A Detailed Worked Example

On this page I work through a complete optimisation example, applying a range of techniques to improve the basic image and arrive at a final composition. The order in which the techniques are applied in this example is a typical workflow for an image.

The Original Image

This is a size reduced copy the original image, from the camera. It was taken in a mixture of natural light with additional highlighting from the right of the mask to create the shadow silhouette using the halogen light that was in the room.

The use of the halogen light, without any white balance correction, creates a slight colour cast which I will have to correct later.



There are two rotations that need to be applied.

The first one is applied to turn the image to the correct portrait aspect ration. This is a 90o counter-clockwise rotation.

After this first rotation the prominant vertical in the image, created by the concave corner in the wall, is not true.

This should be corrected using a rotation by an arbitrary amount. In this case a counter-clockwise rotation of just under a degree was enough to bring the vertical line to true.



We know that rotating an image by an arbitrary amount will leave oddly angled edges to the picture which need to be cropped away. In this case, the image also has a lot of unwanted detail around the ceiling area that needs to be cropped away.

Looking at the proportions of the distance of the mask from the right edge of the image and the position of the shadow silhouette, I decided on a tight crop, removing about two thirds of the wall to the left of the corner as well as the distracting detail at the ceiling.

The resulting compostion is shown on the right.


The next area of the image that needs attention is the edge formed by the corner. In the image above, if you look closely at the corner edge within the shadow, you will see a number of blemishes. These were caused by a poorly fitting wall panel that had small gaps in the join. These dertact significantly from the appearance and it would be better if they could be removed.

The image on the right shows the result after using the healing brush to remove these imperfections.



The next stage is to look at the dynamic range of the image and apply the levels tool to make the most of it.

LevelsIn this case, as you can see in the histogram on the left, there was scope at both the bottom and top to extend the dynamic range.

The result of applying this tool to the image is shown on the right.

Colour Balance

The colour balance of the image also needs a little work. In particular the wall needs to be bluer to more closely match how it looks to the human eye. To alter the colour balance of the wall without affecting the tones of the mask I first used the magnetic lasoo to select only the mask, and then inverted the selection.

Colour BalanceI then used an adjustment layer to modify the colour balance of the selected region. You can see the settings on the left. I only modified the midtones, taking the balance slightly to wards cyan and also slightly towards blue.

The result is shown on the right.


Blurring & Sharpening

It was at this point that I realized that I had done nothing to hide the crack in the mask. Ideally, I should have done this at the same time as I used the healing brush to remove the imperfections in the corner. However, I simply used the same technique on the base layer of the image at this point and blended the crack away.

The final stage of optimising the image involves some blurring and sharpening. Firstly the background has a little too much texture and needs to be blurred. Because I did not want to blur the mask, I used the layer mask from the colour balance layer to create a selection. The Gaussian blur tool could then be applied without affecting the mask.

Using the inverse of the selection, I also applied a small amount of Smart Sharpening to the mask itself.

The result of this is shown on the right.

Exporting the Image

The final stage is to export the image in a format appropriate for its use. In this case, it is going to be used in a web page.

Resize DialogThe original image from the camera was 3648x2736 pixels. This is clearly much too large for putting into a web page. The actual image included in this page is 236x400 pixels. So, the final stage was to resize the image using Image Size dialog on the Image menu, as shown on the left. Notice how the resolution is set to 90 pixels/inch, which represents a good quality modern monitor and the pixel dimensions are set to the required size. When OK is clicked it will resize the image. At that stage it is a good idea to save it again with a different name so that the full size original is not lost!

JPEG Save DialogFinally, save the small version of the image as a JPEG. When you do this you will be given the chance to set the image quality of the resulting file, as shown in the dialog on the right. You should set a value that creates an image of sufficient quality for the page, but which also makes the file size small enough to download fairly quickly.

There is also a tool specifically for exporting images for the web. On the File>Save for Web... menu entry. This tool is discussed here.

Back back to the list of examples.