DV Info Net

DV Info Net (https://www.dvinfo.net/forum/)
-   Silicon Imaging SI-2K (https://www.dvinfo.net/forum/silicon-imaging-si-2k/)
-   -   Hate to break up love fest, 1 CMOS vs 3 CMOS (https://www.dvinfo.net/forum/silicon-imaging-si-2k/68217-hate-break-up-love-fest-1-cmos-vs-3-cmos.html)

Douglas Call May 26th, 2006 06:35 AM

Hate to break up love fest, 1 CMOS vs 3 CMOS
 
I guess what I'm asking is what if Panasonic drops the JPG2000 HD codec into a camera like the AJ-HPC2000 so that you can shoot DVCPRO HD or JPG2000 HD, wouldn't the three (3) 2/3" CCDs of the Panasonic give it a much better light gathering capability than a single 2/3" CMOS sensor?

I sure all you pro's must know the difference by now of using three CCD/CMOS cameras over single CCD/CMOS cameras and yet it's sounding like everyone here thinks this camera will replace well proven 3 chip technology, I don't think there is a direct comparison.

This camera definitely has a place for indies etc, but I find it hard to believe it would quickly replace or compare with a Panavision Genesis or even a Cinealt 900.

I'm not trying to be argumentative, I'm just trying to understand if the people interested in camera are simply trying to find a direct replacement for shooting 16/super16mm film or actually think this camera will compete with broadcast HD tripple chip cameras.

I've bought into tripple chip 2/3" sensors and don't plan on looking back.

Jason Rodriguez May 26th, 2006 12:47 PM

There are some advantages/disadvantages to single-sensor designs.

Of course you do loose a little bit of sensitivity compared to a mono black and white chip . . . but even so, our final camera will be around ASA 350-400 which puts it on the same level as a Cinealta.

Also the 10-bit Cineform RAW compression is much better than the Cinealta's 8-bit HDCAM compression . . . there is no blocking, even over multi-generational use, and HDCAM pre-filters to 3:1:1, where-as we keep the full 1920x1080 raster of the image and decode to 4:2:2.

In regards to JPEG2K, you simply can't edit it. Thomson is learning that the hard-way with the Infinity. Also a J2K stream will have to be 2-3 times the data-rate of our codec (so over 200Mb/s) to keep up with the same compression efficiency we get with the Cineform RAW codec, since we are encoding RAW, and they are encoding the full YUV stream.

So I'd say that if you were pulling straight HD-SDI off the camera head of a 3-chip camera, and going to a full-raster or uncompressed codec, like a DDR, HDCAM-SR, or a high-bit-depth J2K (over 200Mb/s), then yes, the 3-chip camera would have clear advantage over our camera. But once you start to throw tape-based compression or low-bit-rate J2K in the picture, then I feel we have the upper hand with our effiency and flexibility by recording the direct RAW stream from the chip at a very mild compression ratio using a very "natural-looking" and visually-lossless wavlet codec.

In addition to the compression/resolution arguments, single sensors also have a simpler optical path, which reduces optical aberations, and once can use Cinema or B4 glass directly on the chip without using converters.

Is this camera a replacement for next-gen designs like the D-20 or the Genesis? No way.

But I do think that when you compare to the older generation Cinealta, etc. in resolution, dynamic range, frame-rate options, and post-flexibility, we have a number of advantages that camera can not deliver.

Forrest Schultz May 26th, 2006 08:59 PM

hell ya! you tell em Jason

Tip McPartland May 28th, 2006 12:04 AM

The lession of digital still cameras...
 
Most pro photographers that I know have abandoned film for digital cameras, at least partially because these digital cameras make great pictures.

And nary a one is three chip, all the way up the price scale. I've always suspected that inserting a prism into an optical path is a terrible thing to do to the light, and often wondered about the imprecision of extracting portions of what is really a multi-color gradient and aiming those portions, each still containing a gradient itself, into three monochrome chips which represent pure colors, unlike the spectral fragments that hit them. It sounds like a process fraught with inevitable problems, although obviously the technology has been brought to a very high level.

Maybe someday the 3-chip design will be as rare in video cameras as it is in still cameras, and perhaps this camera is part of that process. What's that other one that uses one big chip? Can't remember its name. I bet the good one-chippers has less chromatic aberration than most 3-chippers, as I've always suspected that prisms were the culprit in CA at least as much as lenses.

Tip McPartland

Jason Rodriguez May 28th, 2006 03:00 AM

BTW, One more pesky problem the prism presents is that you can't open wider than f1.6 . . . were-as with the single-sensor design, there's really no limit (except for the microlenses, which I believe limit you to f1 or something in that area, but I'm not exactly sure), and with Zeiss superspeeds, we can get a f1.2 for some really shallow DOF . . . at least very shallow for a 2/3" sensor, and the same DOF as 35mm shot at f4-f5.6 split (standard 4-perf, not Super35).

Balazs Rozsa May 28th, 2006 05:05 PM

When you design a lens for a prism camera, you can take into consideration the prism behind the lens. You have to correct the lens elements anyway for the chromatic abberations. Maybe it is quite easy to corrigate for a prism block. I'm not an expert, so I don't know, but I have always been curious how much harder it is to design a lens with small chromatic abberation when you have a prism behind it.

I remember in a previous thread I read there is no practical visible difference in CA using prism designed and non-prism designed lenses on 3 CCD cameras. So if the prism introduced chromatic abberation is so small, it is possible you can correct it in a lens specifically designed for a prism with only a small fraction of the effort required to correct the CA introduced by the lens elements (the lens elements without CA correction would result in an extremely abberated image). It would be nice to hear an expert about that.

Obin Olson May 29th, 2006 09:35 AM

And don't forget the artifacts from a 3ccd design on your out of focus areas, looks like "VIDEO" when I see the the R G B offsetting, gime single sensor any day, it's like film!

Ari Presler May 29th, 2006 10:32 PM

Reality TV Applications?
 
Quote:

Originally Posted by Tip McPartland
Most pro photographers that I know have abandoned film for digital cameras, at least partially because these digital cameras make great pictures.

Tip,

Can you think of any applications for setting up SiliconHD Mini Camera (Remote Heads) to do continuous shooting/recording for reality TV content? Where else can you get this level of quality/film look with nearly unlimited storage (put 4ea 750GB drives in a system and get over 3-days @ 24hours a day)?

We could setup the recording program to break up the files every 10 or 15min and some snapshot previews. How about we add some motion detectors to start the recorders only when somone goes into a room and now you have a week or 2 of storage.

You can even use a few monochrome version cameras with IR lights for nightime shots!

The first 24-hour HD Show with constinuously posted clips to the Web and a weekly on-air production?

Tip McPartland May 29th, 2006 11:22 PM

Got to think about that...
 
Ari,

Although a part-time shooter (JVC GY-HD100 with optional wide lens), I am also a full-time VP of Development for an emerging TV company TMC Entertainment. We just put three hours of "Walking the Bible" on PBS and now have a deal for six follow-up hours "The Route to Christianity." We also have have some very strong reality and documentary commercial network pitches upcoming.

The next six hours for PBS will be shot in the Holy Land as were the first three. They were shot with Sony HDCAM and looked really, really good. The Wall Street Journal even took notice, calling the show "three splendid hours."

What would be the configuration of your camera's recording unit that would travel very well, survive high temperatures, and give us a more cost-effective yet equally beautiful picture? Bear in mind that we shot 1080i so your 1080p might hold up very well.

As for the reality show, that does start the gears spinning, although nothing has yet come to mind. I'll probably wake up in the middle of the night with a concept that would take advantage of your new capabilities. Hopefully I'll remember it in the morning!

Please feel free to call me any time this week at work -- 310 806-4400 or e-mail me at:

tipsd9video@msn.com

Tip

Ari Presler May 30th, 2006 07:26 AM

Quote:

Originally Posted by Tip McPartland
Ari,

What would be the configuration of your camera's recording unit that would travel very well, survive high temperatures, and give us a more cost-effective yet equally beautiful picture? Bear in mind that we shot 1080i so your 1080p might hold up very well.

Tip


I think the head-on-wire with recording on a notebook could be an excellent option. What is the timleline for the shoot. We may be able to supply a limited number of units of our current configuration (if you are interested). Have you looked at the footage from "spoon"? They just posted a slew of clips (http://indiefilmlive.blogspot.com)


As jason stated: The 10-bit Cineform RAW compression is much better than the Cinealta's 8-bit HDCAM compression . . . there is no blocking, even over multi-generational use, and HDCAM pre-filters to 3:1:1, where-as we keep the full 1920x1080 raster of the image and decode to 4:2:2.


Quote:

Originally Posted by Tip McPartland
As for the reality show, that does start the gears spinning, although nothing has yet come to mind. I'll probably wake up in the middle of the night with a concept that would take advantage of your new capabilities. Hopefully I'll remember it in the morning!

That is exactly what I was hoping you would say...someone has to be first to apply the new technology!

The Real HD World
Joe SchmoHD
Big Brother is watching you in HD
The Surreal HD Life
HD House Rules
Talking HDs
.....

I will give you a call to strategize.


All times are GMT -6. The time now is 11:47 PM.

DV Info Net -- Real Names, Real People, Real Info!
1998-2024 The Digital Video Information Network