For some reason, this debate never wants to go away. Every 6 months or so, someone who has little to no experience shooting EXR sensors posits that shooting L size and downsizing the image leaves one with more detail than shooting M size in the first place.
But this is wrong … because L size creates tiny details that are somewhat overwhelmed by noise and thus are more easily mistaken for low contrast noise by the noise reduction algorithms. This has been documented many times by myself and others.
So here, once again, is the perfect example. I shot this at the Leonard Cohen concert a week ago. The images were shot within 2 seconds of one another and they are obviously lit identically.
Look at the textures and the small details and you will see that surfaces, textures and tiny details are much better preserved in the M size images, whether you upsize the M to match the L or vice versa. These differences are subtle when shown in 100% crops as above, but when they appear in water or foliage, they are obvious even at web downsizing …
So especially if you a JPEG shooter, you are always better off shooting M size. There is no detail to be gained by shooting L size.
Resist the dark side 
