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 :)

Panavision videos...

Look, Listen, Learn think for yourself

Look, Listen, Learn think for yourself

Its good to listen, learn and form your own opinions Red has a vested interest in protecting itself, so does Panavision, so does Dalsa etc. Their is a "new" bayer pattern AGAIN invented by Kodak, this bayer pattern corrects some of the known issues with the original and will first appear in low end devices in 2008 like mobile phones eventually going up the scale.
The MTF video is very interesting and points out that video is very different to film and it has a bearing on finished images (diffrent cycles). As stated elsewhere the CMOS chip has its own light gathering lenses and these can affect the final image as light in the center hits these lenses differently to the edges.
No matter what bullshit manufacturers say about their algorithms they only alter what they can phyiscally get from the chip, if the chip didnt get it then no amount of manipulation will change that and the manipulation is a compromise remember that each time you get a software update as the gurus like to blind people with science as its in their interest to do so.
 
The new Kodak pattern makes the image brighter - that's it. That's why it's aimed at small sensor / pixel devices first. They need all the SNR help they can get.

Nova, for the rest, I don't know what you're saying.....

Graeme
 
1st- Panavision deserves much credit for what the do (and have done) for the industry. But that does not mean that their technical marketing piece we are discussing is right.

2nd- please understand that all CMOS sensors can no longer be lumped together in one big group with the assumption that they perform the same, act the same, all have the same characteristics, advantages and problems. Blanket statements about CMOS are now much more limited than in the past. The differences in individual CMOS designs and capabilities now far surpasses the similarities.

Jim
 
So what is Panavision going to do with RED?

Are they going to rent lenses for RED? Are they going to purchase RED cameras to rent their own Pana-REDs?

Since they were servicing "The Argentine", I thought they could develop a business plan to work with RED cameras.

Jim, you probably have talked to them, due to Soderbergh s films. WHAT ARE THEY GOING TO DO ABOUT RED?
 
Just saw the videos. I have to say, it seems like they went to a lot of trouble to carefully explain image sensors just so they could do a rushed critique of a bayer pattern sensor in video 6 and then turn around and praise their own sensor in the last video.

I understand why some people have said we shouldn't be so quick to dismiss the whole seminar.
I found a lot of it to be very interesting. However, if they had called the presentation "Why we think RGB Stripe is superior" that would be one thing.

Instead, they call it "Demistfying Digital Camera Specifications" and present it as something like a PSA for the motion picture industry. (And they get their facts wrong as Graeme has pointed out).

It's the guise of objectivity that gets me.
 
I am surprised that people are so much disgusted by Panavision videos. I personally thought they were very well done and were very informative. Many of the people that are complaining about it did say that they learned from it. You tell me in overall scope is it good to have these videos posted there on the Internet so that people can learn about the issues that plague digital processing of sensor signals or they should not have been there because some sensibilities were hurt?

I personally would vote for the videos to stay. They are educational, instructive, and explain what Red should also do, and as Graeme Nattress mentioned they will eventually do.

One thing is immediately apparent from the presentations that the two companies who did it clearly thought that they are the market leaders in this area. And in some sense they may be true. But the point to ponder is to what gives somebody that much confidence that they think they are clear market leaders? It has to go beyond simple assembly and packaging of products. Fanboy statements like "Kodak/film is dead, Panavision / Canon are shaking in their boots, Red is the way to go" notwithstanding, what are the elements that shall give Red the respect in the development and scientific community that, say Canon enjoys, or for that matter Kodak and IBM demand and rightfully deserve?
 
Joofa, that's a great balanced post. Of course the overall scope of the videos is great. I just wish as John and Larry are expert on lenses, MTF and such like, that similar care had been taken on the section on Bayer patterns. If Larry had got one of the Canon DSLR guys in to do the Bayer bit, I'd have been thrilled to watch it!

Graeme
 
Thanks Graeme. I agree with you. I am sure it must be frustrating for Red to see how some portions of the talks were presented. I guess we shall wait for similar presentations from you and other folks at Red.
 
Argh!~ my last post was post #1920. I want to post again to move away from HD resolution :-)

Graeme
 
Almost halfway to 4K. And almost 2/5ths of the way to 5K...

Keep em coming Graeme. :)
 
In general I thought the videos were informative. The basic philosophy of the presentation is "what we want is good 4:4:4 1920x1080p". It is quite obvious that RED and GENESIS are both very well suited to this task. Except, the RED philsophy is "we want to get the most out of our sensor" followed by "we can do better than 4:4:4 1920x1080p".

In short, they deliberately avoided saying how a 4k Bayer array was awesome at producing 4:4:4 1080p, and focused on how a 4k Bayer array doesn't produce "perfect" 4:4:4 4k. It was a strange double standard.

The other thing I thought was weird was their discussion of RGB filter spectral cross-talk in Bayer patterns, as if they thought it was a bad thing. If you understand how the human eye works, you would realize that our perception of colour comes largely from cross-talk analysis between our "Long" "Mid" and "Short" wavelength cones, each of which is sensitive to a rather broad spectrum. You get much more information when your filters aren't particularly monochromatic.
 
Spiff said:
The other thing I thought was weird was their discussion of RGB filter spectral cross-talk in Bayer patterns, as if they thought it was a bad thing. If you understand how the human eye works, you would realize that our perception of colour comes largely from cross-talk analysis between our "Long" "Mid" and "Short" wavelength cones, each of which is sensitive to a rather broad spectrum. You get much more information when your filters aren't particularly monochromatic.
Click to expand...

Human vision has overlapping responses because I think it takes advantage of redundancies later in high-level processing. However, for your sensor spectral overlap, typically they are removed as early as possible, by design. When you do white balancing, the negative co-efficients in the matrix also help to reduce the cross-talk factors.

A good source for human vision is Boynton (yes with a B, and not Poynton with a P).
 
The ideal spectral response of a camera would be similar to the spectral response of our visual system. One weird thing is that spectral response is different depending on what part of the eye is involved (!).
http://www.filmlight.ltd.uk/documents/FL-TL-TN-0101-StdColourSpaces.pdf

You can assume that the spectral response of the average human is that of the 1931 CIE measurements, and design a camera so that its sensors have a spectral response similar to the CIE color matching functions. In a sense, that would be the ideal spectral response (assuming that you didn't have to compromise in sensitivity, cost, etc.; of course in practice you do).
 
GlennChan said:
The ideal spectral response of a camera would be similar to the spectral response of our visual system.
Click to expand...
That's like saying "there's no point shooting 4k, because I can't resolve all the information in 1/24 s". Ideally the sensor acquires better spectral information than our eyes.

Furthermore, it doesn't have to do that in the same way we do. You could collect hundreds of colour channels as opposed to just 3 if you wanted - measuring the image spectrum down to a nm, well into the UV and infrared. You could then do colour filtering in post to modify the image.

Heck, you could go a step further, and acquire not only the spectrum but the polarization of the light with your sensor. Then you'd be able to do it all - colour filtering, spectral balancing, polarizing filters - all in the digital realm.

And while I'm dreaming... I'd like a pony. Actually, scratch that, I'd rather a Bösendorfer 290.

Once the spectrum is acquired and known, the next problem is re-generating that spectrum for display - or at least fooling our eyes into thinking it's seeing that spectrum.
 
Theoretically, there might be a small advantage to going with a system with more than 3 primaries, both in acquisition and in display. This would avoid any metamerism (assuming perfect capture + display).

Multiple primaries might possibly have some interesting keying applications.

But in practice, I highly doubt that would take off.
A- We are pretty insensitive to metamerism. If you just take the case of us looking at real world objects (not watching TV), looking at our world is not "color accurate" because there is metamerism from different lighting. e.g. at golden hour, things like golden. So I don't think we pay attention to color inaccuracies as they naturally occur (except where the inaccuracy is extreme, like the extreme IR with stacked NDs example).

The price you pay for a multi-primary system would be very high and it wouldn't work with existing infrastructure.
 
the ideal camera

the ideal camera

Spiff said:
You could collect hundreds of colour channels as opposed to just 3 if you wanted - measuring the image spectrum down to a nm, well into the UV and infrared.
Click to expand...

This is a fun topic! I think Spiff's idea is great for a what-if scenario, just acquire all the information that physics permits. Might as well specify a quantum efficiency approaching 1.0 too (recording every photon) for the theoretical best results in low-light situations. For now this may imply something the size, price, and lead time of the Hubble telescope, but technology keeps advancing...!
 
I, too found the videos very informative. The main message was that a 4K camera doesn't give you a 4K image on screen in a theatre. All elements that treat an image reduce resolution. Even if you use the best lens at 4K, a bad projector lens can turn your 4K source into a 1.5K image on the screen.

Panavision and RED have different approaches to digital cinematography. For Panavision the priority is clean keys for visual effects while for Red the priority is maximum resolution for what the human eye can see.
 
But the RED approach does not lead to non-clean keys....

Graeme
 
You must log in or register to reply here.
Back
Top