Originally posted by Anvh Noise per pixel goes up but you also have more pixels so average noise actually stays the same roughly, you can check the sensor on DXO and it's very decent sensor the 24mp.
Not exactly true. Smaller pixels means there are less light photons reaching each individual pixel (or technically, the sub-pixels). Less photons means higher noise, all other things being equal. It does not matter that there are more pixels, unless you are averaging pixels together to remove noise (smoothing), in which case you lose the advantage of having smaller (and more) pixels in the first place. So there is a point of diminishing returns as pixels get smaller. With current technology, the limit is roughly where we are today, that is 16-18 Mpix in APS-C size format. A 24 Mpix APS-C is going to be a downgrade in almost every way compared to a well-executed 16Mpix APS-C design, like in K-5. Only under ideal lab conditions with good lighting will the 24Mpix APS-C outperform a 16Mpix APS-C, and it will take very good glass to do it. It will lose compared to 16Mpix when lighting is low or higher ISO is needed.
I work in medical X-ray imaging and have intimate knowledge of this stuff as an engineer, we have very much the same issues. Both X-ray and light are dealing with photons, and how to best capture them to create an image with good fidelity while keeping noise to an acceptable value. One significant difference is that in medical imaging, higher noise can be tolerable because we are not trying to create esthetic images, as in photography. Plus we are concerned only with capturing one "color", so there is no RGB Bayer type sensor structure.