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 :)

sony

Andrew M. said:
Yes,
human eye DR without the pupil size change is 6.5 stop.
http://en.wikipedia.org/wiki/Human_eye

So when you look on static image, without allowing your eyes wander around the pictures peripheral you can see only this, 6.5 stops of image DR.

Maybe watching IMAX on 180deg spherical screen will make use of higher DR achieved the way of pupil change.
Click to expand...

I don't think this value means much, practically. The human eye has some incredibly sophisticated auto exposure and auto focus mechanisms that adapt faster than we can notice. So, even if at one instant the eye theoretically only sees 6.5 stops, our brain practically sees 20 stops or so in most situations. So, human vision should be rated at 20+ stops, and this should be the target for cameras.
 
Perhaps a live 4K output module will be made for Epic?! Perhaps RedCode could be sent over fiber to a black box decoder at the truck that would output the 4K using multiple Rockets/Cuda GPU tech/Atomix cards in parallel/repurposed camera ASICs, etc. Just musing...

Would love to do 4K IMag for concerts/shows and obviously 4K sports is a Holy Grail. Fun times...

Cheers - #19
 
Blair S. Paulsen said:
Perhaps a live 4K output module will be made for Epic?! Perhaps RedCode could be sent over fiber to a black box decoder at the truck that would output the 4K using multiple Rockets/Cuda GPU tech/Atomix cards in parallel/repurposed camera ASICs, etc. Just musing...

Would love to do 4K IMag for concerts/shows and obviously 4K sports is a Holy Grail. Fun times...

Cheers - #19
Click to expand...

That sounds almost exactly like the RED ONE prototype rig with off-board recording that they told us about back in the day. I believe they said it was something like "a black refrigerator tethered to the camera by fiber optic cabling". If I remember correctly, they took it out to a nearby skate park and tested it. I don't think they ever posted any photos of it but it sure sounds like a sight to see.
 
Jannard said:
I hate it when companies tell you "all you need is this" when they are quietly "working on that". And I'm not talking about Sony.

Jim
Click to expand...

There's gonna be more companies than Sony and Canon that gets out of the 4k closet the next 12 months...

Or that was one of my initial thoughts when I started testing a very nice 1080 camera this fall...
 
Dear Sirs!

Where can I find the best introduction to the basic principles and concepts of "bayer"and "debayering"? Particularly in relation to RED digital workflow.

With the upcomming Scarlet, I think there will soon be a need for more basic introductions (I work with film talent development in Denmark).

Jim's explanation of HDRX™ in the Recon area is a great and useful example.

All the Best,
Simon Falkentorp
Premiere Pro ACE
 
For the most part, the underlying software in the SDK and REDCine-X, and the internal hardware of the camera handles all things to do with the demosaicing of the sensor data. There's not much you actually need to know unless you're the guy writing the code to do the demosaic (like me). What the software delivers to you are nice images, and as for actually how it does that is not necessary to know to understand how to get the best out of the camera.

Now.... I'm happy to discuss the subject (start a new thread?), but the important thing is that it's really all taken care of for you.

Graeme
 
Michael Olsen said:
Just out of curiosity and a question to those who may actually know - what does an S35 6Kx3K CCD cost? They definitely kept hammering the CCD aspect, which at that size and those pixel counts makes me think this will definitely be a rental camera...
Click to expand...

If you listen carefully to the presentation, he didn’t mention CCD 6Kx3K.
He mentions and the drawing presented the 6Kx3K sampling method.
Look on the description on the top.

“RGB sampling method”

So we still don’t know if the 17MP sensor is CCD or CMOS, correct me if I am wrong.
There are references to CCD pixel pattern versus CMOS pattern.
In this context there is reference to CCD sensor though.


EDIT - Actually in the 9:04 he mentioned that Sony Developed CMOS sensor 17MP.
So the refference of CCD sensor was just for 6.2MP sensor used in F35 and 9000PL

In any case 17MP on S35 with square pixel, diamond pixels or round pixels arranged in RGB pattern or Bayer pattern doesn’t really matter. It is still 17MP sensor.

What does matter how much empty space you have in between pixels and how good are you in massaging this data coming out of it. We are presuming that both hardware sensors technologies are most up to date and equal.
But at the end what really does matter before you start processing the sensor data is how much DR you can squeeze out from these buckets.

If you are collecting rain to the round buckets or square buckets or square buckets but placed diagonally and you would like to be very efficient to collect all the rain, you have to make sure that there are as little gaps as possible between these buckets.

Or you collect the rain twice in to the same bucket and you get double of rain:-) HDRx™™™™™™™™™™?
 
Graeme Nattress said:
For the most part, the underlying software in the SDK and REDCine-X, and the internal hardware of the camera handles all things to do with the demosaicing of the sensor data. There's not much you actually need to know unless you're the guy writing the code to do the demosaic (like me). What the software delivers to you are nice images, and as for actually how it does that is not necessary to know to understand how to get the best out of the camera.

Now.... I'm happy to discuss the subject (start a new thread?), but the important thing is that it's really all taken care of for you.

Graeme
Click to expand...

Ahem…cough cough…

Maybe I take this too literally but reading between the lines, Graeme you are actually offering an early morning fire side chat to discuss various de-mosaicing algorithms?

Hey, I can ask some really moronic interview style questions that I have always wanted ask? Plus some technical ones that might put people’s mind at ease…

Suggested title of thread?

Cheers,

Eric

Algorithms are much sexier than Sony…
 
The measured resolution is lower than the pixel count mostly due to the optical filtering that is necessary in any type of sensor to avoid nasty aliasing.

Graeme
 
...and inversely proportional to the number of Guinness your 1AC had last night.

I loved the old Sony argument about how 3x1920 CCD imagers somehow added up to 6K...especially when paired up with a HDCAM deck...at 3.1.1...
...I still hear that one every now-and-again.
That must mean my high def Panasonic 3MOS handy-cam has 6K resolution!

CMOS trades some (localized) colour fidelity for resolution. We get that.
Resolution is more perceptible to the eye.

OK so they us a high photosite count and then sample at half the resolution.
They trade off resolution for colour fidelity. Ok, fine.
But you can't claim a 17 mp image, you've divided by two... live with it.
But maybe the colours are very good for a standard 8meg demosaic...
...but its not a normal 8 meg sensor.

Global shutter is a good thing though.
 
I'm curious (as a layman): Is the Q67 illustration of a single pixel sharing different colors an accurate depiction of what takes place? How are the opposing colors and their common edges defined?
 
The 3x1920 would hold weight if the image from one actually measured some good proportion of 5760, but when they measure less than 1920 and generally have some aliasing there too, the argument doesn't hold any water, never mind photons. Same for the 3x1920 F35 argument. It measures less than 1920 and has aliasing and chroma moire.

Bill Q67 is Bayer pattern, so interpolation is needed to define colours on edges. With a good algorithm this is not very much of an issue as the luma resolution which is what defines the visibility of edges is the main factor.

Graeme
 
Thanks G, I mistook the Q67 bayer pattern overlay lines (illustration) as suggesting something else, what a dope.
 
Jannard said:
I hate it when companies tell you "all you need is this" when they are quietly "working on that". And I'm not talking about Sony.
Click to expand...

Except that everyone does that, it's simply part of technology marketing. Nobody knows what the future will bring, so companies generally will develop what they can and have a future roadmap of hopefully achievable goals. But until those goals are indeed achieved, what is currently available will be offered as an optimal solution. And while you can point to (insert unnamed German camera manufacturer here) as possibly doing this right now, you could also look at Red about 1 or 2 years ago talking about 4K being "all you need" and then quietly working on 5K and beyond. That's not a criticism, it's simply a fact of being in a fast moving, technology oriented business.
 
M Most said:
And while you can point to (insert unnamed German camera manufacturer here) as possibly doing this right now,...
Click to expand...

Mike,

you mean rumors about that Arri is doing 4K project code named "Alexander The Great" that should be Alexa successor or...???!!!
 
M Most said:
Except that everyone does that, it's simply part of technology marketing. Nobody knows what the future will bring, so companies generally will develop what they can and have a future roadmap of hopefully achievable goals. But until those goals are indeed achieved, what is currently available will be offered as an optimal solution. And while you can point to (insert unnamed German camera manufacturer here) as possibly doing this right now, you could also look at Red about 1 or 2 years ago talking about 4K being "all you need" and then quietly working on 5K and beyond. That's not a criticism, it's simply a fact of being in a fast moving, technology oriented business.
Click to expand...

Agreed... sorta...
It's been no secret that RED was going 5k since I think NAB 2008... That was before most of the "early adoptors" of R1 even had their cameras...

Initially, some felt a bit "sony"'ed back then...

But then again both the development and the actual specs took some more time to figure out...

And RED was the party that initially (and long before I got my R1) made me aware that a 4k sensor won't deliver a 4k image.

Off Topic:
What has been continuosly fun is watching all the guys not delivering a 1080 image bashing RED for not delivering a 4k image... Yup. Different story. Still fun...

While at the same time, the comment on how the "not 4k cam" delivers "quite a nice 1080" is a factor with less focus...

Back on topic:
Marketing is marketing and anyone who believes that anything makes anything obsolete in the tech-division is rathr a fool and has little historical knowledge...

And I know that the strongest feeling I was left with when testing some German camera and an upgraded version of a camera with a colorcode, was that these cameras both will be replaced rather quickly...As in, when compared both had something that the other needed badly... Left on their own (out of a comparative setting), both are quite nice, though...
 
Eki Halkka said:
True, but... isn't the thing that counts actually the photosite area used for an individual end pixel?
Click to expand...

The only thing that counts is the final image, the end result. No approach or design is perfect, every one has tradeoffs, and every company chooses their own approach based on the combination of intended markets, feature priorities, and resulting image characteristics that they find most important. Red has their target market (primarily digital cinema, but also some other things) and their choice of tradeoffs is based on the requirements of that market segment. Sony has their target market (primarily consumer and broadcast, but also some other things like digital cinema) and makes their choices based on that. This is not a contest to see who is smarter, and it is not a contest, quite frankly, to see whose approach to a large single sensor design is "better." Different product criteria breed different image sensor designs, and only the end result - the image - means anything to anyone other than the designers and engineers. Nobody has seen the results of the design choices Sony has made for these new products, just as three years ago nobody had seen the result of the choices Red made. People here ripped apart those who criticized what the Red One would likely be able to do, and now here they are doing exactly the same thing to a company with a pretty darned good technology track record.

There is no one, singular approach to almost any technological problem. There are multiple approaches, all with a different set of tradeoffs. They all have advantages and they all have disadvantages. You can like and believe in one over the others, but that doesn't mean you've got some kind of lock on universal truth. Innovation is all about disproving conventionally accepted wisdom and hopefully coming up with something even better. If you keep an open mind, sometimes things can actually surprise you. I'm not saying that's the case here, but what's the point in ripping something apart that you haven't even seen yet? Personally, I don't know if Sony's approach will prove to be any better than conventional Bayer or not, or if it will have processing speed advantages or not. But I am willing to at least wait until it's shown before I form a final opinion.
 
You must log in or register to reply here.
Back
Top