In my previous articles comparing the Nikon Df to other cameras like D800, D700 and D610, I posted images from the D4 as if they were from the Df in the articles (note that I clearly pointed out that the images were from the D4), because I was pretty sure that the Df had the same sensor. Some of our readers criticized me for doing that, arguing that Nikon’s sensor technology and the imaging pipeline might have changed since the introduction of the D4. I received the Nikon Df last week, so one of the first things I did was compare its performance against the D4 to see if I could spot any differences. Below is a detailed comparison between the two, which shows that both cameras utilize the same or similar sensor technology. So my previous comparisons are still valid and can be referred to for comparing between the different Nikon DSLRs.
I received an email earlier today from one of our readers, who provided a link to DxOMark’s article titled “Nikon Df review: New low light champion?“, which shows the Nikon Df having superior low-light ISO when compared to the D4. DxOMark ranked the Df at 3279, while the D4 scored 2965 before – a close to 10% delta in numbers. Does it mean that the Df is 10% better than the D4? No, those numbers represent ISO, similar to your camera ISO. So a score of 3279 is actually only around 1/10 of a stop better – a very insignificant difference. But why are DxOMarks’s numbers different across cameras, even if their sensors are the same? In this case, one might argue that the Df should have the same score as the D4, just like the D610 that should have the same score as the D600 (DxOMark ranked the D610 worse than the D600). Well, there are a number of factors involved in computing these scores. First of all, DxOMark down-samples / resizes images from cameras to around 8 Megapixels of resolution, as explained in this article. This alone creates a problem, because noise levels look different when software resizing algorithms are involved. Second, there might be differences in lighting, focus, white balance and other camera settings that might make one camera appear better or worse. Third, there are manufacturing / sample variances that also can potentially impact the results. DxOMark cannot account for those differences, so it just reports a different score. In my opinion, DxOMark’s scoring system is quite confusing and sometimes even misleading. Ideally, the Df and the D4 should get very similar scores, just like the D600 and the D610 should. Instead of crowning the Df as the “new low light champion”, DxOMark editors should have stated that the differences are simply lab / sample variances and other errors in calculations. To DxOMark’s credit, the review does state that “the sensor is the same unit in both models”, however, it should be clearly noted that the low ISO scoring system has a rather large error delta. Technically, if the Df scores higher than the D3s, then the D3s should not score higher than the D4. This is contradicting data that confuses a lot of people, including myself. If we take into account error variances and interpret DxOMark’s current sensor ranking based on low-light ISO, then all last generation Nikon DSLRs (with the exception of the D3s) have a very similar ISO performance. So Nikon Df, D3s, D600, D800E, D4, D610 and D800 should not be crowned by numbers from 1 to 7, but should all have the same “1” standing as a group. That would make a lot more sense, since that’s pretty much the conclusion that I came to while testing all these cameras.
Nikon Df vs Nikon D4 ISO Performance
Let’s take a look if we can find any differences in performance between the Df and the D4. The below images were shot in RAW, rendered by Lightroom 5.3 RC. Since Lightroom cannot yet read Df files, I had to cheat it into thinking that it was looking at D4 files. I won’t bother with my notes on each ISO comparison – see if you can spot any differences yourself. All EXIF data is attached. Let’s start out from low ISOs 50-1600:
And here is high ISO performance from ISO 3200 and up:
All four boosted ISO levels (H1, H2, H3 and H4):
Perhaps this needs a better eye than mine, but I spent some time looking at images back and forth and got tired of trying to find differences – I cannot see any!
As I have pointed out a number of times before, take everything you see on the Internet with a grain of salt (including what you see at PL). While I really like what folks at DxOMark do, their scoring system is confusing and can potentially lead to erroneous assumptions. The Df is not better than the D4. The D600 is not better than the D610. They are all pretty much the same in ISO performance.
Don’t get me wrong – I am not here to criticize DxOMark. They do a great job with giving us data that nobody else is willing to provide. The same can be said about us at PL and many other folks that assess and review gear. Too many variables are involved in these measurements and a lot can change depending on what tools are used in what conditions, and how one interprets the data. Canon and Sony folks are never happy with DxOMark’s assessments, because they typically score worse in comparison (and I came to a similar conclusion after looking at data from Imatest). In real life, such differences are very minor, especially on the same generation sensors. The grass might look better or worse on the other side, but in reality, it all depends on how you look at it and through which glasses…