I have an EX1 for a weekend. Anyone want me to run tests?

[Stevet]

A quick review ...http://www.sonyhdvinfo.com/showthread.php?t=11758

 

[Barry_Green]

You've got to wonder if all these companies are on to something with this CMOS thingie.

This CMOS thingie that they're on to is fast frame rates and cheap chips with low power consumption. Makes a lot of sense for cheap camcorders/cell phone cameras. Makes less sense for something like the Infinity. Makes sense for Red in that they wanted a cheap price tag and fast frame rates. Makes sense for the EX1 in that they want massive quantities at a cheap price.

The only holdouts have been Panasonic and JVC, although JVC doesn't make very many cameras.

In bottom-of-the-line cameras, maybe. In top-line, guys like Dalsa are using CCD, and as you pointed out, a Sony engineer said that the strategy is CCD for the broadcast stuff, CMOS for the cheaper stuff. That makes a lot of sense.

By the way, it is possible to have a CMOS sensor with a global shutter. Apparently this is common in industrial cameras for machine vision. I'm sure there must be trade-offs though...probably one of them being higher cost because the pixel structure is more complicated.

Yep, exactly. And the CMOS revolution isn't about more cost, it's about less cost, which is why people haven't done it.

 

[Barry_Green]

Are you planning on a hands on review of the EX1?

I'll evaluate it, but I haven't decided yet as to whether I'd do a full-fledged review. I've learned that people only want to read what they want to hear in a review. If you agree with their preconceived notions they applaud you, if you disagree with their preconceived notions then you're a shill or a liar or some other vile thing. And of course there's people on both sides of every fence, so by and large it's kind of a no-win situation. Especially if you include pictures -- many times they won't even read the words, they'll look at the pictures and then start calling you names, even if you methodically document your settings and procedures...

Regardless, I'll get time on one, and I'll tell people exactly how it performs. If that meets with their wishes or disagrees with 'em, I can't help that. The thing is what it is.

 

[tnle2]

This CMOS thingie that they're on to is fast frame rates and cheap chips with low power consumption. Makes a lot of sense for cheap camcorders/cell phone cameras. Makes less sense for something like the Infinity. Makes sense for Red in that they wanted a cheap price tag and fast frame rates. Makes sense for the EX1 in that they want massive quantities at a cheap price.

In bottom-of-the-line cameras, maybe. In top-line, guys like Dalsa are using CCD, and as you pointed out, a Sony engineer said that the strategy is CCD for the broadcast stuff, CMOS for the cheaper stuff. That makes a lot of sense.

Yep, exactly. And the CMOS revolution isn't about more cost, it's about less cost, which is why people haven't done it.

Actually, my point was that many of the top-of-the-line cameras are using CMOS and I doubt it's about cheap chips. It's about all the other advantages like lower power consumption, lower noise, higher sensitivity, multi-formats, and high frame rates. Examples include those Ikegami EFP cameras, ARRI D-20 and Vision Research Phantom 65. None of these cameras are cheap and each of these companies cite inherent CMOS advantages as to why they chose it.

It seems like the only drawback is a rolling shutter. But if you can achieve a snapshot shutter on a CMOS then I don't see why you wouldn't use it. It could be a technical reason or a business reason. For example, if global shutter is patented by a few companies (like Micron who has their own system) then the other companies who make sensors (like Sony or Canon) might not want to license it.

 

[ESTEBEVERDE]

I'll evaluate it, but I haven't decided yet as to whether I'd do a full-fledged review. I've learned that people only want to read what they want to hear in a review. If you agree with their preconceived notions they applaud you, if you disagree with their preconceived notions then you're a shill or a liar or some other vile thing. And of course there's people on both sides of every fence, so by and large it's kind of a no-win situation. Especially if you include pictures -- many times they won't even read the words, they'll look at the pictures and then start calling you names, even if you methodically document your settings and procedures...

Regardless, I'll get time on one, and I'll tell people exactly how it performs. If that meets with their wishes or disagrees with 'em, I can't help that. The thing is what it is.

That's why I like your reviews.

You just tell it how you see it and let the cards lie where they may! :beer:

 

[KeeganFlynn]

It seems like the only drawback is a rolling shutter. But if you can achieve a snapshot shutter on a CMOS then I don't see why you wouldn't use it. It could be a technical reason or a business reason. For example, if global shutter is patented by a few companies (like Micron who has their own system) then the other companies who make sensors (like Sony or Canon) might not want to license it.

RED uses a global shutter, no?

 

[ullanta]

By the way, it is possible to have a CMOS sensor with a global shutter. Apparently this is common in industrial cameras for machine vision. I'm sure there must be trade-offs though...probably one of them being higher cost because the pixel structure is more complicated.

Sensitivity, too - the added transistors for global shuttering significantly impact the percentage of the area of each pixel that can gather light.

 

[TheMusician]

Sensitivity, too - the added transistors for global shuttering significantly impact the percentage of the area of each pixel that can gather light.

I guess that to compensate, they'll just have to give us bigger chips... say, 2/3rds of an inch? Hey, a guy can dream can't he?

 

[marco0782]

Barry, claiming that the Arri D-20 uses CMOS to save money is ridiculous.

 

[LuckyStudio 13]

When the mini RED comes out in a short few years. Everyone here in this forum will be pimpin' CMOS.

Yes, I can tell the future and you mark my word

 

[Barry_Green]

Actually, my point was that many of the top-of-the-line cameras are using CMOS and I doubt it's about cheap chips.

What top-of-the-line cameras use CMOS? Infinity isn't anywhere near top of the line, it's the bottom of the Thomson line. The EX1 isn't anywhere near the top of the Sony line, it's the bottom of the CineAlta line (and Sony's already said they will use CCD for the top of the market, CMOS for the bottom).

Red, yes -- the top of their line is CMOS. But it's also the only product in their lineup. And what's the big thing about Red? Its price tag. :thumbsup: If the Red was $350,000 (like the Sony F35) would anyone be all that excited about it?

It's about all the other advantages like lower power consumption, lower noise, higher sensitivity, multi-formats, and high frame rates.

Lower noise? Nope, CMOS is higher noise. Higher sensitivity? Sorry, no, CMOS is a lot slower (goes hand-in-hand with it being higher noise; you can't rate CMOS up because the noise becomes an issue). Plus the additional space on the chip for the transistors reduces the light-sensitive area. CMOS is slower than CCD, head-to-head.

Higher frame rates and lower power draw? Yep. And no smear. And the fact that they can be stamped out of silicon using high-yield (and therefore low cost) production techniques. Those are the advantages for CMOS, but there are lots of drawbacks. Dalsa's published a series of papers that compare CMOS against CCD, they make for interesting reading.

In short, there is no clear-cut winner. CMOS has been around for 30 years now, and yet everyone used CCD up until the last couple of years. CMOS was not a competitive technology. It's only now getting to the point where it can be viewed as competitive.

Examples include those Ikegami EFP cameras, ARRI D-20 and Vision Research Phantom 65. None of these cameras are cheap and each of these companies cite inherent CMOS advantages as to why they chose it.

The D20, I'll grant you. EFP cameras? No, not buying it. The Phantom? They had to go CMOS because of the frame rates. And the Infinity probably went CMOS (they were initially CCD) because they had *massive* power issues and heat issues. Going CMOS let them cut the power draw, which would directly correlate to less heat generation too.

It seems like the only drawback is a rolling shutter.

It's a problem, but not the only problem. Again, look at the Dalsa papers for perhaps the best comparison information.

But if you can achieve a snapshot shutter on a CMOS then I don't see why you wouldn't use it. It could be a technical reason or a business reason. For example, if global shutter is patented by a few companies (like Micron who has their own system) then the other companies who make sensors (like Sony or Canon) might not want to license it.

I think almost all of my resistance to CMOS would disappear with a global shutter or mechanical shutter.

 

[Barry_Green]

RED uses a global shutter, no?

No. Rolling.

 

[Stevet]

Red, yes -- the top of their line is CMOS. But it's also the only product in their lineup. And what's the big thing about Red? Its price tag. :thumbsup: If the Red was $350,000 (like the Sony F35) would anyone be all that excited about it?

Still, $17,000 USD for just a RED camera body is a lot of $$$$ for many. I'm not sold on CMOS is only offered as low end solutions, that's for sure.

 

[Barry_Green]

Still, $17,000 USD for just a RED camera body is a lot of $$$$ for many. I'm not sold on CMOS is only offered as low end solutions, that's for sure.

Trust me, I know it's a lot -- I'm due to shell that out this month. (and it's more like $50,000, all in, with some cheap glass and an appropriate tripod.)

It's not that CMOS is "only for low end", I'm not saying the Red One is "low end", but come on -- the Red One is priced less than a half-inch Sony F330. The buzz about the Red is because of the price. I'm also not saying that the EX1 is "low end". But when faced with a choice, every manufacturer that offers both CCD and CMOS has, as far as I know, always chosen CCD for the high end, and only offered CMOS in the lower-cost end of their product line.

 

[Stevet]

But when faced with a choice, every manufacturer that offers both CCD and CMOS has, as far as I know, always chosen CCD for the high end, and only offered CMOS in the lower-cost end of their product line.

True, It's been that way in the lower end market.
I also believe there have been advances in this technology, we'll have to wait and see where this takes us.

I also admit that the EX1, like every othere camera looks great on paper, but it's not until you get your hands on a camera that you realize the tradeoffs.

With all these cameras, (especially sub $10K), we have to learn to use the camera to make the best we can out of it.

Barry, I'm looking forward to hearing your thoughts on the RED.
A coworker is due to get one in two weeks here in the Phoenix area.

I'm looking forward to checking it out. He did mention that the overall cost and the additional workflow requirement really pump up your total cost. I see it has for yourself.

 

[Barry_Green]

My Red is supposed to come out this month, so yeah, we'll see. Should be fabulous, but I'm going into it with eyes wide open. I've talked to enough people who've used it that I don't have the starry-eyed syndrome. To say I'm worried about the rolling shutter would be a true statement, that's for sure. That's about all I'm worried about on it though, and part of why I've become so obsessed with rolling shutters. I gave my HV20 away because I couldn't stand the rolling shutter so I wouldn't use it. It's a fine product for many uses but I just didn't trust it. I HATE the idea that your camera can screw you over when you don't expect it, so I obsess over how/when/where.

For my Red uses, I'm not shooting red carpet events, nor is it going to be strapped to a mountain bike (and besides, the thing's so heavy that vibration won't likely be much of a problem for it). But for handycams, I think it's more of an issue. In any case, Red says they "fixed" the rolling shutter problems, so I'm very curious to see if they have, and how.

As far as the Red goes: people just have to understand that it's not anything like any video camera on the market; in fact it's *not* a video camera. It's a digital film camera that uses an entirely different workflow. No it's not a "true" 4K, it's more like 3K. I'm okay with that, as it's still way better than anything else, surely way better than anything anywhere near the price. Heck, when I put in the reservation I was happy with the concept of 1080p @ 35mm DOF, so 3K worth of "true" res is a bonus. No it's not 1,000 ISO, it's more like 160. That's not ideal, but it's not bad and especially considering you can get T1.3 lenses, it's workable. It's not noise-free, there's some issues with noise. I understand they're working on that and improving it.

It's like I say with all these things -- put away manufacturer hype and go in with your eyes wide open. There are no miracles here, there are only tradeoffs. There are tradeoffs in every product. Know what's important to you, and how each product performs, and you can then be empowered to choose the right product for you.

 

[Stevet]

Thanks Barry.
One thing for sure, the RED is an investment. The company will stand by their camera and get it right for their users. I wish the big guys would follow suit with this mentality.

 

[tnle2]

What top-of-the-line cameras use CMOS?

Okay, let's not say top-of-the-line since we'll be splitting hairs. Let's just say very high performance cameras. So there's the ARRI D20, all Phantom cameras, RED One and the Ikegami HDK-79EC. That covers both the high-end digital cinema and high-end broadcast markets.

Red, yes -- the top of their line is CMOS. But it's also the only product in their lineup. And what's the big thing about Red? Its price tag.

I think Jim would disagree. He said he set out to design the best camera he could--something that even if he only sold one to himself, he would want it. Do you think a billionaire with all the design and financial resources at his disposal would compromise his camera with a CMOS sensor (if it was inferior) just to make it cheap? Makes no sense.

Lower noise? Nope, CMOS is higher noise. Higher sensitivity? Sorry, no, CMOS is a lot slower

You must be thinking of old stereotypes. Times change and technology improves. This white paper states:

• Designers of CMOS-based sensors reduce noise while eliminating the classic noise vs. resolution trade because the pixel-based amplifier’s bandwidth better matches the imager sampling frequency. The CMOS output buffer’s noise is usually negligible. In practice, therefore, CMOS can circumvent the “3 dB per octave” increase in noise that is experienced with CCD sensors and the associated degradation in camera S/N.
• The sensitivity of today’s best CMOS sensors matches the sensitivity of the comparable CCD devices since the amount of light collected at each pixel is now roughly equal. Specifically, when deep submicron CMOS processes are used to design and fabricate the sensor, the photon collection efficiency of a CMOS pixel supporting progressive image formation is significantly higher than for a progressive CCD at the 5 micron pixel pitch that produces 2/3 inch HDTV sensors.

Also, subjectively it seems that people are reporting that the EX1 is rated at ASA 800 and it's more sensitive and less noisy than the CCD based F350. High-end DSLRs also use CMOS. The Nikon D3 is being universally praised for low noise at very high ISO speeds.

And the fact that they can be stamped out of silicon using high-yield (and therefore low cost) production techniques. Those are the advantages for CMOS, but there are lots of drawbacks. Dalsa's published a series of papers that compare CMOS against CCD, they make for interesting reading.

I read those papers and DALSA contradicts some of your own conclusions, especially about CMOS being cheap chips. They state CMOS chips are often just as expensive as CCDs because of the greater process adaption, but they are available for high-performance applications.

The D20, I'll grant you. EFP cameras? No, not buying it.

Ikegami started using CMOS in some cameras back in 2004. The latest Ikegami HDK-79EC is being used on mobile production trucks:

http://ikegami.com/IKEGAMI_and_F&F_US_OPEN_FINAL.html
http://ikegami.com/IKEGAMI_BEXEL_FINAL.html
http://ikegami.com/IKEGAMI_YES_PRODS_FINAL.html

And the Infinity probably went CMOS (they were initially CCD) because they had *massive* power issues and heat issues. Going CMOS let them cut the power draw, which would directly correlate to less heat generation too.

You make it sound like Thomson chose CMOS as a band-aid. That's an unfair characterization. According to this press release: "The 2/3-inch 1920 x 1080 active pixel Xensium HD sensor will first be used in production models of the Grass Valley Infinity™ Digital Media Camcorder, and the Xensium line of imagers will become a standard in future generation Grass Valley camera products across a broad range of applications."

It's a problem, but not the only problem. Again, look at the Dalsa papers for perhaps the best comparison information.

Every high-end DALSA imaging chip used for cinema or photography is CCD. They only make CMOS for machine vision. DALSA apparently likes CCD. Other people like CMOS. Nothing wrong with that too.

 

[Barry_Green]

I think Jim would disagree. He said he set out to design the best camera he could--something that even if he only sold one to himself, he would want it. Do you think a billionaire with all the design and financial resources at his disposal would compromise his camera with a CMOS sensor (if it was inferior) just to make it cheap? Makes no sense.

Well, with all due respect, I don't think Jim would disagree with me at all. There are obviously compromises involved in the Red design, specifically to meet a price point. If he set out to build the greatest thing in the universe, price as no object, would he have settled on where he did? Maybe, but I doubt it. Part of the goal of the Red design, from day one, was to make it affordable so that guys like us (DVXUser members) could buy it. That was always part of the design philosophy. Price was always a foremost consideration in the Red design.

You must be thinking of old stereotypes. Times change and technology improves. This white paper states:

Could be. Most of what I've been reading is the Dalsa papers, and they are a couple of years old by now. Even Sony, though, when they introduced the V1U, admitted that a Z1/FX1 would perform better under lower illumination levels. And every CMOS camera other than the EX1 has proven to be substantially slower than the CCD equivalent. And the EX1, I believe, gets its sensitivity/noise performance through aggressive noise reduction processing, not through innate sensor sensitivity.

Also, subjectively it seems that people are reporting that the EX1 is rated at ASA 800 and it's more sensitive and less noisy than the CCD based F350.

Many people are. Adam Wilt said it was 320. I asked him how, since everyone else says 800, and he said "I have no answer for you, all I know is that every which-way I tested it, it was 320." He will hopefully be providing a more thorough report in the future that will address this 1.5-stop discrepancy.

I read those papers and DALSA contradicts some of your own conclusions, especially about CMOS being cheap chips. They state CMOS chips are often just as expensive as CCDs because of the greater process adaption, but they are available for high-performance applications.

Yes, but when you go with the 3T design, that's when they become cheap. And that's what everyone is using so far, are they not? CMOS can be designed to be a lot more complex, and it can be designed to use a global shutter, but both of those mean more expense.

In any case, you're always going to have a system that's designed to hit a certain price point. That's the fundamental nature of product design. You calculate your projected price point and your desired performance, and then you compromise and tradeoff until you get as close as you can to both.

You make it sound like Thomson chose CMOS as a band-aid. That's an unfair characterization.

I don't think it's an unfair characterization at all. I also don't know if it's true, but it sure seems to be, and if I had to bet on it I'd bet on it. Thomson was screwed, they were stuck with a design that wasn't going to come out. It was way too hot and had way way too much power draw, it was extraordinarily heavy, and people started saying that the joke was "the reason they named it 'infinity' is because that's when it would come out." By going with CMOS they were probably able to greatly cut the power consumption, which should also cut the heat output, which would help them solve two of their three problems. Weight is the third, don't know what they'll do about that.

and the Xensium line of imagers will become a standard in future generation Grass Valley camera products across a broad range of applications."

How is this inconsistent with anything? Of course they're going to use this thing anywhere and everywhere they can now, they developed it, they have it, why not use it? Unless the market has a backlash against it, of course.

Every high-end DALSA imaging chip used for cinema or photography is CCD. They only make CMOS for machine vision. DALSA apparently likes CCD. Other people like CMOS. Nothing wrong with that too.

And Dalsa's where I've been getting some of my info, which may explain why I am quite an advocate of CCD. But in all actuality, it was my HV20 that turned me against rolling-shutter CMOS, and it is the substantial price tag of the Red that I'm getting that has me figuring out anything and everything I can about the rolling shutter.

If you like a rolling shutter, hey, more power to you, I'm sure there will be many more rolling-shutter CMOS choices in the future too. But just because more people are using it, that does not inherently mean it doesn't have issues! The issues will remain, the issues are present. When they actually *do* fix them such that there are no issues, I'll gladly embrace it. Until then, I think people should at least be aware of what these systems do and how they perform, so they can make an informed choice.

I mean, look -- I could turn the argument around and say HDV is the lowest-quality recording format ever introduced to record a high-def signal. And you could say "HDV can't be so bad, because major companies like Sony and Canon and JVC use it, and Sony sells HDV product in its BROADCAST lineup" and etc... Simple fact of the matter is, HDV is the lowest-quality way to record an HD signal that's ever been released, and if you disagree with that then we really can't discuss much further. Just because Sony uses it, does not mean that it was chosen because it's high quality! It was chosen because it fit on the existing tape structure, and the companies wanted to use the existing hardware that they had. THIS DOES NOT MEAN THAT HDV IS USELESS. But it is unquestionably the lowest-quality HD recording system ever introduced. And it was not developed for ultimate quality (even if a camera that costs 50% more than an EX1, the XLH1, uses it!) It was developed, I believe, because they had a 25-megabit target to hit, they had a price point they had to hit, and they fiddled and waggled and finagled until they found a way to cram an HD signal into the available space that was given 'em.

And it works surprisingly well, most of the time. Sometimes it falls apart, but usually it works well. Sound familiar? These rolling-shutter CMOS chips are inexpensive and they work surprisingly well, most of the time. But sometimes they fall apart. It is the tradeoff you have to face.

 

[StMad]

And the EX1, I believe, gets its sensitivity/noise performance through aggressive noise reduction processing, not through innate sensor sensitivity.

What are you referring to here? On 0db (or -3) or when adding gain?

It's difficult to believe Adam's 320iso result is indicative of the production model. If Sony did NR tweaking (after Adam's test) to increase sensitivity, with no apparent side effect...that's a pretty impressive achievement. It'd be interesting to learn more.