Split screen technical explanation

[Douglas Spotted Eagle]

Originally Posted by Marty Baggen

I think the bottomline should include consideration of the underlying CAUSE of the effect.

I'm not a huge fan of JVC, but won't it be ironic if Panasonic has the same issue with their HVX?

..... see my earlier post.

FWIW, I can *almost* guarantee Panasonic won't. I keep waiting for someone to offer a specific, technical explanation for why this is happening. It's there, I believe, I've had two different competitive manufacturers tell me essentially the same story as to the why...however, I don't believe I'm mathmatically astute nor technical enough to explain it with any degree of confidence or authority.

[Chris Hurd]

DSE has just saved me from an immense amount of further typing. What he just said pretty much sums up the way I feel about the situation.

I never know what motivations lies behind someone who has only a handful of posts on this site. Maybe they're a longtime lurker who knows the ropes around here, or maybe they think this is "just another message board." So it's my intention to make this point very clear:

A person cannot camp out in a forum discussing some hardware or software that they have chosen not to buy; they cannot hang out here constantly nay-saying about something they're not using. That has not happened yet; I'm not accusing anybody of actually doing that here. I'm just letting you know now, ahead of time: that game is not played here.

As has been stated, the SSE issue is very real and we're all very well aware of it and I'm looking forward to some kind of response from JVC. Meanwhile, if you've said "I'm not buying this camera," then that's great -- it's the *one* time you're allowed to do that on this board. Leave the remaining bandwidth to the users and to those folks that are genuinely interested in furthering some *productive* conversation. Hope this helps,

[Matt Davis]

Originally Posted by Douglas Spotted Eagle

1. JVC is very aware of the issue, and seems to be dealing with it.

In the UK, JVC are doing another Milk Run with TouchVision. Today in Farnham - Thursday's in London with the Wide Angle on show for the first time.

Bottom line: no SSE, but then no night shots. Some room lighting, but still no SSE (one shot looked like it had in still frame, but it was part of the scene!)

An informed sauce says that there are almost weekly firmware revs from JVC, but there's no point in releasing a firmware update if it's not quite right - c.f. 'Beta testing'. I guess when they've got it licked, they'll go GM. OTOH, there was talks of adjustments over and above the firmware update, requiring a trip to the shop - maybe this is the Dead Pixel Shuffle, maybe a real need to open the unit up and do something (though that sounds very 'Analog' to me).

[Marty Baggen]

Chris....

Were your comments made with me in mind?

If so, I can tell you I am sincere in my curiousity, and was hoping that my questions were put in such a way that the answers would not only provide some concrete design facts, but also make potential buyers of any HD camcorder more aware of how these things are designed.

[Chris Hurd]

Marty, my comments are not directed towards anyone here in particular; they're meant for those people who have said (or are about to say) "I'm not buying this camera." If you are choosing not to use this equipment and feel like you need to say so, that's fine... one time and one time only. Step aside and let the actual users have this board, that's who it's for.

I am specifically discouraging "crusaders." A crusader is somebody who feels that since they're not buying a particular thing, then neither should anybody else. It is not a crusader's responsibility to make potential buyers aware of what they perceive to be deal-killing flaws. Rather it's the potential buyer's own responsibility to evaluate for themselves what the bonafide owners of that gear are saying.

Anything else consitutes noise, which is worse than useless and gets in the way of intelligent and productive discourse between the person doing some serious research and the actual users of the gear. If you *own* this camera and you have something to add about the SSE issue, then I encourage your input wholeheartedly. If you don't own this camera and you have no direct experience with the SSE issue, then what business do you have talking about it. That doesn't help anyone. I hope that's clear. We've had enough meta-discussion about this so let's please try to get back on topic.

[Steve Mullen]

Originally Posted by Robert Castiglione

It is possible you have one of the more recent cameras with new software which has greatly reduced the problem.

To test:

Find a flat surface with only one colour in a room. Close the blinds. Turn on the light bulb as your only source of lighting. Fill the screen with the flat surface. Now pan across the flat surface. Try on no gain and then with gain.If you have the split screen it will be immediately evident to you. If you cant see it then great!

Rob

Doesn't tell you anything! You must get the exposure to F2, or higher.

[Marty Baggen]

Originally Posted by Chris Hurd

Step aside and let the actual users have this board, that's who it's for.

If you don't own this camera and you have no direct experience with the SSE issue, then what business do you have talking about it. That doesn't help anyone. I hope that's clear. We've had enough meta-discussion about this so let's please try to get back on topic.

Is this topic is closed to those who don't own the camera?

Is there another thread to ask about the CCD circuitry design, and the affect of heat on the chips?

I am confused about the rules here.

[Steve Mullen]

Originally Posted by Marty Baggen

Steve....

the symptoms of the issue may be closed, but I am concerend about the underlying cause.

SSE is a result of a slight mis-matching of dual circuitry. To a layperson such as myself, seems to suggest that ALL HD, progressive, 3 x 1/3" CCD pickup systems will be required (if dual circuits aren't needed, why implement them?). It has been suggested that the density of these tiny chips and the heat they generate is one of the factors in this design (at least the choice made by JVC).?

I also suspect it's the fact that 6 A/Ds are used. The first several dozen pixels in each CCD's row get no light. These pixels are sampled and averaged to a Black Level and stored in the DSP. One for each color. Normally, this value is subtracted from every pixel.

But, in the HD100, this one value must serve for both halves of each CCD. Each half is being digitized by a different A/D. Since there are bound to be differences between these 2 A/Ds -- the single Black Level has a probability of being wrong for the second A/D.

Now, when a full range image is captured (say 0-100IRE) -- a tiny percent error doesn't make much difference. But, if the image is poorly lit and uses only half (0-5OIRE) or a quarter (0-25IRE) of the A/D's range -- then the tiny error will show-up as a split.

That is why:

1) We need sufficient light.

2) We need an AVERAGE exposure of F4 to F5.6 in order to cover the possibily of a statistically "non-normal" distribution of light. For example, a very bright light supplies very high value samples to the AVERAGE, yet the rest of the pixels are in low-light. If we setttle for an F2 reading, the average will be F2 -- but the majority of the scene will be way darker. By adding light until we get to F4, we have a greater chance that the rest of the scene will be bright enough.

I would have assumed that JVC would have used all black rows at the top to actually determine the A/D differences and correct the Black Value for the second A/D. That might have been the firmware fix. However, it's possible that such a single correction simply isn't adequate for the dynamics of CCDs, A/Ds, and a huge range of lighting conditions.

THE BOTTOM-LINE -- the weaker the signal, they greater the error will show-up. At least, that's my GUESS as to what's happening!

By the way -- I suspect some one will write software to SAVE critical scenes where this occurs. Simply adjust the slider until its gone.

[Barry Green]

A few things...

but a line down the middle of screen ?

It's not a "line", it's that the left half looks different than the right half; one side will look brighter (and perhaps redder) than the other. The "line" refers to the notion that there is a clear demarcation between the halves of the screen.

JVC is very aware of the issue, and seems to be dealing with it

JVC is most definitely aware of it, and they are concerned about it. I've spoken with them about it.

the reason I see this issue as "closed" is because -- as I've posted -- SSE can be prevented. If it can be prevented, then it is no longer an issue. To keep posting about it -- it simply becomes a dead horse

This is something I cannot agree with at any level. To me, trying to tell people "don't worry about it" is nearly as bad as the crusaders who have no interest other than to discredit the product. Potential buyers need to know what's happening and when (or if) they will encounter the issue. The issue is not by any means closed, nor can it be prevented in a huge percentage of circumstances, so therefore it is not a dead horse.

I assembled my camera, plugged it into my XBR960, and turned it on. First thing I saw was two completely different looks to the halves of the picture on the screen. Not subtle, but massively noticeable. And this is with a firmware version 1.17 camera, which is apparently the latest.

Can it be prevented? Apparently yes, by keeping enough light in the scene. And for those who intend to use the product specifically under controlled lighting circumstances, they may be able to do as Steve recommends and it may become a non-issue for them. For others, who shoot in available light (I'm thinking news, sports, events, wedding receptions, etc) it could be a dealbreaker.

There are a whole lot of types of shooters out there, a whole lot more than just "indie filmmakers".

I flatly reject the idea of someone declaring this a "non-issue" -- that arrogantly dismisses a whole huge sector of the market's shooting requirements. I don't think that in the history of video cameras there's ever been an issue quite like this. Experienced professionals are unprepared for this -- they haven't seen something like this before. The issue needs to be explained, not glossed over or swept under a rug.

Let's come up with a better definition for it: let's determine the light rating that's necessary to avoid it, and then be able to definitively tell people: "if you're shooting at under XYZ footcandles, then you'll see an unusual artifact in the picture, where one half will be noticeably brighter than the other."

I will conduct some testing today to try to determine what FC the situation occurs under, and how many FC are necessary before the split is no longer visible.

[Jiri Bakala]

Originally Posted by Barry Green

This is something I cannot agree with at any level. To me, trying to tell people "don't worry about it" is nearly as bad as the crusaders who have no interest other than to discredit the product.

Thank you! Well put.

[Barry Green]

EDIT: Steve wrote in while I was typing, so I should update it to say that that's more like what I'm talkin' about. Let's determine what's happening, acknowledge the limitations, share info on how to defeat it, and advise people as to how they can determine whether or not it will be an issue for them. For some it will be a non-issue, for others it will be a dealbreaker. It's not for us to tell them whether or not it's a dealbreaker, but it is for us to supply them with the accurate and correct information that they need so they can make that decision for themselves.

[Steve Connor]

How would the reaction to this issue have been if it was a Sony camera?

I really hope they sort it out, I like the look of this camera and I was all set to get one, but as I shoot mostly doccos in available light, I'll be waiting to see what develops.This board has the best information I can find about it, so I'm glad it's still being discussed as it has a bearing on how I spend my hard earned cash!

[Chris Hurd]

Originally Posted by Marty Baggen

Is this topic is closed to those who don't own the camera?

This topic isn't closed at all. It is very much open. I am simply making a very basic appeal for the sake of the integrity of what usable, actual information we have here, to *please* withold comments about things one has no direct experience with. If one does not have the camera in hand and has not seen the issue, then what usable input could there possibly be?

Is there another thread to ask about the CCD circuitry design, and the affect of heat on the chips?

Please, I started this site years ago specifically to get away from that kind of talk... (grin)(that was a tongue in cheek comment from having been to too many trade shows).

I'll grant that CCD circuitry design may be an interesting topic for some (and it is not an uncommon topic around here), but what can you *do* with that information? What does it accomplish? How about another thread discussing how to actually use this thing. How about discussing technique and usability, setting up scene files and sharing them, the practical applications of the camera, and even more importantly, what you're creating with it and how.

I guess there are folks to whom the guts appeal to, but my original intention for DV Info Net was for it to be a *usability* forum. Who is using it, what are you doing with it and how. To me those are conversations far more stimulating, more interesting and ultimately of far greater relevance than dissecting innards.

And when there is an important issue regarding innards:

Originally Posted by Barry Green

For some it will be a non-issue, for others it will be a dealbreaker. It's not for us to tell them whether or not it's a dealbreaker, but it is for us to supply them with the accurate and correct information that they need so they can make that decision for themselves.

Bingo! And that's from an HD100 owner. That's what I was looking for. Thank you, Barry!

[Steve Mullen]

Originally Posted by Barry Green

For others, who shoot in available light (I'm thinking news, sports, events, wedding receptions, etc) it could be a dealbreaker.

There are a whole lot of types of shooters out there, a whole lot more than just "indie filmmakers".

I'm wondering why someone shooting any of the above subjects would buy a 24p/30p progressive camcorder when an FX1 for half the price would do the job -- except, of course, it too is almost 2 stops less sensitive than the PD170/VX2100.

The bulk of HD100s, I expect, will be bought by those who rent CineAlta, HDCAM, and Varicams who shoot 24p. These folks are already filmmakers who are used to working with a crew that knows how to light. And, they rent all the types of lights they need. They want perfectly exposed video. FX are done in post. Indie filmmakers, unles they are of the "run and gun" type -- will be the other buyers.

I'm sure JVC would love to sell to all the markets you have listed, and as long as these folks use light (news and sports always do), they can.

Until recently, weddings were also shot with extra light -- so if someone wants an HD wedding they need to accept more light no matter what camcorder they use. But, I can't honestly see why a wedding would be shot at anything other than 60i. There is no real reason for a 24p wedding!

There is no reason for everyone to find one camcorder ideal.

[Mikael Widerberg]

Here is a tiff image with a split screen example.

Its shot in my office and what you see is my white-painted wall.
The exposuer is set correctly at f4, all settings are at normal, no gain is used.

The light comes 90 degre from the side, so there is no light hitting directly to the lens.

Download the image (2.6mb) from here: http://www.plonk.se/splitscreen_nogain_f4.tif