I took some additional test shots with the 5DS and 24-70mm lens on a tripod with IS turned off. I took essentially the same shot at f8, 11, 16, and 22 in order to evaluate the sharpness of the overall image. Here is a down-sampled version of the scene:
http://www.throughjoeslens.com/Temp/TheScene.jpg
and here is a resulting 100% crop from each image:
http://www.throughjoeslens.com/Temp/Crop100.jpg
Make sure to view them at 100%. Web browsers tend to scale them down to fit the current window size.
The raw files were processed with the exact same settings in Photoshop. Notice the progressive softening of the 100% crop images with the most noticeable hit to the sharpness happening from f16 to f22. I believe that I read that the Diffraction Limit Aperture of the 5DS is f6.7 due to the pixel size (similar to the 7DM2). Plus the lens itself may perform better at specific apertures. I did the test with several runs and two different scenes to make sure the results were repeatable. I used Live View to focus in the area of the scene where the 100% crop was taken from and the 2 second timer to trigger the shots. Basically, I minimized the possibility of camera shake with every trick I know. Since the shutter speeds got progressively slower, ignore some of the softness in the foliage since there was a slight breeze. The rocks don't lie though. They definitely weren't moving. To be honest, with 50MP, the f8, f11, and f16 images could be printed at 20x30" and you probably wouldn't notice a difference other than the depth-of-field difference and that's only in the extreme foreground. I think softness of the one taken at f22 would be visible if compared at a very close distance to one of the others. I plan on repeating a similar test with Canon's 17mm TSE lens to get a better handle on how much of this softening is diffraction and how much was the lens.
Currenly using R5 and R7 mostly plus a variety of Canon RF and EF lens...
"
