S35 comes out, snatch it up and I can have my A and B cam ready, with lenses for both, and some experience with the same type of camera.
Click to expand...

28 isnt really a wide IMO on the 7d

BTW, getting nikon mount you can go to canon, or nikon

other way round does not work, keep options open with nikon mount I would say

different arguments apply for non manual primes where electronic interface could be cool as focus rack etc is developed

S
 
But the 28mm is gorgeous. And ZF's get much slower as they go wider. Their 18mm is a f3.5.

Their 21mm is decent at f/2.8 but then I'd stick with the 28mm because of the image. Maybe pick up the 21mm later.

And you could use them on a 5D with really wide FOV and speed. Or a rented FF35 Red.
 
You must log in or register to reply here.
Back
Top