I don't understand this statement. When you use the raw converters (LR/ACR) to move from colour to B&W, one is engaging an instruction set making the program render the colour data in grayscale, wherein the use of the colour group sliders influences the luminance of the gray tones representing those colour groups. If this is what we are talking about, I have never ever seen any colour noise in printed images from such a process. Nor have I seen this problem in fully converted colour images converted to B&W using the Photoshop CS5 Black and White Adjustment Layer, nor have I seen it using Nik Silver Efex Pro versions 1 or 2. I've made B&W images using any of these techniques. Conversions are neutral and not showing colour noise in print sizes ranging between 13*19 and 17*22 inches. Are there other processes you have been using which do show such a problem, and if so what are those processes and in what viewing conditions do you see it?
Mark and Slobodan
I probably shouldn't have made this comment, but I just assumed that it must be a well-known issue. Specifically, I am talking about my Hasselblad 3FR files, imported into Lightroom and converted to B/W using the grayscale conversion tools within LR. When I first started doing this, I made up a set of B/W LR pre-sets which emulate the spectral response of various B/W film stocks, and some others which replicate the effect of various coloured filters (yellow, green, orange etc). This was by no means an original idea, I simply studied how trial versions of Convert to B/W Pro and Silver Efex worked and rolled my own.
All was good, but then I started to notice that even at 100 ISO or 50 in some
images I had serious colour noise in areas of flat tone and low detail. Clear blue skies are the obvious area where you see this first, and of course you only see it on screen at 100% when applying capture sharpening and noise reduction. With the 'Blad files, there is always a tiny
bit of luminance noise, but normally that is no problem. One area which showed up the colour noise issue big time was skin tones in semi-shade, where massive blotchy noise was apparent. In a complex landscape with lots of grass, trees, and a cloudy sky you will barely notice a problem, as the high-frequency detail masks the noise.
Now the noise reduction in LR is very good, and will deal with the issue, but I couldn't understand why I had so much CN at low ISO. So I examined an area at 100%, and switched the file around through my presets and also back to a flat grayscale conversion (all sliders set to zero). Lo and behold, when the grayscale was set flat, the noise completely vanished. Smooth skies, smooth skin tones. Further investigation showed that the main culprit was boosting yellow and orange in the grayscale panel, which either introduces this noise or makes what CN is already there much more visible, I don't know which. And of course, the spectral response of a typical B/W film (like FP4) has this "kick" in the yellows, which gives it that look. And when I shoot B/W film, 90% of the time I have a yellow filter on as well, which in the English landscape lightens foliage and darkens skies. So all my LR presets tend to boost yellow and darken blue in varying degrees, which gives me a really nice film-like look, but also has this unforseen noise side-effect. So these days I tend to think rather more carefully about how much digital filtration to apply to a given image, and modify my pre-sets to suit.
I had just thought that all of this must be a well-known problem (except to me, of course), and I was trying to save others just starting out from finding out the hard way.