Note that this exercise is based on Photoshop CS4.
Using the Canon 50d, you will find that as soon as your exposure is even slightly off, noise will be your unwanted companion. This post will take you through the steps I took to process an image from A to Z, and in particular, showing you a method to deal with noise in PS. There are of course more options do deal with noise, but this method is available without the need to purchase noise plugins and is available in standard Photoshop.
Here is my candidate. In the empty space of my sun blockers in my living room, a pair of Black Redstarts have taken residence and are raising a nest of 3 young. So for the remainder of this brood, we cannot use the blind, as it will block their entry to the nest.
| Black Redstart |
|
Image details:
|
And this is the setup. The nestsite is in the sunblinds in the far right corner and my perches are mounted on the wire fence running along my garden.
| The Venue |
|
| A lot of concrete and steel to deal with… |
The first steps I take in Photoshop is run the image through Adobe Camera Raw. There are 4 sliders I spend most time on:
- the tint and colour temperature slider
- the exposure slider
- the recovery slider
- the blacks slider
| ACR |
|
| Going through the first steps involves looking at the overall exposure, the white and black details and using the sliders as mentioned above to deal with the RED’s (overexposed) and BLUE’s (underexposed). The recovery slider should be used to take the overexposed areas away (pushing the slider to the right) and the blacks slider deals with the too dark areas, by pushing the slider to the left, black details might be reintroduced. If one of the blacks or whites cannot be dealt with using one of these sliders, the exposure slider should be used to adjust the overall exposure of the image. Have a closer look at your over- or underexposed parts of the image: often you find that there are hotspots in the image (e.g. waterdrop reflections) that can be dealt with in Photoshop, by simply cloning them away. This means that you do not need to frantically try to get all pixels in good order, where some of it can be arranged in PS. The tint and colour temperature slider are their to assist in finding the correct colour tones. Often the tint is too strong, introducing magenta casts, so just use the slider, going left and right, to find the colour tone you find most representative to your image. Also with the colour temperature slider, sometimes you should choose a ‘cloudy’ colour temperature in a midday photograph, to get a little bit of warmth in your image. Use the slider to taste, until the colour tone represents the tone as you remembered it (or want it to be). |
As soon as the basic RAW image setting are completed, press the “Open Image” button, so the image can be loadded in Photoshop. Note that whenever you press “Open Image” or “Done”, the RAW settings are updated in the RAW image file, so the settings can at all times be ‘redone’.
| The first steps in PS |
|
| I use actions to take me through the first postprocessing steps. Adding my name as the author, adding the copyright text, creating a duplicate layer, and creating a threshold, levels, curves and selective colours layer. The latter I sort of need in most images more or less. So my action will include them all and perhaps I don’t need them, but at least they are created using one mouseclick. The threshold layer renders the image black/white. Good practice here is to save your image. “Goto File->Save As …”, select the target directory, where you want to save this workfile and select TIFF or PSD as the image format. I have always saved my images as TIFF, as this is an industry standard, PSD is Photoshop specific. |
The threshold layer I use to find the right white and black balance. Select the threshold layer and move the slider all the way to the right: doing so, you will get to the brightest pixels in your image. Slide as much as you can until you see the last pixels fade. Select the “colour sampler tool”: it is in the same group as the “eye dropper tool”, the 5th tool from the top in the tools panel in Photoshop. Place a colour sample mark on the brightest area and use the zoom tool (fast select “z”) to get a detailed view. Once the colour sample has been placed, ctrl-0 or cmd-0 to full viewing the image, select the threshold layer again and move the threshold slider all the way to the left, until the last dark pixels fade. Zoom in to get a detailed view and place a colour sample in the middle of the darkest area. Now that you have found the darkest and brightest pixels of your image, we do not need the threshold layer no more: we only needed the threshold layer to find the brightest and darkest areas of the image. Take this layer and move it to thrash. The image will get back to its original colours again.
Now that the threshold layer has gone, select the levels layer.
| Levels |
|
| There are 3 colour samplers on the left, a black, a midtone and a white sampler. If you haven’t set your default white and black values yet, double click on the black sampler. You will get the colour selector. In the colour selector you can now set the default black values, when using the colour sampler in levels |
|
| In the R, G, B you need to set the black values as 12,12,12, as this will be your darkest dark. Do the same for the white sampler. You will get the colour selector again, where yo will set 244,244,244 for your whitest whites. Each time when you press Ok and you changed the values, PS will ask you if these values are to be taken as the detaults: you say yes. |
Now you have set the black and white maximum values. In the levels layer you select the back colour sampler again and goto the just selected darkest area of your image. Select the colour sampler point you created in the threshold layer. This will set the black balance. Do the same with the white sampled point of the image. By using this method, you spread the colour intensity over the whole scale from 12-244 and it also deals with colour casts. For example, if your darkest pixel was RGB: 25-18-18, you could have had a red cast in the blacks: by colour sampling it towards RGB: 12-12-12, that red cast in the darks is gone. This, however, could potentially push the other colours up in relation to red, so additional post tweaking might be necessary: it is of course possible that the blacks as selected truely had more red. In such cases you can decide to set your colour sampler default valules to RGB: 16-12-12, so to leave a little red in the blacks when sampling.
Take a break and save your file. ctrl-s or cmd-s. You can set an autosave option in the Photoshop preferences, in Edit->Preferences.
Now that the basic exposure and colour settings have been done, we can now start working on the composition and image cleanup. Obvious area of concern is the wire fence and the heavy perch in the lower right corner. A second compositional issue is the pose of the bird. In this image the male stretches out to the nest, where a more pleasing pose would be a little bit more tilted in a clockwise direction. But before I do that, I make sure that the perch is extended and I can use as much material in the image as possible.
| The Cleanup |
|
| A 10 degree rotation and a crop, where the center of the image is the middle of the bird’s chest. |
Next, the noise. Because I used a high ISO number to be able to shoot at 1/80 @ f8, there is some noise in the background. One can use the noise programs or install plugins like Ninja, but I want to introduce you to an alternative method. Essential with all methods is a very precise selection of what you want to de-noise. All of the noise programs are more or less blurring techniques, that become visible at contrast edges, in this image, at the intersection of bird and background. So the first thing we do is select the bird and perch, with the intention to isolate these objects. I use the magic wand for this to start (shortcut “w”) and select as much with the wand as I can, edging as much of the bird in as possible, avoiding any background. Once I am happy with my selection, I open the image in Quick Mask mode (shortcut “q”). Everything you select in QuickMask mode is coloured red. First, let’s set the QuickMask Mode presets. Do this by double clicking on the QuickMask button in the far bottom of the tools palette. In these options you can set the QuickMask colour (default is red), if your coloured areas are masked or selected. I choose to work with 100% opacity, red colours and my coloured areas are what I want to select.
Select the brush (shortcut “b”) and make sure the brush’s hardness is at minimum. As you will see, everything you just selected by the magic wand is now coloured red. Also, all edges the magic wand selected are hard. Take the brush at your best selected size (you can vary the size of your brush with [ and ], to resp. decrease and increase the size of your brush) and carefully make the hard red edges soft. The intention of this action is to create a gradient area, making sure that when we are going to reduce noise, the application of the noise reduction action will be gradually kick-in and will not be visible. Anything in images done, using hard edges will be visible.
| The Soft Edges |
|
| Now leave the QuickMask mode. What you selected will be marked by the running ants. |
Save this selection as we will need it twice. Goto Select->Save Selection and give it a name, say “aaa”. Now throw this selection in a new layer, by pressing ctrl-j or cmd-j. Your object, here the perch and bird are in a protected layer. Now select the background layer (bottom) again and reselect your earlier QuickMask selection, by going to Select->Load Selection ans select the “aaa” selection. Invert this selection, by pressing ctrl-shift-i or cmd-shift-i: by doing this everything but the object is now in the selection. Also, throw this selection in a separate layer by pressing ctrl-j or cmd-j. Your image should now have at least 6 layers, the “selective colour”, “curves”, “levels”, the bird, the inverted selection and the background layer.
| Gaussian Noise Reduction |
|
| On the inverted selection layer we apply the gaussian blur |
Place your cursor in the inverted selection layer. Now we are finally dealing with the noise. Select Filter->Blur->Gaussian Blur. A popup window comes up with a slider. Use this slider to see the effects, it will blur the layer your are working in: more to the right, more blur and more to the left, less blur. For a successful noise reduction usually a 3-9 pixel blur will do the trick: take as little as possible: Gaussian blurs can leave visible blur lines, along the more contrasty parts of the blurred area.
Finally, a minor but often ignored detail. The eyes. We can pop the eyes with a few easy steps to make the eyes stand out more. In this example we have an additional issue, and that is the line of the horizon. Since we rotated the entire image by 10 degrees, we should actually rotate the eye back by the same amount. To do this, select the eye, using a QuickMask. Select the bird’s layer and select the eye and throw it in its own layer, using ctrl-j or cmd-j. We throw selection in their own layer all the time, so that 1) we are not working on the original (background) and 2) if we don’t like what we did, we can simply trash the layer and start new.With the eye now in its own layer, open the Transform tool, using ctrl-t or cmd-t. A square with selection boxes is now surrounding the eye: use one of the corner boxes to rotate the selected area and rotate it back about 10 degrees or so much that the horizon is horizontal and press enter, to leave the transform tool. Within the eye’s layer, select the darkest area, again using a QuickMask. Throw this selection in a new layer and call the Hue/Saturation tool, using ctrl-u or cmd-u. Desaturate the selection completely, so that all colour is out and then slide the lightness slider to the left, so far that the pupil looks nicely black. When ready press Ok and view to fitting the image on screen. See if the eye’s pupil is not too black. If that might be the case, use the opacity slider in the layers panel to reduce the effect of the layer.
Save your file, using ctrl-s or cmd-s, so your workfile is safe and you can retrieve it anytime for additional work or other corrections.
The steps following your final workfile are usually the steps to publish or print your image. In this case I will publish a reduced version for webdisplay, usually this means a maximum size of 800 pixels and a JPEG formatted file. It includes flattening the image (ctrl-shift-e or cmd-shift-e), converting the mode to 8bits, fitting the image in 800×800 pixels (you find this in File->Automate->Fit Image). At this point in the process, there is one more thing we can do and that is apply a round of sharpening. In this image I used an unsharp mask of 200%/0.3pixels/1threshold. As a final step of tweaking, I flipped this image horizontally and saved the file to a <200KB file size, saving it as a JPEG.
| The Final Result |
|
