Comparing sample images from different sources is not a reliable way to judge signal-to-noise ratio unless the photos were made under exactly the same conditions. Even small differences in lighting, exposure, processing, lens, or scene content can overwhelm the noise differences between cameras.

A better approach is to use independent lab measurements from a sensor database such as DXOMARK. Controlled testing is much more objective and useful for comparing noise performance than trying to infer it from unrelated sample photos.

So, in practice: don’t trust cross-site sample-image comparisons for SNR unless the test setup is identical. Use standardized sensor test data instead.