Panasonic goes CMOS

[Anhar_Miah]

Won't work though. Might on a static object while you're panning, but what if something is going the other way? I mean, let's say you're panning to the right, and a car drives through the frame from the right to the left. Your in-camera compensation might straighten out the background, but it'd double-skew the car...

Damn it Barry your right!

after re-reading the paper again, it seems that it was not merely pixel shifting, what they did was "stack" 52 cameras images and then "collated" the 4d data, so that each "slice" occured at exactly the same time, however each camera was at a different point of the scanline,

I've done some basic calculations:

Let P = orginal point in 2d space

Let P' = point captured on cmos im 2d space

thus delta_Error = tan * theta-1 * Height(pixels)

this is for a point travling +ve relative to P,

if however we have a point Q' such that it is traving -ve to point P, then if we apply the corrective transform the resultant point Q'' becomes:

Q'' = (tan*omega-1*height(pixels)) + delta_Error


Note: note to sure about the theat-1

 

[Blaine]

I've done some basic calculations:

Let P = orginal point in 2d space

Let P' = point captured on cmos im 2d space

thus delta_Error = tan * theta-1 * Height(pixels)

this is for a point travling +ve relative to P,

if however we have a point Q' such that it is traving -ve to point P, then if we apply the corrective transform the resultant point Q'' becomes:

Q'' = (tan*omega-1*height(pixels)) + delta_Error


Note: note to sure about the theat-1

I think my head just exploded...:huh:

 

[BobDiaz]

One of my new cameras is CMOS and I have never seen this EVER.

That's like saying a certain car design is unstable. Then showing the design is indeed unstable when the car goes faster than 550mph. Totally irrelevant to almost anyone using that car design because outside of an extreme situation, it won't ever be travelling that fast.

So show me what you've personally shot with a current CMOS sensored camera where this was any kind of an issue.

All my cameras are CCD, except for the $99 "toy camera" that I play with. Because I know what the rolling shutter will do, I avoid cases where this will screw up the image. For example, I did shoot the last LA Marathon (for fun) while walking the 26 miles. My small "toy camera" was light, but there was no way I could shoot with with all the walking bounce, so I used a larger CCD model.

However, I do have examples of other people's videos:

The black and white shot comes from Philip Bloom's work, Dungeness. The blur that tilts is a train passing by the camera; notice the tilt. I spotted this as it was playing.

The wedding shot, below is when a flash fired as the rolling shutter was closing. You can see it as it plays. The more flashes in the video the worse it gets.

The blur to the right is where someone did a whip pan and caused the image to lean as a result. (OK, this one the person was testing the effect, but any whip pan can do this.)


The worst problem is a flash or many flashes, like at a wedding. It shows up in the video and there is no hiding it.

As long as something takes 3 seconds or more to move from one edge (left or right) to the other edge, the degree of tilt is minor, BUT anything that moves faster than that will start to show. A fast pan or something moving fast will show the tilt. You can avoid this up to a point by avoiding whip pans and tight shots on fast moving objects with vertical edges.


Bob Diaz

 

[craigbowman]

Great. Thanks for posting the photos Bob. I really wanted to see some examples. I tried a whip pan on the Red but didn't get that result and I don't have any strobes to test.

 

[Buck Forester]

Not surprised to see it in the low-end consumer devices. I'll be surprised if Panasonic would put CMOS in a broadcast camera though.

Do you mean in the very near future, or ever? I'd bet very heavily (as in I'm 99.9% sure of it) that eventually they will. CMOS is where things are heading, both in still photography and video. With each successive generation there will be great improvements, much more so than CCD, because CMOS has some cost/effeciency advantages that will make them see the most investment/research money. CMOS has just begun.

 

[Anhar_Miah]

firstly not all CMOS control circuits are the same, RED does indeed suffer from rolling issues, HOWEVER it is greatly reduced because the scan rate is very fast (much fater than Consumer CMOS).

for examples of RED rolling shutter see:

http://ieba.wordpress.com/2007/11/01/red-rolling-shutter-redux/

Technically speaking THE only way to stop rolling shutter is to use a global shutter, BUT that increases costs, because one of the reasons why CMOS is cheaper is because they do not use a sample and hold switch on EVERY pixel (as with global shutters) this reduces complexity and cost.


Rolling shutter correction (but as Barry pointed out, would not work if an object moved in the opposite direction of the pan):

http://mpac.ee.ntu.edu.tw/Exhibition/rolling-shutter.php

 

[Erik Olson]

If RED renders obsolescence obsolete, shouldn't they have started with a global shutter regardless of cost?

e

 

[ecking]

This plus an hmc150 seems like it could be a really cool two camera operation for people with needs below what p2 has to offer.

Panasonic canada told me about this a few weeks ago but no info was supposed to be released this early, I guess someone leaked it so they just put it all out there. Interesting.

For anyone who doesn't know the max bitrate is 17Mbs. It looks like anything past 20Mbs is gonna be considered AVCCAM (at least from panasonic). 17Mbs looks great though so I don't see that being too much of a big deal if the consumer side peaks at roughly 20.

 

[Adam J McKay]

I agree with Erik. You think that Red would have used a Global Shutter, because if and when they come out with a camera that can do that their current cameras will be..."obsolete", however I doubt a RED ONE will be obsolete anytime soon. They may even be crazy enough to offer the same deal as the One to Epic upgrade pacakge if and when they use a global shutter. Yummy speculation.

 

[Kholi]

This plus an hmc150 seems like it could be a really cool two camera operation for people with needs below what p2 has to offer.

Panasonic canada told me about this a few weeks ago but no info was supposed to be released this early, I guess someone leaked it so they just put it all out there. Interesting.

For anyone who doesn't know the max bitrate is 17Mbs. It looks like anything past 20Mbs is gonna be considered AVCCAM (at least from panasonic). 17Mbs looks great though so I don't see that being too much of a big deal if the consumer side peaks at roughly 20.

It's due out in Japan in July, that's next month and there's press release from Panasonic Reps in several articles about these cameras.

 

[BobDiaz]

Great. Thanks for posting the photos Bob. I really wanted to see some examples. I tried a whip pan on the Red but didn't get that result and I don't have any strobes to test.

You are welcome.

CMOS is not a dirty word for video. As long as a user understands the limits of the technology and can live with it or knows how to avoid it, CMOS can produce great results.

I think we all know what happens with CCD when you point a CCD camera at a very bright light source. That's a limit with CCD.

For those who were around in the days of tube cameras, you'll remember the "comet trails" we got from bright lights. I also remember the tube guys being upset with the vertical line we got from CCDs when we pointed a CCD at a bright light.


Bob Diaz

 

[Kholi]

Has anyone mentioned the difference or impact a 3 CMOS system in this size camera will have versus one 1/4" sensor?

I'm assuming these three aren't 1/4" either. Hmm. Actually, doing another search (might've missed it in the article) it's 3 x 1/6" Inch sensors versus one 1/4".

 

[BobDiaz]

This plus an hmc150 seems like it could be a really cool two camera operation for people with needs below what p2 has to offer.

Panasonic canada told me about this a few weeks ago but no info was supposed to be released this early, I guess someone leaked it so they just put it all out there. Interesting.

For anyone who doesn't know the max bitrate is 17Mbs. It looks like anything past 20Mbs is gonna be considered AVCCAM (at least from panasonic). 17Mbs looks great though so I don't see that being too much of a big deal if the consumer side peaks at roughly 20.

On the HMC-150, the Maximum data rate is 21 Mbps average, with 24 Mbps maximum.

I understand it come outs in October this year in the USA.

Here's what I wrote in my web page about the HMC-150:

> Supports 1080 at 60i, 50i, 30p, 25p, & 24p
> Supports 720 at 60p, 50p, 30p, 25p, & 24p

> Uses SD & SDHC Memory Cards and Records using AVCHD H.264

> Records around 3 hours at maximum data rate to a 32GB SDHD Card

> Doesn’t support over cranking and under cranking like the HVX-200

> Does NOT replace the HVX-200

> The higher model is the HPX-170, which uses P2 memory cards

The higher model looks almost the same as the HMC-150 and uses the

same CCDs and Lens.

> Maximum data rate, 21 Mbps average, with 24 Mbps maximum

> Uses 72mm filters

> 13x Zoom (3.9mm --> 51mm??) 35mm equivalent of 28mm --> 364mm??

?? = Not exactly sure, this is an estimate.

> 3 speed switch for the zoom

> Waveform Monitor & Vector scope

> Offers Pre-Record Function (never miss a good shot)

> Panasonic says it will be under $4,500 List.

Now is that $4,499.99 or $3,999.99?



As far as I’m able to see and test, the HMC-150 looks like it’s going to be an exciting camera for its price range. The images look clean, the colors are rich, and even in poor lighting with gain, the image noise is soft. The feel of the camera is wonderful, it is well balanced and not very heavy. Because it uses the same CCD and lens as the higher model, the HPX-170, the images should be about the same between the 2 models. The higher model will offer additional features and P2 recording. Back when Panasonic introduced the HVX-200, there were naysayers who felt it would be a flop, but Panasonic proved them wrong. Can Panasonic do it again at a lower price point? I sure hope so, but we’ll have to wait until October until we can know for sure. Stay tuned...

Bob Diaz

 

[Barry_Green]

Do you mean in the very near future, or ever? I'd bet very heavily (as in I'm 99.9% sure of it) that eventually they will.

Oh, I'm sure they'll do it eventually, if they come up with a satisfactory solution. But Panasonic resisted HDV even when everyone else went there, and I wouldn't be surprised if they resisted rolling-shutter CMOS even when everyone else goes there.

Of course, if people clamor for it, I'm sure they'll make it too. But I'd be surprised if they just offered a rolling-shutter CMOS product from the broadcast lineup. If they do, I'm sure it'd be in the small-chip lineup; I'd be shocked if they came out with a 2/3" CMOS product.

CMOS is where things are heading, both in still photography and video.

In stills, yes.

In video? Where?

(ooh, I know I just made some people pop their stack, but stick with me...)

In the consumer lineup, sure, CMOS is all the rage, but in professional cameras? Red and SI are going all-CMOS, but those aren't broadcast manufacturers. Among the broadcast manufacturers:

Every JVC Professional camcorder is CCD-only.
Every Canon professional camcorder is CCD-only.
Every Panasonic professional camcorder is CCD-only.
Every Hitachi professional camcorder is CCD-only.
Every Sony professional camcorder is CCD-only, excepting the sub-$10,000 lineup.
13 of the 15 Ikegami professional camcorders are CCD; 2 are CMOS.
Every Thomson professional camcorder is CCD (except the Infinity).

Dalsa and Sony, two of the biggest CMOS manufacturers in the world, use only CCDs in their digital cinema products. Sure, Sony will put a CMOS in a $6500 EX1, but what do they put in their $350,000 F35? CCD.

With each successive generation there will be great improvements

Let us hope.

much more so than CCD

Says who?

CMOS has just begun.

CMOS sensors began 40 years ago, as did CCD. CCD was just patently superior in all ways. It's only in the last few years that CMOS has even become competitive.

CMOS has advantages in small-chip HD. In larger chip cameras, the CMOS advantage is pretty much cost, but for performance CCD is still tops. In small-chip cameras, the problem is packing enough pixels onto the chip and managing the heat of reading a lot of pixels off. CMOS's low heat allows them to pack in more pixels and read off at faster frame rates. The tradeoff is that you get rolling shutter complications.

So in small-chip cameras, yes, we'll probably see more and more CMOS models, because that's where CCD struggles. You can get a 1/6" CCD but you have to keep the pixel count down and you can't get the fast frame rates. But in bigger chips, CCD is still king.

 

[Barry_Green]

If RED renders obsolescence obsolete, shouldn't they have started with a global shutter regardless of cost?

e

Red's going to allow existing Red One owners to upgrade to the latest/greatest Mysterium X sensor. And, I imagine, if they someday come out with a global shutter, they'll allow existing owners to upgrade to that. I believe that's the gist of their "rendering obsolescence obsolete" claim, is that you'll never be "stuck" with last-generation technology. There will always be a path forward.

 

[Barry_Green]

CMOS is not a dirty word for video. As long as a user understands the limits of the technology and can live with it or knows how to avoid it, CMOS can produce great results.

Exactly.

I think we all know what happens with CCD when you point a CCD camera at a very bright light source. That's a limit with CCD.

Yes, but even so, it's not a universal limit. The bigger the pixels are, the less likely this is to happen. So you'll probably see more vertical streaking on an HPX3000 than you would on an HPX500. It's really tough to get an HPX500 to streak.

For those who were around in the days of tube cameras, you'll remember the "comet trails" we got from bright lights. I also remember the tube guys being upset with the vertical line we got from CCDs when we pointed a CCD at a bright light.

Yep, the same discussion had to take place back then. New technology brings new challenges, and if you wanted to shoot a super-bright light field, a CCD was the wrong choice back then.

 

[Buck Forester]

CCD was just patently superior in all ways. It's only in the last few years that CMOS has even become competitive.

We both know "superior in every way" does not mean "the way of the future". We've seen this in Beta vs VHS. We're seeing it in CRT vs LCD. What wins out in the future is not which is inherently best from the get-go, it's where the $$ moolah is being spent. Beta was better than VHS but a lot of money went into VHS and it eventually became a more widespread and superior product. CRTs are better than LCDs but LCDs are now catching up (and some might argue just as good or better) but will surpas anything CRTs in the future because that's the push... they're cheaper to make and consume less energy. Sounds like CMOS vs CCD to me. If I had to place money on the table, I'd put it all on CMOS because that's where the biggest gains will be, it has the most room for advancement from a cost/effectiveness ratio, and that's what companies like. Panny and Sony are still in this game for the money. CMOS will bring the most ka-ching and eventually surpass CCD. Do I know this for a fact? No. But I'm darn sure. Plus I saw photos of the new CMOS factory in AREA 51. You didn't hear it from me. Gotta run. *poof*

 

[Barry_Green]

Market factors certainly seem to triumph quality factors, that's true. You may be correct and 10 years from now it may be an all-CMOS world, regardless of whether one has better performance than another, if the price tag dictates it.

 

[Tzedekh]

Market factors certainly seem to triumph quality factors, that's true. You may be correct and 10 years from now it may be an all-CMOS world, regardless of whether one has better performance than another, if the price tag dictates it.

I think that until rolling shutter has been effectively eliminated (i.e., no more wobble or problems with flash, strobe, or HMI lighting) and there are no deficits when compared with CCD technology, there will be CCD-based cameras. They will be stratospherically priced (maybe even more so than now), but when you're shooting a $100 million feature, you're probably not going to try to economize by choosing a cheaper camera.

 

[phd]

ARRI FLEX D-21 have CMOS, but the price :thumbsup: