View Full Version : Split Screen Fixed Yet?


Bill Dunn
November 20th, 2005, 01:40 PM
Sorry to mention a products defects, but has the split been fixed in later released cameras? Can anybody with a new JVC100 confirm this? Starting a project and need to know damn soon.
Many thanks for your indulgences.

WBDunn

Robert Niemann
November 20th, 2005, 04:05 PM
Please read this statement from JVC: http://pro.jvc.com/prof/Attributes/faq.jsp?tree=&model_id=MDL101539&itempath=&feature_id=13#chips.

Bill Dunn
November 20th, 2005, 05:40 PM
Thanks for the official response, but I prefer a response from a seasoned user who has a late issue camera.
I'll wait till tomorrow without assuming its not fixed.

Merci

Nate Weaver
November 20th, 2005, 06:41 PM
It's not fixed. The calibration procedure in firmware that tries to minimize it has been improved reportedly, but it has not been "fixed". It likely will never be "fixed" in this hardware revision.

As a user of the camera, if you have not used the camera and seen what it's capable of on a proper display, then I implore you to do so.

[edit: Robert's response was a nice way of saying "It's never going to be "fixed", as such]

Stephen L. Noe
November 21st, 2005, 12:41 AM
Hi Bill,

I think Robert was trying to help with the link. Anyway, the split "issue" has been minimized for the most part and the subject has given way to the next hurdle, 24p capture and timecode. If you shoot 30p then this likely will not be a problem for you.

Have you had the camera in your hands?

Bill Dunn
November 21st, 2005, 10:03 AM
Stephen I did heft the camera in Toronto at the Dv Shop and liked the ergonometrics of the machine. The owner and I had a discussion on the split fault and he said it had stalled his sales of that item.
Our project requires a reliable camera for a doc shoot around the world. Low light, no light would be the rigour not the exception. So the JVC100HDV not the machine one ought to use?

Thanks

Dunn

Steven D. Martin
November 21st, 2005, 12:12 PM
I got my HD100 about a month ago and I have not noticed the split-screen yet. Can't say that I have pushed it to the limits, but I certainly have no complaints.

+Steve

Michael Maier
November 21st, 2005, 01:31 PM
the subject has given way to the next hurdle, 24p capture and timecode.


But this is not a problem with the camera actually.

Tim Dashwood
November 21st, 2005, 05:55 PM
But this is not a problem with the camera actually.
...just the fact that 720P24 HDV is a brand-new format. Most NLE software companies are still catching up (except for Cineform.)

Shannon Rawls
November 21st, 2005, 06:49 PM
That's because CineForm is awesome!
I'm gonna submit a propsal to them for the HDV Festival next year!

- ShannonRawls.com

Jiri Bakala
November 21st, 2005, 08:13 PM
Well, yesterday I was checking a US camera I was about to buy and it took me all of 5 seconds to see the SSE at 0dB. Of course, it was also there at higher gain but not necessary more pronounced. The suggested trick with changing iris didn't really get rid of it, neither did manual WB. Rather when closing the iris the 'line' began to disappear but it became replaced by a slight difference in the tonality of the two halves of the picture. The right side was darker, which brings up an interesting point; from what I have seen, all affected cameras have their right side darker, hence, the idea that the two processors (sets of) are not close enough in specs is not quite the whole story. If that was the case, some cameras would have their left side darker and other righ, quite randomly. The fact that there is (seems to be a pattern) may indicate that perhaps the automated factory calibration (if there is such a thing) is not set up properly. Or that the two sides for some reason employ different processors to accommodate additional functionality.

Anyway, lots of guessing on my part but based on the above mentioned observation. Please, feel free to enlighten me if you have other experience, more info or better understanding.

Michael Maier
November 22nd, 2005, 07:35 AM
How long was the camera on? It normally goes away at 0db once the camera is warm.

Jiri Bakala
November 22nd, 2005, 09:08 AM
It was still there after 20 minutes.

Silton Buendia
November 23rd, 2005, 04:23 AM
So we all know how the split screen effect is causes, two processors sampling the information on the chip. Half and Half and left and a right. Why not have one read every other line how things are done with interlaced video, one processor can read even lines the other odd. That way the information that is being read is never far from each other and that would eliminate the margin of error. Reading one left half and one right, at one point of the chip scan the two different processors are reading information that is on opposite sides from each other and begin so I can see how things can be calculated differently. Anyways just I tought why they didn't do it like this.

Werner Wesp
November 23rd, 2005, 05:06 AM
Thjat wouldn't help, because it would still be read out in the same time-interval - faster then with interlaced video - and thus creating more heat....

Keith Wakeham
November 23rd, 2005, 01:39 PM
So we all know how the split screen effect is causes, two processors sampling the information on the chip. Half and Half and left and a right. Why not have one read every other line how things are done with interlaced video, one processor can read even lines the other odd.

What your suggesting would require a feat of engineering that doesn't exsist. CCD's have a vertical transfer register. Want to goto the next line on readout - every line gets pushed down. So the charge that was held in pixel 1,1080 is now held in 1,1079 and then next in 1,1078 all the way down to 1,0. What you want means that they would actually have to wire in 2 million more traces into the sensor so that it could skip every other line. Just can't be done without huge gaps between rows and colums so you will decrease sensitivity and increase chip size and cost.

Also, interlaced video is read out just as fast as progressive. You need to read out x number of lines in progressive in y time, but in interlaced you need to read out 1/2*x number of lines in 1/2*y time. Lines/time = some frequency. x/y = 1/2*x/1/2*y. So the ccd operating frequency for progressive and interlaced are the same thing. OR 720x240x60 fields = 720x480x30 frames. Sensor still operates at the exact same pixel clock.

Guy Barwood
November 26th, 2005, 07:54 PM
But interlaced cameras only have to read out 1/2 the number of pixels every field (1/60sec for example), where as the HD100 samples the sensor in full progressive mode at 60fps (hence the 60fps capabilities of the analogue output).

The 60fps (48/50) is used by the camera for its frame blending. Perhaps a better choice would have been to use 30fps native only, therefore halve the data read out requirements and reduce the load on their processor to be able to use a single chip solution only and eliminate SSF altogether. Was there an absolute need for 60fps?

Alternatively a slightly larger body with heat pipes could have been used to dissipate the heat of a single chip processing 60fps. What would you prefer: a slightly larger body with NO SSF altogether or the current body with the chance of SSF? Personally I'd prefer the former. While the small body is 'cute', more important is image quality in all conditions...

Jiri Bakala
November 26th, 2005, 09:04 PM
While the small body is 'cute', more important is image quality in all conditions...
Absolutely!