Sony Intros 8 MP Digicam

[Ray]

Sounds like an interesting product. The fact that it can double as a high quality video recorder is a big step forward. Current systems all seem to fall into the category of either a still camera with inferior video capability, or a video camera with inferior still image capability.

This new product from Sony appears to allow progressive scan 640x480 images at 30 fps continuously, limited only by the size of the memory drive/stick etc. (I wonder if they'll have a 768x576 option for PAL countries   )

The 2.7 micron pixel pitch is impressively small, so I guess we can expect the usual problems of noise and unsatisfactory dynamic range - unless Sony can pull a few rabbits out of the hat

[Ray]

Would you like to bet on it?    Adding an additional colour might have worthwhile benefits that people will rave about, but I don't see any connection there to noise and DR.

[icke]

the just announced canon eos 300d could indeed be a great alternative.
even though it comes 'only' with a 6 megapixel sensor, it sports the same cmos sensor that can be found in the 10d. so the noise could be as low as in the 10d and the list price of ~€1200 including a lens can make this a real killer for similar priced digicams.

[BJL]

8 MP to record 4 colours = 6 MP to record 3 colours. I'd bet the resolution of this camera is no better than the EOS 300D/10D.

I do not see any basis for this division by the number of colours.
Standard Bayer pattern sensors mostly use the green pixels (half of all pixels) to determine the luminance, and then use the red and blue pixels to determine colour details. This makes sense because green is in the middle fo the colour spectrum and is typically the main determinant in overall brightness. Not surprisingly, such Bayer pattern sensors give roughly the same resolution as you get with half as many pixels in a device with all colours measured at each pixel (like a Foveon X3 sensor or a scanner.)

With the Sony design, instead of two green, one red, one blue pixel in every 2x2 group, there are two different shades of green plus red and blue. But since both "greens" are still in the middle of the spectrum, they might both be fairly useful for estimating overall luminance. So quite possibly, the resolution will be about the same as for a traditional three color Bayer pattern CFA.

But really, shouldn't we a be a bit patient and wait for actual resolution tests?

[Marshal]

DPreview has a 9 page preview of the upcoming Sony DSC-F828 advanced digicam with a new 8 megapixel CCD with RGB-E color. It also has 640X480 @ 30fps video mode with audio and takes CF, Microdrive and Sony Memory Stick. It can connect to comps by USB 2.0. Lens is a nonremovable 28-200 equivalent Carl Zeiss.

It has a small pop-up flash, but what it could really use, especially when used by advanced amateurs would be an external flash unit. A wide-angle adapter would also be a real plus when someone wants to go wider to perhaps 24mm. I don't know if Sony will be offering those.  

The camera is expected to retail for $1,200 USD.

Most of you may not be interested in something other than a D-SLR, but I remember Michael trying and reviewing the Nikon 5700 digicam last year, just for fun & curiosity if nothing else.

This could be a fun and addictive camera for anyone too intimidated by D-SLRs and just wanting to get their feet wet. The high quality video capability could be seductive too. The difference between this camera and others in the video department is that it can record non-stop in video if you desire. Most if not all other digicams have a 30 second to 1 minute limit on video clips. You could conceivably put in a 1-4Gb memory card and record a lot of either/both. Sony will inevitably increase the storage capacity of their Memory Sticks, but again, this camera can also use CF and MD.

http://www.dpreview.com/news/0308/03081502sonydscf828.asp

[Marshal]

Maybe that new RGBE chip design will counter some of those problems.

[BJL]

I am avoiding predictions on the performace of the new Sony 8MP 2/3" format sensor with its "microscopic" 2.7 micron pixels, but I am looking forward to the test results, because either (a) Sony, the dominant digital camera sensor maker who consistently get better noise results than competitors using the same (Sony made) sensors, has made a huge misjudgment, or ( this combination of very small pixels and very large numbers of them will perform significantly better than a lot of people have been saying is inevitable with such small pixels, though not necessarily up to DSLR standards.

Either way, we will get some useful new evidence to inject into discussion of pixel and sensor size choices.

One thing I hope is that people will pay little attention to 100% pixels scale on-screen viewing in comparing to sensors with far fewer pixels (i.e. everything under $5000 at least until the mythical EOS 3D arrives), and instead compare more on the basis of print quality.

[Julius]

8 MP to record 4 colours = 6 MP to record 3 colours. I'd bet the resolution of this camera is no better than the EOS 300D/10D. But the size of Canon's image sensor is 5.9 times the size of the Sony one used in this camera. Which one would you bet to have better image quality?

On the other hand, I do like the video capabilities of this camera. Now if only this camera would come with built-in image stablizer just like the Minolta Dimage A1 then I can convince myself to buy a non-SLR camera!

[Marshal]

Yes we should. And not just the lab type tests with a testing chart, but also real world or outside the lab shots.