There’s been a couple of different complaints I’ve seen on different fora in recent weeks that all center around “this camera/sensor produces wrong color.” Typically, these are accompanied by some casual or non-equivalent examples that “show” the problem.
My experience has been different (and not just with Nikon, but across multiple brands). And every time I’ve gone all forensic on someone what ends up happening is that we find that something wasn’t understood correctly, something was assumed, or something wasn’t done correctly by the user. Sometimes all of the above.
So let’s talk about what involves color throughout the photographic process.
- Light. Light can be continuous spectrum, discontinuous spectrum, constant or sequential. So what wavelengths are showing up at the front of the lens are the first variable that has to be understood and possibly managed.
- Filtration. A filter on the lens and the filter over the image sensor can have impacts on the spectrum that eventually reaches the image sensor. You have control over any filter on the lens. Nikon has control over the filter over the sensor. In the DSLR era, the Nikon pro cameras (D4, D5, D6) tended to have more red sensitivity than the consumer/prosumer cameras (D500, D750, D850). I believe that tuning was deliberate to deal with low indoor lighting better (which tends to be red biased). However, in the Z System, Nikon has been pretty consistent with their filtration. I’m not finding any significant changes from model to model.
- Settings. If you’re taking JPEGs your white balance obviously needs to be correct. Nikon’s Auto white balance is not perfect. Indeed, so much so that if you’re taking photos outdoors in natural light, I strongly recommend that you use Natural light auto instead if you require automated white balance. If you control the lighting, you should also control the white balance (using a Preset you measure correctly). The reason why this step in the sequence is important for raw files occurs in the next:
- Processing. Nikon applies what they call white balance preconditioning to the raw Digital Numbers that are recorded in a raw file. The problem is that if your white balance is off, the preconditioning is also off, and that will have impacts further down the processing chain. If you’re using NX Studio to convert a raw file, it does the same thing as the EXPEED chip in the camera; Nikon is absolutely consistent about this. However, if you’re using a different converter you start running into interpretation and math differences. Adobe tends to produce more orange luminance (as well as other shifts), and its math engine isn’t great for big white balance swings (speed is emphasized over accuracy). Color models also play a part in this. Adobe, CaptureOne, and DxO all have different color models, and they build Nikon presets using different sample/test/change processes. It pays to learn exactly what each converter is doing, as it will impact the results. That said, I’ve been able to get solidly consistent and neutral results from all three of these, though not without some attention to detail and in the case of Adobe, some preset changes.
- Workflow. This is one area that doesn’t get enough coverage. For instance, if you use DxO PureRAW to pre-process an image, the process of taking a NEF to DNG and then into a converter seems to change the color values along the way. It’s one reason why I stopped using PureRAW, as it didn’t just add one workflow step, but several in order to correct the color integrity. You might not see it at first, but quite a few adjustments you can make (ACR, plug-ins, you name it) distort color values when you make them. You might notice this if you use huge Dehaze adjustments in ACR, for example, but that’s only one such control/change that may trigger color shifts.
I’ve written about this before, but basically when you get into these “wrong color, bad color, different color” arguments, the way you solve them is by starting with a known input (e.g. ColorChecker in known light) and a known output (e.g. Epson print on known paper, or sRGB JPEG). If the output doesn’t match the input, then the problem is somewhere in the above list. You simply follow a color-managed process and quickly discover the thing that’s making the change.
But that thing is not the image sensor. The image sensor is basically agnostic in this debate. Even the filtration over the image sensor, once understood and potentially accounted for, doesn’t really play in the debate.
———
Just to be complete: I have seen two cameras in two decades where there was an image sensor/filter problem. While I’ll never know exactly what the problem was, the two most likely candidates would be a badly mismatched green pattern (alternate rows of green in the Bayer pattern had different responses) and some issue with ADC timing. However, this type of problem is extremely rare, and Nikon will take care of it under warranty should you encounter it.