I do agree with your overall point. Modern cameras are amazing, perform very well. Noise is much better controlled than a decade ago. While the R5 is not as good as the R6, it has better (lower) read noise at low ISOs than the 5DIV/5DIII.

Couple of quick points:
  • The second you downsample the R5 to the R6 size you have lost the benefit of the extra resolution.
  • Downsampling compared to RAW (which is still from DPP) is a more heavily processed image compared to a less processed image.
  • Read noise as calculated on photons to photos is different than what Bryan did with Read noise being measured under complete black conditions and then values calculated (shown here). Bryan's test is a highly illuminated subject and involve a level of processing, sometimes a heightened level of processing.
  • Read noise is is important for blacks, such as the nightscapes Kari does, but also for shadows and DR. Until recently, this is one of the reasons Sony outperformed Canon in low ISO DR tests, they had lower read noise. Canon mostly caught up at the 5DIV generation and caught up with the R5/R6/1DXIII generation.
  • Interpreting Bryan's tests are somewhat subjective. I do prefer the noise of the R6 over the R5. It isn't a stop, but it is better.


I actually really like what Bryan does with his noise tests, DPreview as well for that matter. But, these tests are very different in that they are well illuminated vs as black as possible. Thus, they eliminate all but the worst of read noise (1-6 electron/photon variance doesn't mean much when your pixel is hit with 20,000 but means a lot when the pixel is only hit with 10 photons/electrons). Which is fine for those that shoot well illuminated subjects. But read noise can become an issue in poorly lit conditions.