How would you like your future camera to have a much higher resolution, a higher dynamic range (enabling you to capture much more detail in the darkest and the brightest areas of your photos) and be cheaper?
It may sound too good to be true, but Edoardo Charbon, of the Technical University of Delft, in the Netherlands has presented a paper at an imaging conference in Kyoto, Japan which promises just this.
Digital camera sensors work by collecting light in “buckets” on the chip, then converting these analogue signals into digital levels. There is a lot of circuitry involved in the conversion, which makes the chips large and leads to noise in the image, especially with higher ISO settings.
Charbon’s idea was the remove the black plastic packaging from the chips, focussing the light fall directly on them. The light causes the memory cells to register as “on” or “off”, allowing a digital image to be decoded from them.
Why is this better than conventional sensors? You can fit around 100 times more pixels in the same space (hence the name Gigavision).
One important problem Charbon had to overcome was that the cells on memory chips are binary – either registering light or dark. Special algorithms, known as spatial oversampling, are being developed that sample an aray of 100 pixels, in order to estimate greyscale values. The benefit of this is the ability to produce an image that doesn’t over-saturate as easily as conventional sensors, giving more detail in the very dark and bright areas of an image. “Gigavision cameras do not saturate anywhere near so easily, so we’ll be able to use it for high dynamic range applications like medical imaging,” says Feng Yang, the developer of the algorithms.
It is hoped to have a working version of the chip working early next year.
