If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Rate Thread | Display Modes |
#106
|
|||
|
|||
OT: Camera resolution (Was: How to connect to a wireless device...)
"Commander Kinsey" wrote in message
news I've always wondered: with hexagonal packing, where every alternate row of pixels is offset half a pixel-width horizontally from the row above, how do they map that offset structure to the rectangular grid of pixels that is assumed for any display or for any image-processing algorithm? Is there a matrix transform which will interpolate the value that a pixel would have if its sensor was directly underneath another one rather than being offset horizontally and vertically by half a pixel? I'd guess it probably messes it up. Remember trying to record interlaced video from a TV onto your computer and getting every other line slightly squint when things moved rapidly in the image? There's a name for it which I can't remember, which some video editors can reduce, but they never get rid of it entirely. I think it's called "combing" because the jagged edge that you get on fast-moving vertical lines looks like the teeth of a comb. Given that many sensors use hexagonal packing of pixels, I would imagine that you get a perfectly good results (ie without obvious artefacts) at the expense of a slight loss of sharpness compared with a true rectangular pixel array due to whatever convolution matrix is needed to map hexagonal to rectangular. That loss of sharpness may be mitigated by the fact that pixels of the same size can be packed slightly more densely, so you probably have more actual pixels than the effective (rectangular-pixel) output. |
Ads |
#107
|
|||
|
|||
OT: Camera resolution (Was: How to connect to a wireless device...)
In article , NY
wrote: Given that many sensors use hexagonal packing of pixels, not many. just fuji. I would imagine that you get a perfectly good results (ie without obvious artefacts) at the expense of a slight loss of sharpness compared with a true rectangular pixel array due to whatever convolution matrix is needed to map hexagonal to rectangular. That loss of sharpness may be mitigated by the fact that pixels of the same size can be packed slightly more densely, so you probably have more actual pixels than the effective (rectangular-pixel) output. the results are quite good. |
Thread Tools | |
Display Modes | Rate This Thread |
|
|