ACTUAL Pixel resolution of "35mm" CCD sensors.

[Phil Rhodes]

Hi,

 

Some well made points. Quite a lot of stuff there is already addressed, or trivial - mounting a monitor, using Digiprimes or the Pro35 adaptor, and there are plenty of lenses in the 7x range that don't breathe very badly - and make more sense for operator focus l and it's worth remembering that a lot of the things that make them long are additional options which just don't exist for film cameras - accurate monitoring, sound recording. F'you want it to be shorter, take 'em off; you're only going to be back where you started.

 

The whole video/data thing seems like nitpicking. It's like saying a pinhole camera is something other than a camera because it has no glass lens. Put data on tape as YUV and people complain it isn't untouched RGB, put it on hard disks and people complain they're chained to a fridge.... YUV is a sensible way to get around the problem in a way that matches human vision well.

 

Prism blocks are very limiting regarding flange depth, or rather they force you into complex retrofocal lens designs which compromise performance. If it's possible to create single-CCD image sensors that supersample adequately to produce true RGB data at the desired resolution, I don't see a problem with that. The reason to do that is to improve lens performance, though - NOT just to allow people to use whatever lens series they happen to have fallen in love with. Very high resolution CCDs will be physically large for a while yet so the use of 35mm film lenses will be less than ideal in any case. The higher sensitivity of a CCD might allow these larger sensors to be used at deeper stops, alleviating focus pulling issues that might otherwise occur.

 

At the end of the day it's ironic that possibly the best widely-available "digital cinematography" camera, which uses a 3-chip block and video lenses and yet has the option to use a tiny, top-mounted RAM recorder.... is essentially the front end of a Thompson HDTV studio camera. Telling, I think.

 

Phil

[John Sprung]

Thanks for your input, but the answer to my question: "Does anybody really know?" appears to be : "No!" ;)

<{POST_SNAPBACK}>

The specific answers so far are:

 

D-20: 3018 x 2200 pixels under a Bayer mask

 

Dalsa: 4046 x 2048 pixels under a Bayer mask (Approx. 3000 x 2048 covered by standard 35mm film lenses)

 

Genesis: 12.4 Megapixels, under a non-Bayer RGB mask

 

I would submit that the above falls somewhere between knowing everything and knowing nothing.

 

 

 

-- J.S.

[John Sprung]

I don't see if it matters so much if you have three CCD's for each color versus one CCD with three times the pixels to record the three colors on one surface

<{POST_SNAPBACK}>

The big difference is that with the three chip separation camera, you can get the dead regions between the pixels to be quite small, but with a single chip each color has large spaces between where it gets sampled. This problem is called undersampling.

 

Consider a red Christmas light far in the background, just about big enough to equal a pixel on the chip. If it happens by chance to land exactly on a red pixel in a Bayer mask, it shows up big and bright. On the other hand, if it lands exactly on a blue pixel, it pretty much gets lost. Move the camera a little and the sucker blinks at you. Soften the image with diffusion or microlenses enough to even out the reds, and you lose the resolution advantage of the more numerous greens in the Bayer pattern.

 

The pixels on a large chip are large compared with the minimum feature size that can be made in silicon. These things together lead me to speculate that they may be using something more clever than just little squares as pixels. The Viper used subpixels that were rectangles, four times wider than high. Even more complex interlocking shapes should be possible.

 

 

 

-- J.S.

[Phil Rhodes]

Hi,

 

I seem to remember that there's a very high pixel count (22Mpel or something) medium-format back that uses hexagonal pixels, or at least pixels in a hexagonal arrangement, and interpolates down to the standard grid of squares. Someone must be pretty sure there's an advantage to even go to the lengths of manufacturing the chip.

 

As for size and shape of the individual pixel, as far as I know every pixel is a single photodiode. While it's presumably possible to control the size and shape of the N-region of the device, you'd expect a greater tendency to overflow if the charge was being built up any significant distance from the anode - which might reduce effective full-well capacity and negate the purpose of having a big pixel.

 

However, I know almost nothing about semiconductor fabrication, so this may all be drivel.

 

Phil

[Guest Jim Murdoch]

The specific answers so far are:

 

D-20:  3018 x 2200 pixels under a Bayer mask

 

Dalsa:  4046 x 2048 pixels under a Bayer mask (Approx. 3000 x 2048 covered by standard 35mm film lenses)

 

Genesis:  12.4 Megapixels, under a non-Bayer RGB mask

 

I would submit that the above falls somewhere between knowing everything and knowing nothing. 

-- J.S.

<{POST_SNAPBACK}>

Well OK, look; if we talk about "4K" film scanning, we're talking about four thousand red, four thousand green and four thousand blue, equal-status pixels across the useable width of the film. Does anybody have any argument with this?

 

Since the Dalsa has only about 3,000 pixels in total across the standard 35mm film area, it can't possibly be a "4K" sensor as they claim. It couldn't even be a "4K" monochrome sensor as used in the Arriscan, but if it uses a Bayer mask, there's no way it could even approach "4K" resolution in the normal sense of the term. To avoid color aliasing, you have to severely limit the resolution of the optical image that's focussed on the chip, and also use some pretty savage signal processing to average out the responses of the individual colored pixels.

 

See, anybody can take a picture that's 800 pixels across and use Photoshop's "Image size" function to convert it to one that's 4,000 pixels across, but effectively all you'd be doing is "re-photographing" a low-res image with a high-res camera. Garbage in - Garbage out. The fact that a video camera outputs a certain number of pixels per line in no way guarantees that those pixels are gainfully employed, but I get the feeling this is something they'd rather their custromers didn't think too much about!

 

Hence the question: "What is the actual pixel resolution?" If I knew what the actual pixel layout of the Genesis was, I could calculate its approximate "raw" resolution. I'm not really interested in what the signal processing spits out the other end.

[Guest Jim Murdoch]

Jim,

Sony have stated that the Genesis is rgb rgb rgb  striped array of 12.1 million pixels.

 

Yeah, I've heard that, but you/they mean like:

 

RGBRGBRGBRGBRGB

RGBRGBRGBRGBRGB

RGBRGBRGBRGBRGB

RGBRGBRGBRGBRGB etc,

 

that is utterly impossible. The RGB triplets have to be "shuffled" and heavily averaged; otherwise you would get massive color artifacts on striped shirts, sharp edges and so on. The result is that single-chip sensors (whatever type of filtering is used) can only produce luminance and chrominance signals in something less than 4:2:2. It is simply not possible for that type of sensor to produce the sort of true RGB signals you need for chroma key and similar. A Bayer mask gets you closer but at the expense of oversampling the Green.

 

The next question is just how far will Panavision go in promoting it at the risk of drawing clients away from using their film cameras.

Arri have already stated that the D20 is not a replacement to film.

Maybe Panavision will follow suite, with the Genesis being a competitor to f900 rather than 35mm film?

For the last 5 years or so, Panavision's principal shareholder Ronald Perelman has been desperately trying to flog Panavision off to some sucker. If you ever read any of the reports he is required to submit to the S.E.C you'll see exactly what this all about! Anyone would think he was talking about IBM!

Summed up as  "having your cake and eating it too".

Mike Brennan

<{POST_SNAPBACK}>

AKA "Shooting yourself in the foot". Once you've put film-making in the hands of CCD and Hard disk manufacturers, who needs Panavison or Arri? If this crap really worked, people would have started making "Digital Cinematography" cameras that take 35mm stills camera lenses long ago!

[Phil Rhodes]

Hi,

 

> ....something they'd rather their custromers didn't think too much about!

 

I don't mean to keep harping on about this, but their reaction to my contention of exactly the same point suggested that this is indeed the case. Really, really hostile.

 

Phil

[David Mullen ASC]

Jim, you seem to be imply the this "crap" (the Genesis) DOESN'T work, when it does, and if movie camera makers are going to add digital video cameras to their line, how can they NOT be involved with CCD and hard drive makers, et al.? Or are you suggesting that they should never get involved with any aspect of digital cinematography and leave it to companies like Sony & Panasonic?

 

Personally, if I had to choose, I'd rather see a future where movie cameras were still being made by Panavision and Arri even if that means they have to get into making digital versions rather than have Sony & Panasonic dominate all movie production equipment.

 

This is really not an issue of technology -- there are only a certain number of companies making CCD's, or CMOS, etc. -- it's a question of attitude. Companies like Sony and Panasonic are off in Japan and they make every decision based on mass-production of electronic gear on an assembly-line basis. Camera companies like Panavision and Arri make everything more or less hand-assembled, a few cameras per year, because the market simply is not that large. So you can actually TALK to these companies and they will modify equipment according to user demands. It's not like some plastic mass-produced camcorder being handed to the film industry by Sony that we just have to make due with.

 

With companies like Dalsa and Kinetta forming, we may see a day of small movie camera companies forming like we used to have in the 1940's, 50's etc. because they can use off-the-shelf components to create digital movie cameras.

 

As for why we haven't seen single-chip movie cameras before, just like the digital still market has, the problem has always been response time -- those digital still cameras can't snap a 12MP image 24 times a second.

 

I wouldn't write off the Genesis, the Arri-20, and the Dalsa as "crap" just because they use a single sensor...

[Michael Most]

Once you've put film-making in the hands of CCD and Hard disk manufacturers, who needs Panavison or Arri?

 

The same people who need Panavision and Arri in the film world. CCD's and hard disks are imaging/recording components, just like film except in two separate pieces out of technical necessity. Both Panavision and Arri make devices to use those components, not the components themselves. Their relationship to the CCD and hard disk manufacturers is essentially the same as their relationship to Kodak and Fuji.

[John Pytlak RIP]

Don't forget, Kodak is a state-of-the-art OEM supplier of digital image sensors too:

 

http://www.kodak.com/global/en/digital/ccd/sensorsMain.jhtml

 

Kodak's Bryce Bayer was the inventor of the "Bayer" color filter for digital sensors.

Edited December 19, 2004 by John_P_Pytlak

[Guest Jim Murdoch]

The same people who need Panavision and Arri in the film world. CCD's and hard disks are imaging/recording components, just like film except in two separate pieces out of technical necessity. Both Panavision and Arri make devices to use those components, not the components themselves. Their relationship to the CCD and hard disk manufacturers is essentially the same as their relationship to Kodak and Fuji.

<{POST_SNAPBACK}>

Well Arri certainly don't make lenses, their entire expertise in film cameras pretty much consists of making quiet, steady and relaible film transport and exposure mechanisms. Panavision likewise, although of course they do make good lenses too. Not too many other firms have the expertise to build quality motion picture cameras, however:

 

There's a large number of small companies that make large-area CCD and CMOS sensors, mostly for specialist/ custom applications. There's also any number of firms that make hard disk and optical drives for computer applications.

Apart from a lens mount, and digtal processing chips what else do you need? Do you even need PL-mount lenses?

 

In the case with digital still cameras, although the big names in mechanical cameras are all prominently represented, digital cameras have precious little in common with their film-based ancestors, so any "legendary" expertise they may have had in the past really counts for nothing. In fact most current digital still cameras are simply "OEM'd" by companies that specialize in digital equipment.

 

Why would it be any different with digital movie cameras?

[Guest Jim Murdoch]

Jim, you seem to be imply the this "crap" (the Genesis) DOESN'T work, when it does,

 

Sigh. All I want for Christmas is the actual number of horizontal RGB pixels on the Genesis sensor. Not what somebody imagines there are.

 

As I said, by "work" I mean "produce an overall image quality good enough to replace 35mm movie cameras in mainstream applications". I'm sure they produce some sort of picture, but so does the CineAlta.

 

The problem is, I cannot see how a chip with that number of pixels can produce the image quality claimed. Perhaps all that book-learnin' has clouded my idealism, but there it is.

I wouldn't write off the Genesis, the Arri-20, and the Dalsa as "crap" just because they use a single sensor...

 

Since I haven't seen any actual evidence of any significant improvement over technologies that have been around for years, I would!

 

Or are you suggesting that they should never get involved with any aspect of digital cinematography and leave it to companies like Sony & Panasonic?

Call me an old cynic (hang on, I've got the number here somewhere:-) but on past experience, that's eaxctly what I would suggest! But hey, that's just me.

[David Mullen ASC]

Since the Genesis is designed to allow an HD resolution image, it is not equivalent to 35mm resolution. You're asking for some sort of proof that it is 35mm quality; well, the "proof" is that it isn't, not if it's only a 1920 x 1080 pixel recording. I don't think anyone at Panavision is claiming that the Genesis achieves 35mm resolution because they know better.

 

But calling something "crap" based not on how the actual image produced looks but only based on a preconceived notion, well, if that's the way you like to do things, fine. I'm a little more of the "shoot it and see how it looks" school; specs and theory take you so far then you have to use your eyes. It's not like anything but 35mm is "crap". Other formats can produce nice images for the right applications.

 

I really don't get all of this hostility towards Panavision and Arri for their experiments in digital cameras. If you want to shoot on Sony cameras, no one is stopping you. They'll be making them for a long time to come; you can avoid Panavision and Arri if you'd like.

 

The majority of HD cameras are 2/3" 3-CCD designs with prism-blocks, a system you seem to feel is superior over any single-sensor approach. Well, you're in great shape because there are thousands of those cameras out there while there are only a few prototypes of the Genesis or Arri-D20 -- yet for some reason you resent the fact that these prototypes exist even if some DP's want to shoot with them. Why? You've got what you want... but that's not enough -- you don't want anyone else to be allowed to use these single-chip designs because you don't approve of them? What's up with that? Why limit the range of designs for these cameras to just one approach? If ths single-sensor idea is the disaster you seem to think it is, then the results it produces will obviously be inferior and the whole bad idea will be dropped once people see the results.

 

As for why Panavision and Arri, who are in the business of making motion picture cameras, should get into making digital versions of motion picture cameras, well, it's because they are in the business of making motion picture cameras. They are experimenting with a new approach to video cameras, maybe ultimately fruitless, but I don't see the harm in trying alternate approaches to things.

 

The 1950's produced all sorts of alternative approaches to standard 4-perf 35mm spherical photography, like anamorphic lenses, 3-camera Cinerama, 8-perf horizontal 35mm, 5-perf 65mm, not to mention all the experiments in sound formats. Not all of them worked or stuck around but that doesn't mean the whole decade of technical experimentation was a bad idea. I'd rather see more wacky approaches to making digital cameras than for everyone to just admit that the 2/3" 3-CCD prism-block approach was the best and only worthwhile way to build a video camera and lock us into that approach for the next two decades...

[Guest Jim Murdoch]

You're asking for some sort of proof that it is 35mm quality; well, the "proof" is that it isn't, not if it's only a 1920 x 1080 pixel recording.

 

Well as it happens, I don't see how the sensor could achieve even that resolution. Sure, the camera outputs data at 1920 x 1080, but how different are those two-million-odd pixels from each other?!

 

I don't think anyone at Panavision is claiming that the Genesis achieves 35mm resolution because they know better.

 

I've heard the claim "indistinguishable from 35mm film" made many times by various PV people.

 

I really don't get all of this hostility towards Panavision and Arri for their experiments in digital cameras.

 

I don't have any problem with Arri, since they don't seem to be all that serious about the D-20. My problem is that Panavison and Dalsa are making claims of big technological breakthroughs, but either go dead silent or on the offensive when you ask for the hard data to back up their claims.

If this single-sensor idea is the disaster you seem to think it is, then the results it produces will obviously be inferior and the whole bad idea will be dropped once people see the results.

But hopefully, not before some poor fool is conned into sinking more money into one of these companies!

[Guest Jim Murdoch]

Don't forget, Kodak is a state-of-the-art OEM supplier of digital image sensors too:

 

http://www.kodak.com/global/en/digital/ccd/sensorsMain.jhtml

 

Kodak's Bryce Bayer was the inventor of the "Bayer" color filter for digital sensors.

<{POST_SNAPBACK}>

Which leads one to wonder why Kodak haven't jumped into the 35mm single-sensor fray. With their money and resources, I'm sure they could come up with a really nice design, that's if they thought it would actually work!

 

I doubt it's because they think they would be shooting themselves in the foot, since most of their income comes from supplying film for release prints; actual cinematography film is not that a big a part of the market.

[Michael Most]

I've heard the claim "indistinguishable from 35mm film" made many times by various PV people.

 

Have you ever actually seen any of the results from the Genesis? Have you ever even looked at the intercut test that Allen Daviau shot? The statements about "indistinguishable from 35mm" have come largely from NON-Panavision people. People like Daviau, Steven Poster, and yes, David Mullen and myself. None of those people work for Panavision or have anything to gain by praising their product. We just know what we saw - and apparently you haven't.

 

I really don't know what your problem is with these new approaches, either. The end result is all that matters. The numbers mean less than nothing, and theory is, well, theory. I prefer the reality to the theory. And the fact is that these new devices will be used, regardless of your opinion or your hostility towards them. Why not take a look at the actual results before you cement that opinion?

[Michael Most]

Which leads one to wonder why Kodak haven't jumped into the 35mm single-sensor fray. With their money and resources, I'm sure they could come up with a really nice design, that's if they thought it would actually work!

 

What makes you think they're not? Just because they haven't announced an actual product yet? You are aware that Kodak has made line array CCD sensors for scanning and telecine use for a number of years, aren't you (the CCD's in the Spirit telecine are Kodak parts)? Do you have actual knowledge that they aren't doing any R&D in this area, or are you voicing an opinion based on a lack of such knowledge (as you apparently are doing with your opinions on the "big chip" camera designs)?

[Mike Brennan]

Yeah, I've heard that, but you/they mean like:

 

RGBRGBRGBRGBRGB etc,

 

that is utterly impossible. The RGB triplets have to be "shuffled" and heavily averaged; otherwise you would get massive color artifacts on striped shirts, sharp edges and so on. The result is that single-chip sensors (whatever type of filtering is used) can only produce luminance and chrominance signals in something less than 4:2:2. It is simply not possible for that type of sensor to produce the sort of true RGB signals you need for chroma key and similar.

 

snip

If this crap really worked, people would have started making "Digital Cinematography" cameras that take 35mm stills camera lenses long ago!

<{POST_SNAPBACK}>

 

Well obviously they are doing something different as it seems utterly possible, to the naked eye at least. The camera does not seem to have any serious problems with artifacts on vertical lines. This evaluation is based on examinations of the film-out test and hands on experience with the camera and a 24 inch HD monitor. I'm sure there are some artifacts in there somewhere but I couldn't see them and having shot HDCAM for 3 years and suffered artifacts from iQ Avid f500 Xpri HDLink ec t ect I hope that would see something as obvious as you suggest that would be a issue.

 

In respect of green screen, lens fringing is a big issue with wide angle lenses both 2/3 or 35mm.

Genesis has a means of correcting lens fringing to aid chroma keying.

 

Is your agrument that a RGB RGB RGB array has a colour detail only equivalent to 4:2:2 three chips? I sugges that it is valid argument if (each of the 3) chips were the same size as the single chip.

 

Sure 3 x 35mm size chips will be better than 1 chip with RGB RGB RGB array, but the Genesis chip has almost twice the number pixels as a 3 chip camera, (although we don't know how many are active)

 

You raise a valid point, why haven't there been digital cameras using 35mm lenses before?

The simple answer is that there was no money in developing a single chip of PAL or NTSC broadcast quality. 7 years ago who would have bought such a camera? Broadcasters? who would then have to buy new lenses?

Those with 35mm lenses eager to move away from 35mm film to PAL quality video?

 

Was the technology even available a decade ago to make, in large numbers a 35mm format ccd chip?

If so why didn't the still photography world do it? Sony and Ikegami's first HD cameras had prisms. Of course they would have prefered a single chip and 35mm lenses.

 

Your premise that because it hasn't been done before is an indicator that it shouldn't work now ignores the progress in A/D chip minaturisation ect and economies of scale that exist now that did not exist a decade ago.

 

This "crap" that you refer actually works. Nikon is making 90000 single chip "crap" cameras (just one model!) a month.

 

A study by Info trends/CAP released this week predicts digital camera revenue will reach US $24 billion this year.

Sony and Sumsung did a deal this week to share the 24000 patents they have between them.

 

Evidence of progress!

 

But seeing is believing so book yourself into a Genesis demo.

 

Are Sony playing tricks with numbers and doing a dodgey shuffle, probably, but no one on this forum is saying that Genesis is as pure as the driven snow.

 

 

Mike Brennan

[Guest Jim Murdoch]

This "crap" that you refer actually works. Nikon is making 90000 single chip "crap" cameras (just one model!) a month.

 

Still photography CCD cameras can get away with an enormous amount of picture "crispening" and other "enhancement" because you're only looking at a single image. The problem with any sort of artificial image enhancement with motion pictures is that it has to be absolutely consistent from frame to frame, otherwise you get the sort of detail correction "twitch" that's plagued color TV cameras since they were invented. On a fifty-foot screen the problem would be massively magified.

 

 

But seeing is believing so book yourself into a Genesis demo.

If I hadn't heard almost exactly the same sorts of claims made about the CineAlta five years ago, we probably wouldn't be having this conversation. (Or the same sort of claims made about NHK's abysmal "Hi-Vision" HDTV cameras in the late 1980s.

 

I've been a veteran of far too many "video indistinguishable from film" demonstrations that were more "film indistinguishable from from video". Maybe you've been priveleged to see something different, but frankly I'm sick of it. No matter where you go, the conversation always seems to be dominated by a hard core group of low-end "professionals" who seem desperately in need of the services of either an optician or a psychiatrist. The less-vocal amongst us tend to restrict our "conversations" to sideways glances and mutterings of "WTF?"

 

And woe betide anybody who asks embarrassing questions of the people running the demo, ads people here have pointed out.

 

"Seeing is believing" That's the trouble, we [don't] see, so we don't believe. But"this one is going to be different"; yeah, it always is.

 

It's a bit like watching the out takes of one of those "World Idol" tye programs; you know: "Does that guy actually think he can actually sing - at all??!!

 

If Panavision wanted to give one of us curmudgeons one of those cameras to play with, we'd soon show you how "indistinguishable" it is, which is precisely what happened when the CineAltas got out into the real world.

 

And I'm only talking about HD-scanned-off-35mm-film vs HD-sourced-from-the-Geneis.

 

I'm sorry; I'm a scientist, not a Theologian: I want hard data, not anecdotes and homilies.

[Guest Jim Murdoch]

What makes you think they're not? ... Do you have actual knowledge that they aren't doing any R&D in this area, or are you voicing an opinion based on a lack of such knowledge....

 

I'd be astounded if they hadn't, but I tend to suspect that (like Arri) they quickly realized that this was just a technological dead end. Unlike PV, maybe they still employ people there who know somehting about the subject!

 

As for my "knowledge of 'big chip' camera design" my problem is that PV and others are claiming to have made enormous advances, without bothering to tell us what they've actually achieved. I mean, if somebody suddenly managed to produce a workable 100 megapixel sensor, they could surely at least tell us that they'd done that without needing to give away trade secrets.

Edited December 19, 2004 by Jim Murdoch

[Mike Brennan]

I'd be astounded if they hadn't, but I tend to suspect that (like Arri) they quickly realized that this was just a technological dead end. Unlike PV, maybe they still employ people there who know somehting about the subject!

 

As for my "knowledge of 'big chip' camera design" my problem is that PV and others are claiming to have made enormous advances, without bothering to tell us what they've actually achieved. I mean, if somebody suddenly managed to produce a workable 100 megapixel sensor, they could surely at least tell us that they'd done that without needing to give away trade secrets.

<{POST_SNAPBACK}>

 

We'll all have to wait for the hard data. It has taken years for the dirt on HDCAM to be delivered. Only recently a head of a camera department of a national broadcast admitted that he thought HDCAM recorded 1920x1080.

 

I don't disagree with your opinions of the film vs HD and HD vs film tests. The Genesis test had scene with ND gelled windows and a 12kHMI as a key, yet the view out the window was hot and bleached with minimum detail... from my perspective it gave the illusion that the camera was holding more in the highlights than it was. More to the point why expose it that way if you have resources (gel and HMI) to make a better balance... The film example was also exposed in the same way with the same key and gels so the comparison itself was not unfair.

 

For most I would say the important aspects of the Genesis are its film-like depth of field and low compression 10 bit recording, these improved (over f900) aspects are quite apparent to the naked eye with a taste for film.

 

I can't see huge gains in dynamic range, just modest improvements overall that narrow the gap with film.

 

I'll put the kettle on while we wait for the tech details from Panavision :)

 

In UK Panavision are not making a song and dance about it yet, whats happening in the states?

 

 

Mike Brennan

Edited December 19, 2004 by mike

[David Mullen ASC]

The Genesis is basically a 4:4:4 HD camera like the Viper or F950 but with the added distinction of having 35mm depth of field characteristics plus having a built-in recorder. Some improvement in dynamic range apparently. Color and resolution should probably be similar to the Viper and F950. The Viper has one advantage over the Genesis in that you can obtain 2.35 without cropping the 1920 x 1080 pixel recording.

 

Arri hasn't lost interest in their D20 project; they are just trying to get it right before they show it off again, probably at NAB 2005. It has been somewhat redesigned from just being a 435 with a single chip stuck in the back.

[Guest Jim Murdoch]

The Genesis is basically a 4:4:4 HD camera like the Viper or F950 but with the added distinction of having 35mm depth of field characteristics...

 

It couldn't possibly be 4:4:4, not with a 12 megapixel sensor. (Unless you're talking about 525 line downconversion).

 

The only way the Genesis cold produce the sort of results you've described is to "cheat" by using detail correction. And once they start doing that, depth of field goes out the window, because the processing electronics sharpens up everything, whether it was sharp in the original scene or not.

[Guest Jim Murdoch]

I don't disagree with your opinions of the film vs HD  and  HD vs film tests. The Genesis test had scene with ND gelled windows and a 12kHMI as a key, yet the view out the window was hot and bleached with minimum detail... Mike Brennan

<{POST_SNAPBACK}>

For me the real test would be to put a 35mm film camera and a Genesis (with all automatic exposure controls disabled) side by side, start with correct exposure of a daylight scene, and progressively open up the iris a stop at a time. With movie film, you can easily go 12 stops over exposed and still recover a workable image with a good telecine.

 

In my experience, video cameras begin to lose it at just four stops over and long before you got to 12 stops, there'd just be a big white rectangle. And there's no way back from the "white rectangle of death" as there is from the "murky grey rectangle" of overexposed film!

 

Video cameras have all sorts of automatic exposure control mechanisms, but they can only control the overall sensitivity of the sensor. If the manufacturers could up with a chip that had individual gain controls for each pixel, then that would behave more like film. But such a device is far beyond our present technological capabilities.

 

I frankly don't believe that the Genesis could have some individual pixels that are illuminated 4,096 times as much as other pixels and be able to capture both without distortion. As far as I know, there aren't even Standard Definition cameras that good!

[Phil Rhodes]

Hi,

 

> With movie film, you can easily go 12 stops over exposed and still recover a workable image with a

> good telecine.

 

Is anyone else here willing to agree with that?

 

Anyway, if you're going to cite the ability of a medium to help you get away with a mistake, you could just as well argue that video is more tolerant of underexposure.

 

In any case, while it's quite easy to overexpose film and know nothing about it, you're never likely to make that mistake on a camera whose output you are monitoring in real time, so the entire argument is meaningless.

 

Phil