Imagine an image sensor that can expose different parts of an image differently, not by using multiple shots and blending them intelligently, but all at once on a per-pixel basis. Seem like science fiction? We thought so too, but that’s exactly what Sony is proposing in a recent patent.
Dug up by one of the posters on SonyAlphaForum, the patent was filed in July and published on October 30th. It shows a sensor that can expose individual pixels differently depending on where the darker parts and lighter parts of the scene are.
So dark shadows would become ‘long exposure pixels’ while bright highlights would become ‘short exposure pixels,’ essentially spitting out a high dynamic range image without the need to take multiple exposures and combine them.
The obvious challenge here, as pointed out by Sony Alpha Rumors, is that taking a picture of a fast moving object could turn into something of an abstract nightmare when some pixels are exposed longer than others. But Sony knows this and address it in the patent.
They propose using a ‘sensitivity-specific interpolation process’ to come up with a final image by combining an entire short and entire long exposure version and thus eliminate blur — more like traditional multiple exposure HDR.
To dive deeper into the technology at play here, you can read the full, not entirely easy to understand patent by clicking here.