A small filter could be made of a thin glass that would have a precision matrix printed with alternative transparent and neutral gray areas. The filter would slide in front of the CCD, very close to it's surface to be as close as possible to the CCD's focal plane. For this to be achieved, the printed side should be the one closer to the CCD.
Advantages:
-The pixels under the transparent areas of the filter, could then be used to capture the medium-high to low lighted areas of the image, while simultaneously the pixels under the dark areas could be used to capture the highlights preserving detail and color information.
- More filmlike picture, 5 stops or more CCD dynamic range extension.
- Easier and faster exposure, less need to switch ND filters.
- Capture in situations that otherwise would be impossible, either to preserve highlights, or avoid crushing blacks, or both.
Implementation:
The areas of the matrix would be rectangular. Each area would cover 4 pixels (Green,Red,Blue,Green, as many as needed normally for a full Bayer color decoding). By covering half the pixels, the resolution will drop to about half, but because of the "holes" in between, to avoid alias (like moire) a res reduction of about 3 times might be desirable. So I guess it will be excellent for 720p and SD.
[Alternative Implementation (very high precision):
Instead of a matrix with 4 pixel squares, there could be used a pattern of pairs that cover the GB and GR pixels, spread in alternative directions to reduce the alias effect and retain the maximum resolution (half+).]
The major difficulty of either implementation is the alignment of the matrix pattern with the CCD pixels. This either requires the filter and the slide mechanism made with very high precision, or could be done electronically each time the filter slides in, by using the CCD as the sensor that watches what pixels of it's Bayer pattern are covered, with the help of an internal illumination source (led) . The sliding could be done with a small motor, or even a lever and the X-Y micro-alignment with just a couple of small electromagnets and maybe one more to lock the position. Of course the whole process should take about half a second or less.
I posted this idea because I think the dynamic range is the last and major difference between film and digital. A camera with double the dynamic range and an S curve would be more "filmic" IMO than with any other trick...
Advantages:
-The pixels under the transparent areas of the filter, could then be used to capture the medium-high to low lighted areas of the image, while simultaneously the pixels under the dark areas could be used to capture the highlights preserving detail and color information.
- More filmlike picture, 5 stops or more CCD dynamic range extension.
- Easier and faster exposure, less need to switch ND filters.
- Capture in situations that otherwise would be impossible, either to preserve highlights, or avoid crushing blacks, or both.
Implementation:
The areas of the matrix would be rectangular. Each area would cover 4 pixels (Green,Red,Blue,Green, as many as needed normally for a full Bayer color decoding). By covering half the pixels, the resolution will drop to about half, but because of the "holes" in between, to avoid alias (like moire) a res reduction of about 3 times might be desirable. So I guess it will be excellent for 720p and SD.
[Alternative Implementation (very high precision):
Instead of a matrix with 4 pixel squares, there could be used a pattern of pairs that cover the GB and GR pixels, spread in alternative directions to reduce the alias effect and retain the maximum resolution (half+).]
The major difficulty of either implementation is the alignment of the matrix pattern with the CCD pixels. This either requires the filter and the slide mechanism made with very high precision, or could be done electronically each time the filter slides in, by using the CCD as the sensor that watches what pixels of it's Bayer pattern are covered, with the help of an internal illumination source (led) . The sliding could be done with a small motor, or even a lever and the X-Y micro-alignment with just a couple of small electromagnets and maybe one more to lock the position. Of course the whole process should take about half a second or less.
I posted this idea because I think the dynamic range is the last and major difference between film and digital. A camera with double the dynamic range and an S curve would be more "filmic" IMO than with any other trick...
Last edited:
