Like in my understanding in an image a pixel contains RGB values, but in the camera, each R G B is an individual pixel or a group of RGB is a single pixel, Can someone help me to understand this?
In: 5
The other answers here are missing one important part.
Unlike display screens that can produce colors, a camera image sensor cannot detect color. Only brightness.
So how do they get color?
They put a grid of “colored glass” in front of them. Many small pieces of red green and blue glass. If green glass is in front of a certain pixel, even though it’s ONLY sending a brightness signal, the camera will know what color it’s SUPPOSED to be based on its location.
All these brightness signals combined with the pre-programmed layout of the colored glass grid, the camera reconstructs what colors are supposed to exist where based on a input of purely brightness-only data.
Now the reason pixel combining happens where it takes “real” 3-4 pixels to be “one pixel”, is because all colors (except pure R G or B) are actually combinations of other colors. It requires input from at least 3 of these “fake color” pixels to re construct one region of this data into the combined final color. So it just blurs all 3 together and produces a single (blurred) “real color” output pixel.
This is a mosaic image sensor. They will always produce slightly blurry images due to this, unless the camera resizes the image after the fact or something. But the raw output will always be blurry.
This is also the reason when you see VERY RED scenes (or G or B) scenes in some videos, the resolution seems to drop and it looks pixelated. This may be due to the encoding codec, but often it’s a result of the mosaic in the camera. If something is pure red, only a single pixel in that mosaic grid on the sensor (the one with red glass in front of it) was picking up anything. The result ing output is a huge red pixel square (2×2 at least) with no blending. Newer cameras are better about blending the resulting output for this tho.
The only cameras that DONT have this built in blurry ness use a different sensor layout.
They have 3 full-sized sensors stacked one on top of the other. So every single pixel is actually 3 sensors deep, one of each color. Same thing as the mosaic, they ONLY sense brightness, ~~and have a glass filter~~ and take advantage of silicon’s natural color separating properties (R G and B penetrate silicon at different depths) and they embed a layer at each depth. except instead of one sensor using regions, it’s 3 fully dense sensors stacked.
Foveon was the first one of these 3 layer sensors that comes to kind. It was famed for its pixel-sharp images because physically it was able to actually produce them genuinely because every single pixel was 3 colors deep. Again it is only brightness signals but, the camera knows which one is supposed to be what color based on the depth of silicon covering it (since silicon separates colors at different depths naturally)
Latest Answers