nazdravanul
Offline
• •
Upload & Sell: Off
|
 p.6 #9 · Nikon/Canon/Sony Image Quality vs Panasonic's Multi-Shot High-Res Mode | |
RoseandCharles wrote:
Maybe I'm misunderstanding what you guys are saying, but maybe you're missing some of the point too. When the camera is using multi-shot, it's doing two things: cancelling out the bayer pattern for full color information at each channel, and also shifting the sensor by a half pixel and effectively sampling another point halfway between each pixel (again with full color per pixel). If it was quite that simple, it would in fact be true extra resolution up to the limitations of the lens/diffraction.
But it's not quite that simple. Each pixel is sampling some area of photons, like a bunch of buckets on the ground trying to catch rain water. In other words, they're sampling light within some area, and there is some edge area that is being missed. When you then shift the bucket pattern by half, you're now sampling a different area (likely the exact area that was being missed before, but there is also likely some overlap. This means it is absolutely catching more real detail, but it's also doing so with some area of confusion (or blur). This is still better than any single shot with no bayer cancelling, and almost certainly better than without the half-shift with bayer cancelling, but it's physically not the same as actually having a sensor with the true higher resolution (I withhold judgement as to whether it's better or worse).
Note also that for every extra shot, assuming the same exposure, you're capturing twice again (of the original) as much information about the scene. That alone (like average stacking to reduce noise) is going to give a better sample of the truth. In other words, more exposures, even just averaged together, is (essentially) always better. Using those extra exposures to cancel out the bayer pattern is a pretty clever way of using that data.
Is everyone arguing based on this same model or am I missing something?...Show more →
Nope, you are not missing anything, that's pretty much it 
Also we are arguing about the real life implications of not being able to preview the full 16 pixel shot, as the assembly is not done in camera. Essentially, in camera you are assessing the 61 MP file at 100% which is essentially assessing DOF at 25% magnification of the 240 MP file. The DOF limits are going to be significantly different on the larger resolution file vs the smaller one, at 100%. That, combined with the fact that you have no idea, until you reach a PC, if the assembly is successful (checkerboarding, un-fixable artefacts etc.) , raises serious concerns about the real life usability of the multishot, in the field (outside a studio with a fully tethered solution).
Also one shot bayer 100 MP or 150 MP vs perfectly executed 200 MP true colour multishot - better or worse ? On paper, the multishot should be better, primarily due to the true color capture vs the bayer capture. But, of course, the one shot doesn't suffer either from subject movement artefacts, or in the field effective IQ assessment (this last one seems to be done right by the Panasonic multishot implementation, with in-camera assembly of the multishot).
|
