Better color or monochrome at 25% resolution?

[StoryinPictures]

I can't imagine this hasn't been discussed (and if it has, I would appreciate links if you have them), but I can't figure out search terms to find it.

I have been wondering if, instead of demosaicing an image the traditional way, if there might be an advantage to taking four individual sensors on a Bayer and converting them to one pixel. This would, of course, cut the resolution by 75%, but it would, in essence, create a larger "individual sensor" which reads all the color information by combining the values.

If this worked, the red channel would only contain information from the red sensors, the green channel would contain the information from combined green sensors and the blue channel would contain only data from the blue sensors.

Would this create a file with cleaner colors since they are not interpolated?  Would it create better monochrome files?

[Mark D Segal]

I think what you are referring to is "binning" pixels. There is literature on it.

[kirkt]

You mean half-size image rendering?

http://chrisjones.id.au/Half%20Size/halfsize.html
for example.

Some raw converters permit this option (dcraw, libraw, Raw Photo Processor, probably Raw Therapee).

kirk

[StoryinPictures]

You mean half-size image rendering?

http://chrisjones.id.au/Half%20Size/halfsize.html
for example.

Some raw converters permit this option (dcraw, libraw, Raw Photo Processor, probably Raw Therapee).

kirk

Exactly that!  Thank you.

Which application are you using in that screenshot?l [found it: Raw Photo Processor]

[Bart_van_der_Wolf]

Would this create a file with cleaner colors since they are not interpolated?  Would it create better monochrome files?

You'll (probably) trade fewer false color artifacts for more Luminosity aliasing, and lower resolution. You could create a similar effect by downsampling with sub-optimal filtering.

Cheers,
Bart