Hi - sorry I forgot about this "tutorial" thing. Here is a brief explanation. You obviously are able to make a mask that protects the subjects and isolates them from the wall with the shadows.
The next part is to deal with the shadows. As others have shown, you can clone them out completely if you choose, as long as the areas of the wall from which you sample are relatively evenly lit (no gradient or light falloff across the sampling areas) and the perspective is uniform so that the mortar joints in the sampled areas line up with the joints in the areas into which you would like to clone.
In your shot, you have the luxury of a uniformly lit surface and pretty uniform perspective, so you can go which ever way you choose. However, if you were unable to clone detail from one part of the wall into the shaded part, you could at least remove the color cast from the shadows and deal with the shadow tone separately.
In my edit, I removed the color from the shadow using Lab color space and editing the a and b channels directly. The easiest way to do this is to create a copy of the Lab document background layer (CMD-J) and then work on the a and b channels of this layer - if you mess up, no biggie, throw that layer away and make another copy of the background and start again.
I do not know how familiar you are with how Lab (more properly known as "L*a*b*) works and how it is represented in Photoshop, but the very brief (oh dear) explanation is this: in the Lab color model, Lightness (detail) is separated from color. The "L" channel is like a monochrome, grayscale representation of the scene and contains all of the detail and tonal information, while the a and b channels contain all of the color information. The a and b channels contain color opposites in their channels - the a channel contains the magenta-green information and the b channel contains the yellow-blue information, much like the white balance controls in some raw processors. The amount of a particular color in the a or b channels is indicated by how far away from neutral (no color) a particular value is in that channel. The two channels are effectively combined to make the final color and then superimposed on the L channel to create the actual color image. This is a very loosey goosey explanation, but it gives you the idea of how to understand Photoshop's representation of Lab.
In Photoshop, the a and b channels are grayscale images with middle gray (pixel gray value of 128) representing NO COLOR in the a or b channel, located midway between black (0 gray) and white (255 gray). The value of a color in the a or b channel of the Lab color model can range from -128 to 127, with 0 representing neutral. This can be confusing because the a or b value (-128 to 127) does not correspond to the grayscale value in the image of the a or b channels (0 to 255). Just remember to take the grayscale pixel value of the image of the a or b channel in PS and subtract 128 to get the actual a or b value in the Lab color model. As the a or b values get further away (either negatively or positively further) from 0 (neutral) they become more colorful. In the a channel, -128 corresponds to really greenish and 127 corresponds to really magentaish - likewise, in the b channel, -128 represents really bluish and 127 corresponds to really yellowish. Lab can also be confusing because, although it is meant to represent all of the visible colors (in terms of the human visual system) it can produce imaginary colors (what is dark yellow?) and it can also produce real colors that display devices cannot display. A relatively small gamut color space (like sRGB) fits neatly inside Lab (as do all other RGB color models) with the most saturated colors in sRGB located fairly close to the origin of the a and b axes on a 2D plot of the Lab color model. Attached is a 2D plot of a and b axes in Lab with primary and secondary sRGB colors located in Lab for reference (at an L value of 50 - the L value in the Lab color model can range from 0 (black) to 100 (white) corresponding to 0 to 255 in the grayscale pixel values in the L channel in PS).
Anyway, the bottom line is, you can manipulate color in Lab independent of tone or detail, making things like neutralizing (or enhancing) color very easy compared to doing the same thing in an RGB color model where color and lightness are inextricably coupled.
In your shot, convert a copy of your image to Lab. Make a copy of the background layer and then take a look at the channels in the Channels panel. Because most colors live close to the origin of the a and b axes, most a and b channels are fairly low in contrast. However, strong colors against a neutral or opposite color will have pretty good contrast in the a and/or b channels and make editing and masking easier. In your case, you already have a mask to protect the subjects, so this will be about getting rid of the shadow color.
In the second attached figure I have depicted the basic strategy. The lit wall has a certain color (it is fairly neutral) that you want to introduce into the color cast in the shadow (the pukey yellow). As you might imagine, the b channel depicts a lot of contrast between the shadow and the lit wall because the yellow is fairly saturated and intense (and yellow, especially a saturated yellow, lives fairly far away from the neutral origin in the a b axes). As shown in the figure, you want to select the gray value in the b channel that represents the lit wall (without the color cast) using the dropper tool (with a large sampling radius) and then paint this gray value into the area that is affected by the color cast in the shadow. You do this operation directly on the b channel with the mask (selection) protecting the subjects active. Same for the a channel - select the gray value for the lit wall, activate the selection that protects the subjects and paint away.
Now when you view your composite image, you will see that the shadow still exists in lightness, but the yellow color cast is gone. By painting on the shadowed area in the a and b channels with the gray values that represent the a and b color of the lit wall, you have made the shadowed area of the wall the same COLOR as the lit wall, neutralizing the color cast.
As far as completely eliminating the shadow, you can do your cloning on the L channel of the working layer. I would suggest that for your cloned version that will end up with no shadow at all, you duplicate your cast-corrected layer and then clone on this new, duplicated layer. That way you will ultimately end up with your original, a cast-corrected (but still shadowed) version and then a version with no shadow at all. You can then blend these together as you see fit to get the look you want.
Sorry for all of the nerdage - it is a very informal description of a color model that occupies terabytes of data on the web. If you want to learn more about Lab and editing in Lab particularly, Google "Dan Margulis" or search for him on Amazon. He, and his books, are the de facto reference for this stuff.
please log in to view hosted photos in full size.
please log in to view hosted photos in full size.