Sorry for the prolonged gap in posting. Back now from holiday and have processed most of the resulting pictures so normal service is resumed.
This post summarises what I've learned over the past few months about difficulties in reproducing accurate reds. It's a technical point but not an abstract one, as I hope that what I've learned will help me get more faithful colour reproduction in my pictures over time. I hope that it may help others who are also struggling with this.
The issue is that it is comparatively easy to lose definition in strong red tones so they look over-saturated and it is difficult to faithfully reproduce certain intense reds, especially blue-toned ones. This is a particular problem for me as:
- I love colour;
- red is my favourite colour; and
- I want to make very high quality images.
I have spent some months trying various shooting and processing options to get good reds, and also consulted the Nikon Owner technical helpline, and here (with thanks to Simon Stafford) are my findings to date.
Digital sensors have difficulties reproducing pure primary colours. They record everything in red, green and blue. so when photographing a pure bright red (for example) the red channel will contain very little tonal information, or detail. That is contained in the blue and green channels, which when added to the red give detail, tone and contrast. If the shade of red (or indeed blue or green) is very pure then it will be difficult for the sensor to capture tonal detail.
In addition strong intense reds come close to the infrared end of the colour spectrum, and despite recent advances in infra red filters, IR pollution can also affect the way the camera records red.
So there are inherent limitations in what can be recorded. There are however a number of steps that can be taken to ensure that tones are captured and reproduced as faithfully as possible.
Firstly, it's important when taking pictures to ensure that the exposure is not causing the red channel to clip. This requires checking the individual channel histograms as an individual channel might be clipped without necessarily showing on the overall histogram. The same applies for blue and green also of course, but problems occur less often: in the real world there are not so many intense blues and there are twice as many green sensors as either red or blue and so there is more latitude in these.
Shoot Raw, and set picture controls to neutral with no sharpening. It is comparatively easy to increase contrast and sharpen a picture in processing but very difficult to take it out if already present. Look at my shots of Anthuriums at Kew below to see an example of this. These pictures came out of the camera so bright and sharp that I thought they already looked almost over-processed. I had had this problem before but hadn't quite got to the bottom of it. I have since changed my picture settings from standard to neutral. Although pictures now require a little more post-processing I have greater control over the end effect.
Set the camera to record in 14-bit depth and in Adobe RGB.
Set the Raw converter to ensure that it works in 16-bit mode.
It is possible to work in wider colour gamuts than ARGB in processing software. I have tried this but can't then reproduce the effect on internet forums, though I have found that it is possible to improve the colour quality of prints if you have a good printer.
In the Raw converter, select the red channel and make adjustments to luminosity and saturation there - it's much more effective and easier to correct colour at the raw stage than in Photoshop or similar programs. I confess that I don't feel I've got this stage right yet. I have quite deliberately concentrated in recent months in improving my initial capture skills and have some further work to do on post-processing.