View Full Version : HD Progressive CCD Concerns
Marty Baggen October 2nd, 2005, 06:49 AM Many have been following the reports from the field about the "split screen" effect in the new JVC.
The underlying cause has been explained to be a result of the high density, physically small chips which are progressively scanning... thus requiring a dual circuitry of sorts to be able to process the images, handle the heat, etc (I am paraphrasing with reckless abandon).
Is the HVX is going to be designed any differently? Or is this going to be a concern for all HD, 1/3" progressive scanning CCDs, and could we see the same "split screen" issues with the HVX?
Just reading some of the tech explanations of what these chips have to endure, it makes me wonder what their life expectancy will be.
Thomas Smet October 2nd, 2005, 08:48 AM Did the JVC HD1 or HD10 have the split screen issue? I do not remember it having this issue and it was progressive scan 1280x720.
Marty Baggen October 2nd, 2005, 09:05 AM Wasn't the HD10 a single CMOS chip? I'm not certain, I don't follow that closely.
But the way I understand the situation, is that there are certain design characteristics of the 3 x 1/3" CCD configuration in HD that demands dual circuitry. That means two separate white balance and black levels that are impacted by even the slightest variance between those two separate circuits... such as voltage, which is speculated to be the likely culprit for the "split screen" problem on the HD100, reportedly seen at all gain levels, but more pronounced at +9db and +18db.
JVC has been kicked around over this issue, and although I am no fan of JVC, it is appropriate to ask if Panasonic will likely have a similar problem, or if they have come up with an alternative design in their pickup circuitry.
Barry Green October 2nd, 2005, 06:10 PM Did the JVC HD1 or HD10 have the split screen issue? I do not remember it having this issue and it was progressive scan 1280x720.
I am pretty sure it didn't; I don't remember anyone ever saying anything about there being a split-screen on the HD1.
However, we should keep in mind that the HD1 operated at only half the speed of the HD100. The HD100 can scan its CCDs at 60fps, vs. the max of 30fps for the HD1. Perhaps that explains why they had to move to dual processors?
Barry Green October 2nd, 2005, 06:14 PM Wasn't the HD10 a single CMOS chip? I'm not certain, I don't follow that closely.
Single CCD. I suspect that it's the same CCD that's in the HD100.
JVC has been kicked around over this issue, and although I am no fan of JVC, it is appropriate to ask if Panasonic will likely have a similar problem, or if they have come up with an alternative design in their pickup circuitry.
That's the question. I talked with a rep of JVC and he expressed a similar concern -- how will Canon and Panasonic avoid it? He said that the "magic number" of pixels you can extract from a CCD is about 1,000 across. Which is why he suggested that Sony stopped at 960 wide -- to avoid crossing that magic number. At 1280 across, the JVC is thoroughly over that line, and he said that's why they needed to implement the dual processors.
Canon is claiming 1440 pixels across. The JVC rep expressed curiosity over how Canon will deal with 1440 pixels (even more than the JVC) without having dual processors. I haven't seen any hint of splitscreen in the Canon footage...
As for Panasonic, we don't know what the pixel count of the chip is, nor do we know how they're addressing reading the chip. The footage shown at RESFest certainly didn't display any detectable split-screen.
Thomas Smet October 2nd, 2005, 06:34 PM Ok how about the Varicam then? That uses 3 chips at 1280x720 and I never heard anybody complain about a split screen. That camera also clocks at up to 60 fps and while dvcprohd only has 960 pixels wide as a format the camera still processes the 1280.
If the Varicam can do it I am sure the HVX200 will also manage without a split screen.
Marty Baggen October 2nd, 2005, 07:13 PM I believe I read that the Varicam actually uses an interlaced system, then combines full frames within the interlaced stream. I could be completely off base on that, and I'm sure Barry could chime in with the correct info.
Keep in mind, we are also talking about the demands being place on the 1/3" CCD. I suspect most Varicam models are using 2/3" chips?
Barry, you mentioned the Canon.... but isn't that some sort of interlaced system. They refer to their "progressive" as "F" frames, not true "P". I wonder if that is a result of their method to bypass the issue we are discussing?
If Panasonic or anyone else, is using 1/3" chips in progressive mode, and avoiding any imaging defects such as the split screen.... I would think it would be beneficial for all of us to understand how it is done.
It is apparent to me that the 1/3" chips are having a lot more physical demands placed on them than any other pickups that I know of. I think that it is pertinent to ask if those operational conditions will have a predictable affect on failure rate, and usable lifespan. Granted, there are no moving parts, but heat can have a serious impact on delicate circuitry.
Barry Green October 2nd, 2005, 08:08 PM Ok how about the Varicam then? That uses 3 chips at 1280x720 and I never heard anybody complain about a split screen.
VariCam chips are 2/3" though. According to JVC the issue is one of heat, and it's accentuated by the smaller chips. Bigger chips can have more, bigger pixels (the CineAlta's 2/3" chips are 1920x1080). JVC was saying that the issue is because they're reading so many pixels out of such a tiny chip.
I don't know whether that's all 100% accurate or not, it's not something I have any way of verifying. But I do doubt that they would have gone with a split-processor system for the fun of it; I'm sure they had to have a reason.
Barry Green October 2nd, 2005, 08:10 PM I believe I read that the Varicam actually uses an interlaced system, then combines full frames within the interlaced stream.
No, VariCam is 100% progressive all the time. It only does progressive-scan 720p, there is no provision in the VariCam to do anything interlaced.
Barry, you mentioned the Canon.... but isn't that some sort of interlaced system. They refer to their "progressive" as "F" frames, not true "P". I wonder if that is a result of their method to bypass the issue we are discussing?
Well, there's something to think about there, sure. A 1440-wide interlaced chip would only be reading every other field, instead of the full chip. However, JVC specifically said that it was the # of pixels per line that was giving them the problem, so... I don't know.
Thomas Smet October 3rd, 2005, 07:56 AM Barry good point about the 2/3" chips. I hadn't thought heat was the issue. I am curious to know then how 1080i cameras can suck in 1440 pixels across if it is an issue of only being able to use 1000 pixels across. Granted 1440x540 is less then 1280x720.
Marty Baggen October 3rd, 2005, 09:15 AM Now that we have excluded 2/3" chips from this equation, and re-indentified the heat issue, I want to come back to the original purpose of this thread.
1 - is a requirement of a 3 x 1/3" HD progressive CCD block, dual circuitry?
2 - is the above normal heat generated by the CCD going to impact lifespan and/or performance?
Barry Green October 3rd, 2005, 11:24 AM JVC's position is that they went to dual processors because of heat, yes.
And as to how someone can extract 1440 pixels off a 1/3" chip -- they are quite curious to see how Canon is doing that, because as far as their research showed, it's not possible. And they have the same question for Panasonic. They believe it's not a coincidence that Sony stuck to 960 pixels, they think it's this same heat issue. Same with Canon and the XL2, how its CCD is maxxed out at 960 across.
Like I say, I don't know what the absolute truth here in any of this is.
I do think JVC had reasons to adopt their dual-processor approach. I also am extremely disappointed with the split-screen effect -- it seems my HD100 can't shoot a low-light scene without excessive split-screen plaguing it in every frame. Under bright light conditions there's no split, but shooting available-light shots, there's huge split (and yes, that's on zero gain, manual exposure, manual knee, etc). If this split-screen is the price we pay for having a native-res chip, then (and this is subjective) the price is probably too high. You gotta be able to trust your camera. Right now, I don't. I'm still working with it to see if I can explore the full extent of the issue, but just knowing that this can happen to your footage is worse than MPEG artifacts, IMHO.
Marty Baggen October 3rd, 2005, 11:57 AM Well said Barry. I guess the motivation behind my pursuit of this issue is to find out if we are talking about a universal limitation of the progressive HD strategy.
The real scary part of this is, what if 1, 2, or even 5 years from now.... we see a plethura of CCD failures or performance problems.
I guess we just have to wait and see. It is a bit of a riddle, but if Panasonic has a working model, the answers are out there. Do we even know for certain that the Panasonic chips are progressive?
|
|