You would end up with very little light gathering capacity. The number of photons exposing each pixel would be only the ones emitted from the source that managed to strike the pixel directly. The same as a pinhole camera.
The thing about a lens it is lets you grab a relatively huge area of light and sort it all out so the right bits go to the right pixels.
I already did arithmetic once today (erroneously as it has transpired), so I'll pass on this, but although a metric dumptruck load of photons fly off ov things, when you are 10 meters away, and have an image element the size of a CMOS pixel, there aren't a lot of photons per second to let you sort out 256 levels of intensity without random noise.
The thing about a lens it is lets you grab a relatively huge area of light and sort it all out so the right bits go to the right pixels.
I already did arithmetic once today (erroneously as it has transpired), so I'll pass on this, but although a metric dumptruck load of photons fly off ov things, when you are 10 meters away, and have an image element the size of a CMOS pixel, there aren't a lot of photons per second to let you sort out 256 levels of intensity without random noise.