The secret to behind the Panavision Genesis

[Ted Johanson]

Greetings everyone! This is my first post, but I have been reading the forum for

around a year now. Anyway, back to the topic...

 

Many people seem to think that the Panavision Genesis has made giant

technological strides in solving the problem of limited contrast range for digital

acquisition systems. While the results themselves have been improved, the

actual sensor technology has not improved (at least nowhere near as much

as they'd like you to believe).

 

"Well then, how were the results improved?"

 

To answer that question, let's first consider the specs of the Genesis. The CCD

is claimed to be of the RGB CFA type and has 12 million photodiodes (sensors).

The output "video" resolution is 2 million pixels (6 million photodiodes used,

considering the RGB CFA design).

 

As you've noticed, when using an RGB CFA type sensor, you only need 6 million

photodiodes to produce a 2 megapixel color image.

 

"So, what about the other 6 million photodiodes left on the Genesis' sensor?"

 

Here's where the "magic" comes in. The first 6 million photodiodes are never

correctly exposed, they are actually always underexposed to maintain highlight

detail. The other 6 million photodiodes are always overexposed to maintain

shadow detail. Both exposures are made at exactly the same time. Thus, two

separate records of the exact same image are produced; one underexposed,

the other overexposed. These two records are then combined digitally, in-

camera, to produce a single image.

 

So there's your answer to the "excess resolution" of this camera. All of those

photodiodes are actually needed to produce the two-megapixel image...take

any of them away, and you WILL degrade the quality of the output "video"

in either contrast range or resolution. Try to increase the output "video"

resolution, and you will lose contrast range. All in all, it is a very inefficient

design; but effective.

 

Now, in reality, it doesn't work quite as simply as I have explained it above.

I used phrases such as "first 6 million photodiodes" in the hopes of providing

an easier to understand explanation. This seems to imply that the top half of

the sensor underexposes and the bottom half overexposes. Of course it

wouldn't work that way. It actually works on a line-based system. The

sensor underexposes every first line of photodiodes and overexposes

every second line of photodiodes.

 

For those of you who don't believe that this works, try it for yourself. You'll

need to take two pictures; one underexposed and one overexposed. Bring

both images into an image editing program such as Photoshop. Layer one

image on top of the other. Then reduce the layer opacity of the top image

to 50%. And there you have it, an image with increased contrast range.

 

No, I didn't hack into PV's computers or anything like that. I just figured this

one out all by myself. It's the most reasonable explanation for the resolution

differences.

 

I hope you enjoyed reading my post!

 

 

-Ted Johanson

[Landon D. Parks]

What proof do you have to back up this statement? What makes you think that the CCD works the way you think? Are you a video technition? Have you examined the Genesis well? Was you involved in the design process?

 

Somehow you need to back up your "Claim". It's sort of like me saying "I know how the Dalsa sensor works!" When really I have no freaking clue.

Edited March 14, 2005 by Landon D. Parks

[Phil Rhodes]

Hi,

 

> Are you a video technition?

 

I don't know, but he might know how to spell "technician", and he'd be quite within his rights to take exception to criticism from anyone who doesn't!

 

For what it's worth, the concept of having compound photodiodes to improve dynamic range, if that is indeed what the Genesis is doing, is not new anyway.

 

Phil

[Ted Johanson]

I have no concrete proof that it works this way...but I do have the knowledge

obtained from many years of studying CCD and other digital acquisition technology

in great detail. Do the simple math yourself. Isn't it a great coincidence that

everything just happens to fit perfectly? Why do you think they would build

a sensor with 12 million photodiodes on it when they knew the system as a

whole would only output 2 million pixels?!

 

Anyone care to attempt to give another explaination for this overkill? Perhaps

you, Mr. Parks?

 

 

-Ted Johanson

[Landon D. Parks]

I don't know, but he might know how to spell "technician", and he'd be quite within his rights to take exception to criticism from anyone who doesn't!

come on Phil....I'm just wondering how he knows this? How does he know what no one else seems to know or can find out? Don't you find that bit fishy?

 

Why do you think they would build

a sensor with 12 million photodiodes on it when they knew the system as a

whole would only output 2 million pixels?!

maybe they wanted 12mp so when the technology comes about, they can use ALL 12mp instead of 2mp? I'm not a video technician so I don't know a lot about it. I just find it strange that you know so much about the way the camera works, when no one else can seem to find out much info at all about.

 

Actually, I'm not going to make any "Predictions" on stuff I dont know. So no, I don't care to explain my version of how it works, because I don't know.

Edited March 14, 2005 by Landon D. Parks

[Ted Johanson]

maybe they wanted 12mp so when the technology comes about, they can use ALL 12mp instead of 2mp?

 

You're talking about a scalable camera, that's something I'm sure everyone would

like to have. Nevertheless, I don't see any mention from Panavision about the

Genesis being scalable. Has anyone read otherwise?

 

Also, isn't it funny they chose to use an RGB color filter array rather than the very

widely accepted Bayer pattern CFA? The RGB color filter array just happens to work

better for the method I described.

 

For what it's worth, the concept of having compound photodiodes to improve dynamic range, if that is indeed what the Genesis is doing, is not new anyway.

Exactly! There's nothing new about this. Lot's of people have been using the same

general method for a long time.

 

And let's not forget, I did conclude my first post by saying...

I just figured this one out all by myself.

In other words, I'm not giving a 100% guarantee that the camera truly does

work the way I described.

 

 

-Ted Johanson

[Eric Steelberg ASC]

I'm afraid, Ted, your speculating. According to everyone I've spoken with at Panavision, it is intended to be scalable...enabling them to have a built in upgrade path once recording technology improves.

 

Your idea is not a bad one, but old nonetheless. Fuji has a patent on the technology you mention and incorporates it into their SuperCCDs used in digicams such as the S2 and newer S3. They developed this based on the way rods and cones in the human eye work. The Genesis is actually a Sony camera with a Panavision designed body and optics.

Edited March 14, 2005 by Eric Steelberg

[Ted Johanson]

I'm afraid, Ted, your speculating.

 

Yes, I know. I believe everyone here would know that if they read my first post

in it's entirety.

 

According to everyone I've spoken with at Panavision, it is intended to be scalable...enabling them to have a built in upgrade path once recording technology improves.

Scalable...yes. But scalable in exactly which ways? Did they specifically say the

resolution was scalable or are they just being misleading by not telling people

that only things such as bit depth and compression quality are scalable?

 

I know all about FujiFilm's SuperCCD. It works using a different method than the

one I explained. Their photodiodes for capturing highlight details have a small

area, while the photodiodes for capturing shadow details have a large area.

The physical size of each photodiode has a direct impact on the sensitivity.

Thus, actual size of the photodiode is what is causing the under/over exposure.

I made no mention of different photodiode sizes in my specualtion about the

Genesis. I believe that it works by controlling the gain of individual lines.

 

Also, nearly every digital still camera manufacturer uses a Bayer pattern

CFA. If my memory serves me correctly; Kodak, Canon, Nikon, Minolta,

Sony, and HP all exclusively use Bayer CFA's for their digital still cameras.

Why would they use that type of pattern if it wasn't better, especially when

it requires an extra photodiode per pixel?

 

Let's assume the resolution truly is scalable. Why did Sony and Panavision

choose to use an RGB CFA if it isn't as accurate as a Bayer CFA? Clearly

resolution wasn't the biggest concern. If Sony and Panavision were trying

to achieve the most accurate sampling possible, the RGB CFA method was

NOT the best way to go. Obviously, it was chosen for another reason. It's

more accurate than a Bayer CFA would be; but ONLY IF you are using my

speculated multi-sampling method. It would also certainly allow them to

work-around FujiFilm's patent.

 

Go ahead and call Panavision again...ask them specifically if the resolution

is meant to be scalable; and make sure you are talking to someone who

truly knows this camera.

 

 

-Ted Johanson

[Ted Johanson]

PS. I already knew about the cooperation between Panavision and Sony.

It's just like the cooperation they had when creating the F900; and we all

know how misleading they were with that camera!

 

 

-Ted Johanson

[Elhanan Matos]

I don't remember Panavision ever saying that their camera was scalable. I do remember them saying that they will be working on a way of getting uncompressed HD out of it later on.

[Kevin Zanit]

But the bigger question is: Who cares?

 

I don't understand the purpose of your thread. Test the camera for yourself, if you like it, then congratulations you found a new tool for an old job. If you don?t like it, then move on.

 

A company touting its new product as something different and better than their competitor?s is nothing new, its just business.

 

The sensor in the camera is a very old Sony chip from what I am told. Your theory could be 100% correct for all I know. It is an interesting camera system. The little that I have played with the camera, it looks very promising and if it is works better than what exists now, then I can?t complain.

 

 

Kevin Zanit

[John Sprung]

The Genesis is actually a Sony camera with a Panavision designed body and optics.

<{POST_SNAPBACK}>

.... or you could think of it as a Panavision camera with a Sony chip and electronics in it. ;-)

 

It is interesting that the sensor has 1920 RGB triplet across, and 2160 rows of them. It could be thought of as 1920 x 1080 groups of 3 wide by 2 high. Maybe they do alternate high and low sensitivity, I really don't know.

 

But if the idea is to mimic rod and cone vision, the way to do it would be to use a modified Bayer pattern, with one green of low sensitivity combined with the red and blue to act as a cone, and the other of high sensitivity and covering a wider band of wavelengths to simulate a rod. Maybe even move the "rod" spectral peak down to 505 nanometers and give the camera a Perkinje shift just like human vision. ;-)

 

 

 

-- J.S.

[Ted Johanson]

But the bigger question is: Who cares?

Well, I suppose anyone who is interested in how this camera works (or may work).

Everyone seems to want to know.

 

Some people seem to think the extended contrast range is some sort of great

technological breakthrough; but if I'm correct, it's nothing new. It could be quite

simple. That's my point.

 

On the other hand, this camera does have significant advances in the signal

processing area. That is why it can be "over-cranked" to 50 FPS for example.

I'm sure everyone already knows that though.

 

With that said, yes, it is a great tool; certainly better than it's predecessors.

 

 

-Ted Johanson

[Kevin Zanit]

From what I hear, the 50fps feature is not functional on the current one in existence (nor is a few other features, such as playback on camera).

 

 

Kevin Zanit

[Ted Johanson]

That must be a limitation of the recorder. If they can make the recorder "faster",

then 50 FPS would become available. Perhaps that is one of Panavision's ideas of

the camera "resolution" being scalable.

 

 

-Ted Johanson.

[Elhanan Matos]

That must be a limitation of the recorder. If they can make the recorder "faster",  then 50 FPS would become available. Perhaps that is one of Panavision's ideas of the camera "resolution" being scalable.

 

The tape transport used in the Genesis is the EXACT same tape transport in the Sony SRW-1, actually the "Magazine" on top of the Genesis IS the Sony SRW-1. So yes the recorder can spin faster (twice as fast to be exact).

[Guest Jim Murdoch]

Thus, two separate records of the exact same image are produced; one underexposed,

the other overexposed. These two records are then combined digitally, in-

camera, to produce a single image.

 

So there's your answer to the "excess resolution" of this camera. All of those

photodiodes are actually needed to produce the two-megapixel image...take

any of them away, and you WILL degrade the quality of the output "video"

in either contrast range or resolution. Try to increase the output "video"

resolution, and you will lose contrast range. All in all, it is a very inefficient

design; but effective....

 

....No, I didn't hack into PV's computers or anything like that. I just figured this

one out all by myself. It's the most reasonable explanation for the resolution

differences.

 

I hope you enjoyed reading my post!

-Ted Johanson

<{POST_SNAPBACK}>

 

Actually, that's damned good. I don't know whether you're right or not, but even if you are, it still leaves me wondering why Sony/PV didn't simply tell us all this in the first place! If you can work this out, I'm sure Sony's competitors could as well:-)

 

In practice there would be a number of limitations to this technology, and although it would theoretically improve the dynamic range a bit, it still wouldn't take it to anywhere what you get from film. With film the entire dynamic range is still stored in the emulsion, whereas even with this improved technology you'd merely be "massaging" the image data to fit into a 10-bit range. And that's a one-way process.

 

But hey, this is the first time anyone here has come up with something other than "secret processing technologies unknown to anyone else" or "if you think it looks like film, then what's your problem?"

[John Pytlak RIP]

FWIW, film has been blending different speed sensors (silver halide grains) to extend latitude and adjust the characteristic curve for DECADES. Sometimes these are blended into a single layer, and sometimes individual layers (fast-mid-slow) may be used for each color. Modern color films may have over a dozen individual emulsion layers coated.

[Ted Johanson]

...it still leaves me wondering why Sony/PV didn't simply tell us all this in the first place!

Sony and Panavision don't want to tell anyone because if they did, their

camera wouldn't look so technologically marvelous! And, of course, a

technologically marvelous camera is good for business.

 

Also, if you told a potential buyer that your camera had a 12 megapixel

chip and that it can only output a 2 megapixel image---AND you didn't

give them a reason for the difference---what do you think the buyer is

most likely going to assume is the reason for the difference? Obviously

their first assumption would be "it must be scalable for when recording

technology is improved". That would make their camera more attractive

to the buyer. The buyer would then think "Wow! It's got this improved

contrast range AND it will be able to output 12 megapixel images some-

day!"

 

Those are pretty good reasons for them to not reveal the specifics of

their camera.

 

I'm sure some of you will disagree with me on this, but I don't think

Sony is the best business partner for Panavision. We all know how

misleading and erratic Sony can be (if anybody needs examples, let

me know). Here, I'll give this example right now...oh wait, I can't. :unsure:

Okay, does anyone else here have enough memory capacity to

remember all of the different media formats Sony has created or

helped to create in the last 30 years? :D

 

 

-Ted Johanson

[Elhanan Matos]

If you want to know how the Imager in the Genesis works just look up vertical stripe filter CCD's. I think its the same kind of filtering that was in the old Sony Trinicon cameras.

[John Pytlak RIP]

If you want to know how the Imager in the Genesis works just look up vertical stripe filter CCD's.  I think its the same kind of filtering that was in the old Sony Trinicon cameras.

<{POST_SNAPBACK}>

 

Weren't those old cameras quite prone to flaring artifacts from bright speculars? Some weird aliasing artifacts? And easily damaged by shooting toward the sun?

 

http://www.mancini99.freeserve.co.uk/Sony_1640_b.html

[Elhanan Matos]

I heard they had alot of issues, and so did the early RGB Striped CCD's. I only mentioned the trinicon camera because they used similiar filtering. I guess you could say the Trinicon is like the tube version of the Genesis's sensor.

[Phil Rhodes]

Hi,

 

Really, it's not a new idea. It might be a higher-end implementation than we've seen before, but once again, the Panavision logo doesn't make it special - especially if Sony built it!

 

Phil

[Guest Jim Murdoch]

Also, if you told a potential buyer that your camera had a 12 megapixel

chip and that it can only output a 2 megapixel image---AND you didn't

give them a reason for the difference---what do you think the buyer is

most likely going to assume is the reason for the difference...

 

...but I don't thinkSony is the best business partner for Panavision. We all know how misleading and erratic Sony can be (if anybody needs examples, let

me know). Here, I'll give this example right now...oh wait, I can't. :unsure:

Okay, does anyone else here have enough memory capacity to

remember all of the different media formats Sony has created or

helped to create in the last 30 years? :D

 

-Ted Johanson

<{POST_SNAPBACK}>

 

First of all, Panavision don't sell cameras, they only build them for their own rental operation, so scalability isn't really an issue for their customers.

 

Second, last year Panavision suddenly sold their interest in eFilm apparently to raise money to buy back Sony's interest in Panavision, dissolving their "business partnership". I think the Genesis is still coasting along on its own momentum, since it's been in the works for several years. According to my contacts at a certain large video rental house, Sony representatives are no longer welcome at Panavison! It's all very weird, really.

 

We all know how

misleading and erratic Sony can be (if anybody needs examples, let

me know).

<{POST_SNAPBACK}>

The alleged superiority of the Betamax format is one of the most refractory urban myths in our recent technological history. Any competent video engineer can/could easily demonstrate its shortcomings compared to VHS but it never makes the slightest difference.

 

And before any of you leap blindly into the fray, f*cking Betamax is NOT Betacam! They're two completely different things.

[Guest Jim Murdoch]

Weren't those old cameras quite prone to flaring artifacts from bright speculars? Some weird aliasing artifacts?  And easily damaged by shooting toward the sun?

 

<{POST_SNAPBACK}>

Just about all camera tubes could be damaged by pointing them at the sun. The most famous (and public) example of this was on the Apollo 12 moon landing, when they fried the vidicon tube in their camera a few minutes after emerging from their spacecraft!

>>Some weird aliasing artifacts?

 

That's the $64K question. I would have thought the Genesis would surely have used an alternating RGB pixel structure viz

 

RGBRGBRGBRGBRGB

BGRBGRBGRBGRBRG

RGBRGBRGBRGBRGB

 

and then use line averaging to cancel out colour artifacts, which is what they've always used in single-chip RGB colour cameras before, both Sony's and those made by other manufacturers! (I think they all used Sony CCD chips anyway).

 

But when I asked John Galt at Panavison about this, he obviously didn't understand what I was talking about. After a couple more similar technical "howlers" I realized it was a waste of time trying to get any real technical information out of him.

 

I'm beginning to wonder if he's really gotten his facts straight, since the only diagram of the Genesis sensor I've seen is the one he drew himself, obviously using PC Paint!