After finally releasing their score for the OnePlus 6T yesterday, DxOMark have followed up releasing their Pixel 3 grade. Sneak preview: best Pixel ever!
After the OnePlus 6T scored an impressive 98, the same as the Pixel 2 last year, the Pixel 3 bested it by scoring 101. Considering that the Pixel 3 is often rated the best smartphone camera available it is surprising it did not fare better. Of note though is that DxOMark did not take into account Night Sight – all photos and videos were captured using default settings.
In the end DxOMark rated the Pixel 3 as the best single lens Android smartphone camera but let’s be honest, what do they have to beat when every other manufacturer is putting dual and triple lenses in their flagships?
Once again Google was commended for the improvements to their computational imaging along with their bokeh and zoom functionality. Unfortunately it may be time for Google to look at putting a second or even a third lens on their flagship phone because even though they had improved in those areas they were still behind the top multicamera smartphones in these areas.
Google scored well in autofocus, detail preservation and accurate target exposure but did not fare so well with its ghosting and colour fringing. At this level of camera you are splitting hairs to separate one from the other so the differences can be very small.
There was no mention of the Pixel 3 not saving images when exiting the app before the image processing has finished – – although we have seen this on other phones including The Huawei Mate 20 Pro so this may inherently be an Android issue.
At this moment in time buy any flagship and you are going to get a smartphone camera good enough to great pictures most of the time. So much so that DxOMark may well be irrelevant for flagships these days. From what we have seen the Pixel 3 can take some amazing photos, although the Huawei Mate 20 Pro still takes the cake for us.
The first thing you need to know about dxo mark is it is a reasonable guide at best. While I am sure they keep improving their testing methods at least in proper camera there could be as much variation between different copies of a model as between models. So some cameras are often ranked lower than they should and others are higher as the variations in their testing models and error tolerances are not good enough to accurately separate models. I am not sure how well computational photography actually translates to image quality.