Here's a raw conversion from a Sony A7R2 image shot in low light at ISO 2000. The difference between the two images is substantial.
Here we have lifted the shadows of an image with high dynamic range taken at ISO 5000.
It's hard to believe that we're actually looking at two different raw conversions of the same file!
With Frame Averaging, we make several identical exposures of a scene, import them as layers into Photoshop, then average the layers. Sensor noise is "averaged out" in this process, leaving only the pixels which represent the actual subject. Before the days of sophisticated noise-reduction software, this was considered an effective solution, albeit only for stationary subjects.
Performing a test with my 42-megapixel Sony A7R2 set to ISO 10,000, I lifted the deep shadows by 100% in Adobe Camera Raw. DxO performs better than frame averaging, until we reach around 12 frames. Past that point, frame averaging is better, but only in some ways. To see any differences at all, we need to view the image at 100%. How often do we display at 100% ?
The DxO image holds its own but an 18-frame average actually retains more detail in some places. However, even 18 averaged captures are not enough to completely remove the sensor noise that DxO removes in a single pass.
DxO is certainly more convenient: only one exposure is required. Most importantly, it works with photos of all subjects, not just stationary subjects shot on a tripod. Move over, frame averaging. Hello, advanced noise reduction software !
Large Sensor Under Ideal Conditions
If we compare properly exposed and un-cropped images shot with a full-frame 42 megapixel Sony A7RII at base ISO 100, the difference in noise levels between Adobe and DxO Pure Raw2 raw conversion is modest. Therefore we don't need this tool if we're shooting under ideal conditions with a large sensor and we're not cropping. But how often is that ?
Free Camera Upgrade
With this tool you can shoot your camera at much higher ISO, under previously unworkable conditions. You can revisit high-ISO rejects from years past and process them into grainless images.