[Tutorial] The making of an image

[Editor's note: Click on images for full-size versions.]

I have been asked to reveal my secrets on the processing of this image I recently posted to the gallery; since there was nothing too different from my normal workflow involved in its creation, I decided to use it as an example in a tutorial about it. This is meant to help beginners at post-processing; experienced users might find it boring, redundant, and even inefficient in some of its stages. I'm posting it because it worked for me in this particular image, and it can provide helpful ideas for those trying their hand at post-processing. It is freely given, and no offense is taken if it is freely ignored. This is a long post, so brace yourself!

The software I am using is Photoshop CS 2; you can do pretty much the same thing in older photoshop versions (except perhaps for 16-bit processing), PSP, GIMP and God knows how many other packages.

Here's the picture I posted:



KONICA MAXXUM 7D
15mm 1/25th F22 ISO100



So, how de we end up with an image like this? Well, this is not the view out of my window, so for starters you have to get your butt out there. I also didn't stumble on this location by chance; a bit of research was involved. Although I didn't know I would find this particular spot on the shores of Pyramid Lake in Jasper National Park, I knew that the best time of the day to photograph reflections was in the morning, when the sun hits the mountain ranges head on and the mid-morning breeze hasn't yet disturbed the still waters of the lake. Whenever photography is the prime motive of a visit to a particular place, I at least take the time to look at the map, identifyithe various attractions and landmarks, and figure out whether they will be best lit in the morning or afternoon.

The second element is the light; you only get this kind of light in the early morning or late afternoon. To obtain images like this, there is no substitute for getting up early, period. This particular picture was shot at about 8:30 am, and that was the second session of the day (the first one had been shooting the sunrise at Mount Edith Cavell at 5:30 am). Sleep in and you'll miss the light.

Once you are on location at the appropriate time, your next task is to look for a pleasing composition. There are no hard and fast rules about although there are some helpful guidelines. In this image, two things work particularly well. The first one is a judicious use of the rule of thirds:






Notice how I divided the field of view into thirds, both horizontally and vertically; most elements of interest are placed near the intersections of the blue guide lines, and all are far from the middle section of the picture. The boulders, the mountain ranges, the clouds and the rocks seen trhough crystalline water all add interest to the scene, and the eye keeps wandering from one to the other; place one of them smack in the middle and it will capture all the attention, detracting from the rest of the scene and making it boring.

The second compositional element present in this frame are triangles:






The eye naturally follows the lines along the sides of triangles, which also evoke a feeling of balance. If you want to add a sense of strength and stability to your images, try circles instead. If I had placed all lines (shore, mountains) parallel to each other, the scene again would have been boring.

Finally, at the 15mm focal length I used, every distant element becomes really small, so it is imperative to add an interesting foreground, in this case beautifully provided by the boulders.

Once you are happy with the compostion, it is time to shoot your frame. In landscape photography, you want to get as much of the scene in focus as possible. That means stopping down your lens as much as you can without encountering diffraction effects; in my Sigma zoom, that happens at f/22. At such small apertures, however, shutter speed drops big time, so a tripod is mandatory. Even if you get a shutter speed higher than that, you will still want to use a tripod to preserve fine detail. Your 20x30 enlargements will thank you!

Next, figure out the optimal focal distance, which in landscape photography is the hyperfocal distance of your lens. Focusing at hyperfocal distance ensures that everything behind the focal point remains in focus, and a good bit in front of it too. You'll need a hyperfocal distance chart for your lens; in this case, the hyperfocal distance was somewhere in the boulders. Notice how the mountains are still in focus, although I am not focusing at infinity.

Set your camera to RAW mode, then shoot. And, for heaven's sake, BRACKET YOUR EXPOSURES! I am not going to be back in this location anytime soon, it wouldn't have been nice to get home and find out a bunch of nicely overexposed pictures, would it? Memory is cheap, moments are priceless. BRACKET!

OK, so we are back home, pictures downloaded to the hard drive. Grab a cold one (thanks Cramski for the tip!) and start post-processing.

Since we shot the image in RAW, we will need to use a RAW converter. I will use Photoshop's Adobe Camera Raw in this example. Open the file and you will find a screen like the one below:






A few things to note here:

- I will be working in 16-bit mode (lower left circle). This will allow me to preserve most of the information captured by the sensor, and will minimize any post-processing artifacts.
- Most other values are set to what I found works best for my camera; experiment with yours. ACR has a few standard profiles for the most common cameras.
- I normally don't touch saturation at this stage. I found that the saturation command in Photoshop itself works much better than the one in ACR.

The two commands you will use more often are the temperature and exposure ones. Here I am deliberately underexposing the image by 85% of a stop; this is because I will use two different versions of it to expand dynamic range (more below). The other is the temperature command; I am shooting for a "warm" look on the stones here, and so I am increasing the temperature to a hefty 6000. This value is way too high for most images; you will want to stay at 5100 for sunny shots, and 5500 for ones in the shade.

I said I would develop two versions of the image, one underexposed and the other overexposed. This is done to capture maximum detail in both shadow and highlight areas. My technique here is one of many; some people may manage it out of a single development; others will use more than two developments. Others yet will shoot two different frames (but in landscape photography this is tricky, as the slightest movement of the tripod between frames will show).

Here's the second conversion:






Note how it is now overexposed by half a stop; since this version of the image highlights everything but the rocks; I am dropping the light temperature to a more normal 5700.

Once you have your two conversions open in Photoshop, select the entire dark image and paste it on top of the light image. This will create a second layer, called "Layer 1." Now we will try to combine the best elements of both layers using layer masks; the goal will be to have the darkest parts of the overexposed image and the lightest parts of the underexposed image surface in the final product.

Select this layer and add a layer mask by hitting the "add mask" button (shown with a red circle). The layer mask becomes linked to Layer 1.






The next step is to select the bacground layer, select all its contents, copy them, and paste them onto the layer mask. To do this, control-clik on the white square representing the layer mask (Macintosh: option-click). You should see this black and white image representing what you have pasted into the layer mask.






The layer mask contains the luminosity information of the light frame; the dark areas of the light frame are the ones we want to preserve from the overexposed image; by using them as a mask on the dark frame, we filter them out of the underexposed image.

If we were to keep the mask as is, the image would look unnatural. To smooth things out, we will blur the layer mask. Select Image -> Adjustments -> Gaussian blur. For a 5-6 MP image, a value of 40 should work fine.






Next, click on the background layer, and you should see the blended image. It should already look better than any of the original frames, but we still have some ground to cover. The first thing is to get the brightness and contrast just right. To do this, forget that mistake of programming that were the brightness and contrast sliders; we will use the curves tool instead, which gives you much more control with minimal loss of information.

One of the advantages of this digital blending techniques is that you can use separate curves on the light and dark frames. Select the background layer and go to Image -> Adjustments -> Curves.






Here I am applying a classic "S" curve, darkening the dark areas and lightening the light ones. This adds contrast to the image while, if used judiciously, preserving the highlghts (not that we care much, the highlights are coming from the dark frame anyways).

Then select the dark layer (Layer 1), and use the curves command again.






Here the curve is a little different. Since this layer contains the highlights information, I am not touching the upper part of the curve. I am still darkening a little the dark areas, so that they will blend a little better with the background layer. Play it by ear (eye?) and see those tones come alive!

It is time to flatten your image now (Layer -> Flatten Image). From now on, we will work on a single layer.

There are a number of ways to get more vivid colors, all the way down to sophisticated and expensive actions. We'll take the easy road here: Image - Adjustments -> Hue/Saturation. Set the saturation slider to +10.






Be careful not to overdo saturation, or your images will look unnatural.

Next for color correction. Those rocks look awfully yellow to me, and I want a reddish effect. Go to Image -> Adjustments -> Color balance.






You can work in three different luminosity areas; the midtones are the "general" playground, and will suffice for the correction I want. I am increasing the Red and Blue sliders (hence subtracting Cyan and Yellow), which has the effect of turning the image a bit redder without warming it up too much (because I am adding blue).

Voila! The only thing left to do is to sharpen your image. What kind of sharpening you do depends on the intended use of it. Suppose you want to print the image; in that case, I suggest you set the viewing size to 50% of actual size; this will give you a good approximation to the sharpness of the printed image. Choose Filters -> Unsharp Mask.






I tend to set the radius to 0.5, using higher values only in extreme cases. 100% is a good bet for a 6 MP image, give or take. Look at the rocks for reference, and take care not to oversharpen.

Before saving, do not forget to convert your image to 8 bits, otherwise you won't be able to save a JPEG file.






If, instead, you want to post the image on the web, go to Image -> Image size. Make sure the "resample image" and "constraint proportions" boxes are checked, and set the image size to your liking. Here I am going for 800 pixels wide. Don't pay attention to the DPI box.






Now you have to sharpen your web image. Since you are dealing with a smaller file, you need different values for the unsharp mask filter. I found that 50, 0.5, 0 work best for my web images. Feel free to try your own settings and let me know what you come up with!






And that's it! Thanks for bearing with me, and I hope this has been useful. Above all, remember that the most important thing in all this process is to have fun creating your images. I enjoy every step, from the planning to posting them online. I hope you do too.

---

From: /forum/remark,14125972