Fresh ideas For New Cameras

[thierrylegros396]

Perhaps the Best Future Camera will use a lot of Actual Techniques.

To obtain good resolution, good DR, low Noise, manufacturers could use the following Receipt.

Certainly our eyes and brain use that receipt (not the first item).

• CMOS Sensor with integrated amplifier for each pixel.
• We don't need high resolution in very bright areas, so use small "Luminance Only" Pixels in very bright areas.
• We don't need high resolution in very dark areas, so use some "Pixel binning" in dark areas.
• Use all Pixels in "mid light" areas, but not too small Pixels to obtain a good chrominance resolution.
• Place red and blue Pixels to obtain small distance between others to achieve acceptable resolution when "Pixel binning".
• Use "quad (or more) pixel binning" in very dark areas.
• Continue to propose "Raw output files" to profit of future improvements in processing.

And why not use those techniques in new "raw converters and processors".

Thierry

[dalethorn]

Perhaps the Best Future Camera will use a lot of Actual Techniques.
To obtain good resolution, good DR, low Noise, manufacturers could use the following Receipt.
Certainly our eyes and brain use that receipt (not the first item).

• CMOS Sensor with integrated amplifier for each pixel.
• We don't need high resolution in very bright areas, so use small "Luminance Only" Pixels in very bright areas.
• We don't need high resolution in very dark areas, so use some "Pixel binning" in dark areas.
• Use all Pixels in "mid light" areas, but not too small Pixels to obtain a good chrominance resolution.
• Place red and blue Pixels to obtain small distance between others to achieve acceptable resolution when "Pixel binning".
• Use "quad (or more) pixel binning" in very dark areas.
• Continue to propose "Raw output files" to profit of future improvements in processing.

And why not use those techniques in new "raw converters and processors".
Thierry

Sounds like a prescription for moving some of the current software functions into hardware for improved performance. There was a previous discussion on LL about the differences between re-programmable chips and permanently coded chips  If the permanent types (like CPU's) can perform better, then the next issue would seem to be what happens when the chip needs to have its "firmware" upgraded.

[thierrylegros396]

Sounds like a prescription for moving some of the current software functions into hardware for improved performance. There was a previous discussion on LL about the differences between re-programmable chips and permanently coded chips  If the permanent types (like CPU's) can perform better, then the next issue would seem to be what happens when the chip needs to have its "firmware" upgraded.

Are you happy with your LX3 ?

I like my G10 very much, the only drawback is noise and associated DR limitation.

About 10 F-stop using Lightroom, what about the LX3 ?!

Have a Nice Day.

Thierry

[dalethorn]

Are you happy with your LX3 ?
I like my G10 very much, the only drawback is noise and associated DR limitation.
About 10 F-stop using Lightroom, what about the LX3 ?!
Have a Nice Day.
Thierry

The LX3 works well indoors without flash up to ISO 800, so I use it at shows and forums etc. for candid shots. DR is OK - haven't measured. Color can be a problem with artificial lighting, partly fixable by changing WB.  Outdoors the LX3 needs good light, unless the subjects are high contrast, otherwise the images become muddy.

[Ray]

• CMOS Sensor with integrated amplifier for each pixel.
• We don't need high resolution in very bright areas, so use small "Luminance Only" Pixels in very bright areas.
• We don't need high resolution in very dark areas, so use some "Pixel binning" in dark areas.
• Use all Pixels in "mid light" areas, but not too small Pixels to obtain a good chrominance resolution.
• Place red and blue Pixels to obtain small distance between others to achieve acceptable resolution when "Pixel binning".
• Use "quad (or more) pixel binning" in very dark areas.
• Continue to propose "Raw output files" to profit of future improvements in processing.

Thierry,
That's very complicated. How about a simple pixel of great quantum efficiency that varies its efficiency according to the amount of light impinging upon it. The first photons are received with great alacrity and full efficiency. Successive photons are welcomed with increasingly less enthusiasm, and the very brightest, photon saturated parts of the scene are positively discouraged.

That's the ideal. Whether or not this concept could ever be practicable, I wouldn't know.

[thierrylegros396]

Thierry,
That's very complicated. How about a simple pixel of great quantum efficiency that varies its efficiency according to the amount of light impinging upon it. The first photons are received with great alacrity and full efficiency. Successive photons are welcomed with increasingly less enthusiasm, and the very brightest, photon saturated parts of the scene are positively discouraged.

That's the ideal. Whether or not this concept could ever be practicable, I wouldn't know.

Why not, doing that is possible by changing the bias during capture.

More bias at first, decreasing progressively !

Good idea !

[EricV]

How about a simple pixel of great quantum efficiency that varies its efficiency according to the amount of light impinging upon it. The first photons are received with great alacrity and full efficiency. Successive photons are welcomed with increasingly less enthusiasm, and the very brightest, photon saturated parts of the scene are positively discouraged.

That's the ideal. Whether or not this concept could ever be practicable, I wouldn't know.

Take a look at http://www.photonfocus.com/html/eng/cmos/linlog.php

[spidermike]

How about a curved sensor? It would remove edge distortion and although it would need a new range of lenses, they would be smaller and cheaper.

Mike

[Plekto]

That's very interesting, actually.  By curving the sensor to a specific lens, you could solve many problems that plague smaller P&S type cameras.  And even more advanced method would be to have micro-motors curve the sensor according to the lens you use.  This would be a welcome feature to people who use prime lenses and want to extract every last bit of quality out of a lens.