A Few Words About This Page
This tutorial page is not structured around explaining what every tool and every function in Photoshop does. Such information can easily be gathered by simply pressing ‘F1’ and reading. That will teach you where, for instance, the Clone Stamp Tool is and roughly what its function is, so that won’t be explained here. There will at times obviously be procedures explained in here that could be gathered from Photoshop’s own documentation, but not at the cost of making the tutorial too convoluted to easily follow.
The premise and structure of this tutorial is the presentation of a specific and/or common type of problem and then working through a range of different solutions. Learning how to use a Photoshop tool is easy, but learning what it could be used for is not quite so straightforward, hence this tutorial.
Cheers,
disgracian@deviantart
Table of Contents
1. Photo-restoration 101
Not every photo you take comes out the way you want it. With cheap point‘n’shoot digital cameras you often don’t have a say in it either. The photo below will require brightening, hue adjustment to fix the skin tones, and noise removal. ‘Noise’ is another word for grain, a common phenomenon with digital cameras and accentuated by shooting in low-light conditions.
Figure 1a: Indoors snap with a malfunctioning flash. Figure 1b: Levelling reveals high levels of ISO noise and unnaturally purple skin tones.
First, press Ctrl+L to bring up the Levels tool. By looking at the graph you will notice a large mass on the left and virtually nothing on the right. As you can see from (2), this graph measures the spread of pixels in the image from darkest to brightest. What this graph is showing is that the highest concentration of pixels are on the dark end of the scale. We already knew that from looking at the picture, but hopefully now we can interpret the graph’s meaning.
Figure 2: The Photoshop Level Tool.
Simply drag the three notches at the base of the graph to the positions shown in Figure 2 (or type out the Input Levels to the values shown) to lighten the image. The next step shall be to restore the skin tones to a more natural colour. This can be done in many different ways but for now we are going to use the Hue/Saturation Tool, the reason for which is because this tool is suitable for the majority of situations and thus should be your first port of call.
Figure 3: Hue shifts a specific range of colours through the colour spectrum,
while Saturation affects the intensity of the colour.
Figure 4: Before & after.
Change the Edit range from “Master” to, in this case, “Magentas” and use the circled eyedropper tool and select an area of skin for further fine-tuning. The reason we do this is because we only want to change the hue of the skin and nothing else. Then drag the three sliders according to whatever gives, in your view, the best results. Ensure that the Preview box is ticked otherwise you won’t see the results of your tweaking.
The last step is to remove the noise from the image. Which tools you use depends on, among other considerations, the severity of the noise. With extremely grainy images there will always be some amount of overal sharpness that is lost on the process. The quickest and least sophisticated method is Gaussian Blur, located in the “Filter” drop-down menu under ‘Blur’. This blur applies evenly over the entire image so that, while reducing noise it also dulls any and all sharp edges. Smart Blur (same location as Gaussian Blur) is one potential solution to this problem, in that it uses a threshold system to only blur areas with slight variance in adjacent pixels so that, in theory, hard lines are preserved while grainy textures are smoothed over.
Figure 5: Blur properties.
Figure 6: Blurring by hand.
The final method is to get out the Blur Tool (press ‘R’) and do it by hand. For this procedure it is advised to create a new blank layer and tick the Use All Layers box. Doing it this way preserves the original layer from damage. If the experiment turns out to be detrimental then the topmost layer can be discarded and the image returned to its original condition. The technique is to avoid all edges such as the hairline and any facial detail, instead covering the forehead, cheeks, neck, etc.
Figure 7: The finished product.
2. Selective Exposure Enhancement
There are times when one needs to alter image properties such as levels, hue, saturation and soforth to either correct the exposure or aim for an artistic effect. Some shots may contain diversely lit areas and in correcting one area, you ruin another. The most common example is probably landscape shots where the sky is either too light or the ground is too dark.
Figure 8a: Ground is badly under-exposed. Figure8b: However re-levelling ruins all sky detail.
There are a number of methods to overcome this. Some of these methods can be used in conjunction with each other; it all depends upon the individual circumstances. The first method involves Adjustment Layers and Layer Masks:
Method 1
Figure 9: Layer Window.
Looking at Figure 9, the fourth icon from the left is the Adjustment Layer icon, which will give you a pop‑up list from which you should select Levels. Paying no heed to the whitewashing of the sky, adjust the levels until the areas of land have been appropriately lightened.
Adjustment Layers are preferrable in many cases to directly applying effects to a layer because they make no permanent change. The effect of an adjustment layer is carried out to all the layers below.
You will notice in the adjustment layer a white panel to the right. This is a Layer Mask, which is a greyscale image that determines which areas are affected by the adjustments. Black areas are excluded from the effects of the adjustment layer while white areas denote maximum effect. Likewise in areas of 50% grey, the adjustment layer will only half affect the layers beneath.
Click on this white panel to select the layer mask, choose a large black brush and begin painting over the sky with it. Start off by using a large-sized brush to quickly remove the bulk of the sky, then graduate to smaller more precise brushes as you get closer to the horizon line.
Figure 10a: This demonstrates using a large brush to remove most of the sky. Figure 10b: Looks good doesn’t it? You’re right, it doesn’t. Don’t worry though, we’re not finished yet.
Figure 11: Brush properties.
Next you will want to use a soft-edged brush and some greatly lowered Opacity/Flow values. What we will be doing is fixing up the distant mountains to the right, which are currently too bright and blue, by softly and only partially reducing adjustment layer effect so more of the original darker mountains show through and ultimately blend seamlessly between sky, distant ground and close‑up ground. Remember to switch to a white brush to apply more of the adjustment layer if you accidentally make part of the image too dark. The end result should look something like this:
Figure 12: Behold the mighty misty mountains!
Method 2
Figure 13: Quick Mask.
This is a quick‘n’nasty method that may prove suitable for some situations, though it probably will not produce results comparable with Method 1. But, as implied, it does have the virtue of being very quick and painless to implement.
- Press Ctrl+A to select the entire canvas and then Ctrl+C to copy it to the clipboard.
- Switch over to Quick Mask Mode either by clicking on the icon (which is shown in the figure to the right, highlighted by a red circle) or pressing ‘Q’.
- Press Ctrl+V to paste the image into the mask layer. Note: You can skip this explanation on the Quick Mask if you want, and read on to Step 4 instead. The Quick Mask mode uses greyscale painting with standard tools like brushes, pencils, gradient fills, etc. to select pixels gradually (represented by varying shades of grey). In the Quick Mask mode, black denotes areas that are not selected is represented by a red hue.
- Return from Quick Mask either by pressing ‘Q’ again or clicking on the icon next to the aforementioned highlighted icon. Then press Ctrl+Shift+I to inverse the selection. Then, for the sake of convenience (though it is not strictly necessary), press Ctrl+H to hide the selection outlines if you find them distracting like I do.
- Begin applying Level effects until the desired result is achieved.
- Press Ctrl+D to deselect everything. Be Aware: If you made use of the hiding function at the end of Step 4, it is easy to forget that you have an active selection area.
You will notice that the effects of levelling are only being applied to the darkest areas of the image, thus preventing the sky and the clouds from becoming overexposed. So what just happened? First we pasted the image into the Quick Mask to give the mask selection values according to the brightness of each pixel in the image. Then we inversed the selection so the darkest pixels were those most strongly selected because these are the ones we need to brighten. Accordingly, the brightest existing pixels (the sky) were not selected and the mid-range pixels (the distant mountains) were roughly half-selected. When the level effects were applied, they only affected the darkest areas and not the lightest. The result (for me anyway) was this:
Figure 14: Inferior colour depth compared to Method 1.
3. Fixing Artificial Sharpening ‘Halos’
Figure 15: The familiar ugliness of sharpening halos.
We’ve all seen them. Whether applied by the digital camera itself or through overzealous use of the “Sharpen” filter in Photoshop, they’re ugly and they cheapen the entire photo. There is, however, a remarkably easy and quick fix for what ails you. For this example, I have exaggerated the halo effect around the tree, though it is not uncommon to see natural example at least this bad or even worse.
The solution is no more complicated than creating a new layer and adjusting the Blend Mode for the new layer from “Normal” to “Darken”. The way Darken works is to only affect underlying pixels if they are lighter than the brush you are using.
Use the Eyedropper tool and select a nearby water pixel as shown below. Select a mid-to-small sized brush and begin painting over the white halos. You don’t need to worry about not painting over the tree since the properties of the Darken layer prevent this from being an issue.
Figure 16a: Select the appropriate colour. Figure 16b: For best effects, only the outside edge of the brush should touch the halo; the rest of it should paint over the tree sections (i.e., areas that will be unaffected).
There are many occasions where applying a Sharpen filter is desired. The most common is to add clarity to a shot that may be slightly out of focus in some areas. Doing this will, however, add those halos to areas that are already sharp. But more often than not, these halos are the result of crude sharpening filters used in many cheap digital cameras. Using this method we can now turn the original example into this:
Figure 17: Ironically, discarding the halo creates a heavenly result!
4. The Wide World of Panoramas
When one, or two, or even three or more shots aren’t enough to capture what you want, it’s time to start stitching. The first rule is one that must be obeyed before one comes within 374‑yards of Photoshop, and that is to create significant overlap between your photos. This is essential. The images I shall be using for this example are as follows:
Figure 18: The exposure gets progressively worse from right to left, a feature of my old Canon A20. This can create a serious challenge for a seamless shift.
In any operation of this magnitude, a strategy must be devised before the physical work begins. Firstly, you should decide where to start. I would advise starting with the most important feature of the panorama. In most cases, this will probably be the centre of the image. For this example, I have chosen the rays of light on the right edge. Also in this example, we have the issue of varying exposure to reconcile. I have decided to opt for a consistent light-level throughout the entire panorama. With my starting point and my objectives clear, let us begin.
Start by loading the initial image in Photoshop. Double‑click on the Background layer and select “OK”. It now becomes “Layer 0” (by default) and is independent from the canvas. This is necessary because our next step is to enlarge the canvas to accomodate the remaining sections of the panorama. Do this by selecting Canvas Size from the “Image” menu and entering the following settings.
Figure 19: 500% width is more than enough given the amount of overlap between the 7 pictures used. Remember to move the anchor point so all the new space is generated to the left.
Load the next image in the sequence into this newly enlarged canvas and position it so that it overlaps the section in the layer beneath it. It can help to reduce the “Layer Opacity” so you can see underneath the section you are positioning until they align as closely as possible.
Figure 20: Making the new layer semi-transparent helps you to align the two images.
Return the new layer to 100% Opacity and repeat the procedure for the remaining images. There may be occasions where additional manipulation may be required to make the images fit correctly. The Rotate and Distort tools located in the “Edit” menu under ‘Transform’ can generally accomplish whatever is required. In this case, I found it best to begin by aligning the treeline, and then stretching/distorting everything else to fit.
Figure 21: The next image required some distortion in order to fit. While it is not necessary for 100% accuracy, the more closely they fit now, the easier your work will be later on.
The nature of panoramas means that the required levels of distortion will increase exponentially, so don’t be too surprised or disheartened if it ends up looking lopsided. A simple solution to restore proportions is the Perspective tool, found in the same location as Distort.
Figure 22a: At this stage you might be thinking you made a wrong turn somewhere. Figure 22b: All you need however is a sense of perspective to restore balance.
The next task is to create a consistency of colour and brightness. As it stands, each image is slightly off in colour balance and horribly off in light levels. The easiest way to do this is step-by-step: modify the first segment of the panorama to your liking, and then alter each subsequent segment to match the first. Because of the growing lighting disparity between the sky and the ground, you will need to work on them separately.
What is required is to duplicate each segment of the panorama and use the techniques described in Tutorial 2 to erase everything above the horizon line in the newly created layers. See Figure 23: It shows the first 3 segments of the panorama with the sky erased. Once this is completed you will need to adjust each subsequent segment to match the previous one in both colour and lightness. As with the initial placement of the segments, perfect matching is not essential (nor is it possible), but it does need to be fairly close.
Figure 23: Because fixing the landscape and fixing the clouds are two entirely separate processes, a duplicate was made of each layer and the clouds erased.
Another tool that you should become familiar with is Selective Colour. You may also need to use multiple tools if the dissonance is particularly great, such as the last frame of this example panorama which I shall use as a step-by-step example.
Figure 24a: And so it begins…
Figure 24b: Use Levels to brighten the image to near what is required.
Figure 24c: With some final retouching with the Selective Colour tool, the illusion is complete.
The first step is to bring up the Levels tool and drag both the midpoint and whitepoint sliders. You can’t use the Selective Colour tool because it will drain too much colour in the process of lightening the image. Levels, on the other hand, will preserve the colour and if anything lead to oversaturation.
The Levels tool is a general purpose tool, so you will usually use this one initially for the largest corrections.
The Selective Colour tool is for subsequent minor corrections, since it is more versatile. It works by isolating individual colour channels such as RGB or CMYK and altering the lightness and colour balance selectively for each channel. In many cases, you could use this tool interchangably with the Hue/Saturation tool explained in Tutorial 1. The Selective Colour tool is more advanced however. With the Hue/Saturation tool, you can only shift the target colour range along a linear spectrum. The Selective Colour tool gives you far greater control than this, allowing you to bleed or drain any specific colour from the any available channel (RGB or CMYK). For example, if the grass in one frame of the panorama looks too dried out you could infuse more green and perhaps remove some red. Or alternatively if it looks too lush you could add some yellow and red and bleed some green at the same time. For precise colour-matching, this is the tool to use.
I found at times that I needed to adjust the colours and lightnesses of the foreground grass and the hills in the background independently of one another. A simple hue adjustment would not have been able to do this.
For whatever reason, I found easiest to go back the other way with the clouds, starting left to right. Obviously you can ignore any dissonance below the horizon and just concentrate on smoothing over the seams between the sky. The most painless method is to use Layer Masks because it ensures that no change made is irreversible. If your changes go awry, you can start over again with one mouse-click. Bring the underlying layers out of hiding and behold our half-completed panorama. One of the first things you will notice is the same issue dealt with in Tutorial 2.
Figure 25: If all has gone well, this is where we all should be. Only the clouds to go.
Add a Layer Mask to the leftmost layer and, using a large soft brush and low Opacity and/or Flow values, gradually erase the right edge until the two layers blend seamlessly.
To be continued…
© Copyright, Mustafa Smith, 2005
All rights reserved. So there. 