I am back from Indiana where I gave a seminar on Photography, Lightroom and Photoshop. Many thanks to a great group of people at the Goshen Photography Guild for their support for the program which sold out the available seats. The venue was great and the experience was wonderful and fun.
The day after the seminar a group of us went after images at Michigan City along the lake with the Chicago skyline visible (slightly) in the distance. Some were hoping for a sunset with the skyline, but cool air and wind were there instead.
As is typical when giving a seminar, you never know how the questions and overall flow will go until it happens. I wanted to include a section on image sharpening which time did not allow. So, I want to direct people to my Learning Page for more information and in particular the two articles, Smart Sharpen , and the High Pass Filter for sharpening images. There are, of course, many other ways to sharpen images, but these two are both powerful, versatile, and the most used.
The two gulls in the photograph above were surprisingly cooperative considering the exposure time for the image was 20 seconds. This was done with the aid of a neutral density filter that allows the moving water and clouds to appear as smooth tones rather than fine detail, while stationary content is registered in a normal manner. Great fun.
On my Learning Page (see link in the main menu) I have posted a new article, the second in a series on blend modes. This time we explore two members of the Darken group. These apply to both Photoshop and Elements users. They are simple to use, and valuable additions to your tools for controlling your images. Please read the article, which is short and to the point, and try the techniques yourself. I will be running seminars very soon on advanced techniques for processing fine art images in Photoshop and Elements. The blend mode article will be good foundation material for getting you up to speed before attending one of these seminars, especially if you are an Elements user.
I recently spoke to the Hershey Camera Club on the subject of lighting and on-camera flash. This is a subject that can get pretty involved and I was hopefully able to impart a sense of the scope of the subject to those who attended. Feedback on the meeting was good. I would like to run a hands-on shootout on flash photography at some point in the near future. If you think you might be interested in such a seminar, please send me an email using the link on the Learning Page. No obligations at this point, but I would be interested to know how many people may find this interesting. It may be that there will be two events, one for beginners who do not have hand held incident light meters and want to learn to use the flash actually on the camera, and a second more advanced seminar for meter owners including off camera flash. The techniques for marrying flash with ambient light can be confusing, but I can teach you the concepts of the technique pretty quickly if you are interested in more control over your lighting.
Local area events coming up very soon include the Harrisburg Camera Club Spring Workshop April 13, and the Camp Hill Plein Air event May 18. More detailed information on both of these events are in the newsletter on the Learning Page of my website.
A common failing in many images is non-existent or badly done sharpening. I hope if you are working in digital imaging that you already know to avoid the three filters that actually start with the word “sharpen” in the filters list. Smart Sharpen and Unsharp Mask are much better choices. In addition is a technique not shown in the sharpening options, but using a filter that “passes” high frequency information. In real world terms that means “edges” to a photographer, and is the opposite of a “low pass” filter, which is Gaussian blur. This is a fairly simple and easy to use sharpening technique, especially for portraits.
The technique requires a blend mode be applied to the layer as well as the high pass filter itself. This is best done on a merged visible layer at the top of the layer stack saved as a psd. That makes it easy to remove and replace should the image need to be repurposed, resized, or otherwise manipulated. In a finished image it can be applied to a duplicate layer. The primary decision for all sharpening is output destination and size. The sharpening for a small web image will be insufficient for a high resolution file aimed at producing an 11×14 inch inkjet print. Conversely, the print sharpening will be far too aggressive for the web image.
The High Pass technique essentially finds edges and when coupled with a contrast blend mode like Overlay results in increased contrast at the edges giving the appearance of sharpening of the image. Smoother areas of the image are ignored which helps to make the process even more valuable. For example, High Pass sharpening of about 1 pixel on a portrait will sharpen the eyes and hair while ignoring the skin.
The effect can be reduced by using the Soft Light blend mode and/or reducing the opacity of the sharpening layer. Small images can use High Pass settings as low as .3 pixels, and some larger images can benefit from larger radius settings. Experimenting and analyzing the results is something the beginner needs to do to gain an understanding of how the process works.
Find the dialog under Filter | Other | High Pass and choose a setting appropriate for the image size. When you click OK the layer will turn gray with slight lighter and darker edges defining your subject edges. Change the layer blend mode to Overlay to start, and click the layer on and off to see the effect. This is best done with the image on screen at 50% or 100%. A 50% view more closely shows the result as it will appear on a print. Edges you would prefer remain unsharpened can be masked off if needed.
A full article explaining the technique with illustrations is on my Learning Page (see link at top). Check in there (refresh your page if needed to get the latest version).
You sometimes need a current representation of your file to make selections. I work a lot with Color Range for selections, and it is convenient to make a merged visible layer to use as source material for this. If you are not familiar with this technique, simply make the top layer of your psd the active layer and then use “merge visible” in the layers drop down adding the (<alt>) [<option>] key to prevent the layers from flattening. The keyboard shortcut is (<ctrl> <alt> <shift> E) [<cmd> <option> <shift> E]. This creates a new layer at the top incorporating all the layers below into a single layer, much the same as flattening the file but retaining the file layers below it.
I use this merged visible layer as the source for making the selections. If there are several adjustment layers at the top of your stack you will discover that selecting with a non pixel layer as the source will result in no pixels being selected. Making the merged layer and using it for the source will solve the problem.
Once your selection is made you can add an adjustment layer, or jump the selected pixels to a new layer for more work. With this done I drag the merged layer to the trash. This keeps you from expanding the file size for one thing, but also eliminates the loss of the ability to modify the layers under the merged layer. Keeping your pixel layers at the bottom helps you maintain flexibility.
Another good use for the merged visible layer is sharpening. I use a merged visible layer at the top of the stack for sharpening at the end of the process. This layer can easily be removed if necessary to make changes in the file, or modify the sharpening technique. Sharpening can also be adjusted based on output and sizing parameters. Adding a mask to the sharpening layer allows local sharpening, and adding layer style modifications using BlendIf can protect highlight and shadow areas from clipping with some sharpening methods.
When you modify the combined RGB values in an image with Levels or Curves your changes are not strictly brightness changes, even if that is your intention. They also modify the relationship between the color channels which result in shifts in color saturation and occasionally in the hue as well. This is easily brought under control with a simple change in the blend mode.
I burn corners with a curves layer, dragging the top or bottom end points depending on how I want to effect the image. Drag the bottom end point to the right and the image gets more contrasty as well as darkening the bottom end. Dragging the top point down darkens the image highlights and reduces overall image contrast. So, a combination approach usually works best, and mid-points can be modified as needed. If you darken a blue sky or other area with a strong color you will not only get a darker image, but a shift in saturation. Get severe with the changes and you can get a slight hue shift as well.
To remedy the situation change your blend mode to Luminosity. This restricts the effect of the changes to the brightness values in the image and eliminates the saturation and hue shifts. In most cases the effect is not radical, and if you like what happens, just leave the blend mode in Normal. I always check the effect to see which blend mode gives me the look I want.
The same control can be used anytime you modify something to lighten or darken with an adjustment layer. Quickly check the blend mode to see if a color change has occurred that may not be desirable.
There are other useful blend modes, but this one is especially useful and an easy step toward learning to use them to your advantage.
Do you find that having set black and white neutral points in an image, especially with Levels, modifying a mid-tone value with curves, such as a skin color in a portrait introduces a color cast? This is not surprising as the pivot points for the colors are at the far ends of the curves so there is nothing to prevent shifts. To prevent the shift you simply need to “lock down” certain values so that the pivot points are away from the ends and extreme values are protected.
To try this, open a curves dialog and place an additional point on the curve one block up and to the right of the bottom black point (assuming you are using the finer ten block detail in the curves dialog). Do the same one block down and to the left of the white point.
Those values will be at approximately 24 at the bottom and 230 at the top end. Additional points above and below these values can be added to lock in specific values if needed, and there is no rule as to where you lock down the values. Additional points help to minimize the shifts as the two initial points simply create new pivot points. Once your critical values are locked down, mid-range value changes will have a considerably lesser effect on the black and white neutrals.
Selection tools, especially Color Range and others with a tolerance setting will make pretty specific hard selections. I avoid feathering as it is not interactive or easily modified after the fact. I prefer to make a layer mask based on the selection and then blur the mask. Blurring the mask is interactive as you can see the effect the blur is imparting on the selection in the Gaussian Blur dialog, and other modifications can be made to the mask with Levels and painting.
Layer masks made with the Luminosity of a pixel layer or from a channel should also be blurred in most cases if you are making an adjustment or using a blend mode to alter the appearance of the image. An unblurred mask will reveal its existence somewhere in the image either as an edge or as a strange-looking veil over the image content. Blurring the mask eliminates this problem by blending the modifications into the image in a more subtle fashion.
The trick is in how much to blur, and that is a question without an answer. A hard edge along a building roof can be blurred as little as three-tenths of a pixel, essentially faking an anti-aliased edge on a mask where a sky has been changed, added or modified. In other cases, blending in content or a change of density can require a stronger blur of 2-5 pixels. In some cases rather more severe blurring is called for in an attempt to hide the effects of the mask. The answer is to boldly move the slider in the dialog and use the Preview check box to see what happens. You will find that there are few rules for the technique, but closing in on the proper setting is not really that difficult.
When I paint on a mask, even with a fairly soft brush, I generally check the results and add an additional blur at the end to see if the transitions become more believable and subtle. Not leaving evidence of your modifications is the goal.