View Full Version : Low light performance?
04-05-2005, 05:23 PM
I was wondering, what will Panasonic (and JVC for this matter) do about the bad low light performance normally associated with 1/3" HD cameras? Maybe Panasonic will use pixel shift again? I sure hope they both have a work around because I have heard so many complaing about the Z1 in low light. It seems it looks great, as long as you have lots of lights. That's sure a very limitating factor for 10k cameras. Specially because they are being markted as a pro tool. What do you think will the work around be?
04-05-2005, 05:30 PM
Jan mentioned there are a lot of things you can fix in DSP.
04-05-2005, 05:32 PM
If you want to know exactly what those things are talk to the designers:
Sako Kohara and Katsuyuki Taguchi in Japan. Good luck getting past the quizzes with Mr. Taguchi, though.
I suspect we will see pixel shift. At least to achieve 1080p. 720p is still up in the air in my mind. The lower compression should help out a bit. Juan's uncompressed DVX mod has much more latitude than standard DV and I believe DVCproHD/50 will do better as well. Maybe they have come up with some new high bit DSP? Guess we'll find out at NAB! There is a reason, however, that the camera is going for such a low price and 1/3" chips would be a reasonable conclusion.
04-05-2005, 05:41 PM
Yeah, I think you are right. Pixel shift is basically a no brainer here. But I think they sure have something up their sleeves. Specially after all the complaining about the Z1.
04-05-2005, 07:03 PM
Maybe Panasonic will use pixel shift again? I sure hope they both have a work around because I have heard so many complaing about the Z1 in low light.
I'm sure hoping there's some new tech to accomplish it as well. Yes, the Z1 has worse low light performance than the PD150, but -- it's a whole lot better than the JVC HD1! So those 18 months between introductions must have helped them come up with some tricks. Hopefully the new JVC and the Panasonic will have similar tricks up their sleeves.
04-05-2005, 07:05 PM
Well, the HD1 was a single CCD using a Bayer pattern, do it really can't be compared here...
Actually I imagine that the single chip would be better for low light sensitivity. WIth a 3chip camera you have to split the light three ways; one for each CCD. 3CCD is higher res and has better color reproduction but I don't think it's a winner for low light.
04-05-2005, 08:01 PM
Yes it is.
In a single CCD camera you suppress the light by color filters, in a 3 CCD camera you use a prism to split the light into its primary colors.
You loose a lot of light in the colorfilters, a prism is very effective.
Also if both systems are 1/3" CCDs you'd have to squeeze 3x the pixels onto the single CCD, so the pixel size would be at least 3 times as small and need 3 times more light...
I see no reason for light loss in the filtration process. It's all electronic anyway after the image has been captured.
04-05-2005, 08:23 PM
CCDs are inherently monochromatic. A single CCD camera has a filter over ever pixel which allows it to only capture red, green, or blue, and these filters block more light than a prism.
Still, I'd take a 2/3'' rockwell CMOS over three 1/3'' CCDs if what I hear about it is true.
Yeah I'd love to know how they are doing the bayer filter. I know it can be done either in post or directly. All you would have to do is assign each pixel a color value after the image was captured. I know this is often used in high res CMOS cameras. You're probably right though. For low end stuff it probably is a physical filter.
EDIT: by in post I mean in camera after the image was captured although it certainly could be done in the traditional meaning of "post"
04-05-2005, 08:59 PM
Every color single CCD and CMOS imager is monochrome by nature, and has a physical color suppression filter, usually in a RGBG Bayer pattern. Out of the imager you get a coded monochrome image that you will have to decode and interpolate to make it into RGB data, but there is no "filtration" done in post.
Usually there is also a low pass blurring filter in-front of the imager to suppress any moire from high frequencies.
I'm not sure man. The guys over at DVinfo building their own HD cameras have been implementing the bayer patter in post production. Maybe I'm just misunderstanding?
EDIT: I do realize the CCD is monochrome :)
04-05-2005, 09:13 PM
Yes, you misunderstand.
The guys you talk about use single chip CMOS imagers with a physical bayer pattern.
What they capture to a hardrive is a 10 bit single channel RAW frame that is encoded by the bayer pattern.
The software they have written will decode that bayer pattern into a 10 bit three channel (RGB) file. If you have not filtered it physically first it would be impossible to decode afterwards as all pixels would be monochrome.
It would be the same to try and decode colors from an old B/W picture, it's impossible...
04-06-2005, 03:01 AM
Wonder how good those home made HD cameras will be. Specially now with the 2 new releases from JVC and Panasonic. I mean, will it still be worth the trouble when you can get the new cameras for a fair price? Will the home mades even perform on pair?