A little later than I'd hoped but here's part one of my promised how-to on how I achieved the "edgy" look I was so pleased with in my Montreal shots. I'm not claiming to have hit on anything new here, but it was something new for me and it really helped achieve my vision for these shots and I felt like I hit on some good workflow items as well in trying to make this process non-destructive. This becomes important because it lets you come back to the image and make adjustments over and over again until you get things just right. In this first image that wasn't as important but in other images I made use of this in order to tweak the look of the effect after I saw the initial result. I hope to cover more of this in a follow-up to this initial post..
I shoot in Raw and it was important that I be able to adjust the Raw development settings after I saw the effect. This meant using smart objects was the way to go. In Lightroom, you can edit an image in Photoshop as a smart object so that's the root that I chose to go. Not only does this allow me to adjust the development settings later (using Adobe Camera Raw in Photoshop) but it also allows me to use smart filters, which as we'll see are another important part of making this a truly non-destructive workflow. If you're not working with a Raw image and will be applying the effect with an adjustment layer (see below) then there's no need to conver the layer to a smart object.
The key to this effect is treating the colour of the image separately from the light and dark values of the image (or the luminance). This involves creating a black and white version of the image which will replace the luminance values of the colour image. I think I originally heard of this technique on one of the photography podcasts I listen to, unfortunately I can't remember which one it was as I heard it some time ago but only thought to try it out when I was working on these images. Since you're treating the colour separately, you don't really need to worry about the overall contrast of the colour image, that will come from the black and white version of the image. That said, I think it's best to work with a colour image that doesn't have a lot of contrast so that you can better see the colour you're working with.
In this case, I was trying out a 15 day trial version of Nik Software's Silver Efex Pro 2 Photoshop plug-in so I used that to create the black and white image. Fortunately, Silver Efex will work as a smart filter. This is one reason why using the Photoshop plug-in is preferable to using the Lightroom plug-in. Previously I'd found that I wasn't very happy with my black and white conversions and I'd noticed a lot of other people seemed to be using Silver Efex so I decided to give it a try and I was extremely pleased with the results. I want to point out that I have no relationship with Nik Software and they have in no way compensated me for saying this. I was so pleased with the trial that I have purchased the Complete Collection.
There's no need to use Silver Efex, if you already have another method of producing black and white images that you're happy with then the effect will still work. In this case I'm using a filter which is why I wanted to use smart filters. If you're using layers or adjustment layers instead then you may not have to use a smart object, although I still recommend doing so if working with a Raw image (see above).
Now you simply need to merge the light and dark values (or luminance) of the black and white image with the colour values from the colour image. This may sound complicated but it's actually just a matter of changing the blending mode to luminance. If we were working with a normal layer or adjustment layer this would just require changing the blending mode next to the opacity slider on the layers palette. Unfortunately, for smart filters it's not nearly as obvious that this can be done but there is a little icon next to the filter that will allow you to adjust this (see the screenshot).
If you are going to use a filter to create the black and white image then I highly recommend using smart filters so that you can adjust that black and white image once you see the final effect. I went for a very contrasty black and white image in this series of images and I think that works for these particular images but it might not in all cases. One thing to remember is that you're taking the luminance values from the black and white image, this means that very dark values in the black and white will tend to show very little colour (they'll be so close to black) and lighter parts of the black and white image will tend to make the colours look more washed out. Similarly, a very light colour in the colour image might start to look a little garish if it's mixed with the wrong luminance value. All of this can be fixed, either on a global level or more locally but that's why it's so important to keep as much flexibility as possible in the workflow.
In part 2 of this series I'll go into this tweaking in more detail and show how I used the flexibility to adjust the image after I saw the initial effect.