Tone map

One common mistake assumed by many starting amateurs in photography, especially the more purist, is believing that all that matters is what the camera can capture, and all post-processing is evil. It is a point of view that all correction should be made in the camera and if they are correctly performed, the image that we will obtain is the best possible outcome of the process. I also went through this phase, a little bit after the “eccentric filter phase” and just before the “HDR hole phase”. It is true that good parameters make a picture good, and it’s difficult to obtain a good quality picture without the proper measure of focus, exposition and many other scene-dependent parameter, but that is not the whole of the story.

Something you learn after taking many many and many pictures is that what you see is just an approximation of what reality really is (not everybody see the same amount of green-blue tones, for example), and that what the camera capture is just an approximation of what the eye really sees. For example, the dynamic range (the difference in luminosity between what the eye or the sensor consider pure white and pure black) ranges from 5 to 8 f-stops for most compact cameras and can increase to around 11 f-stops for high end cameras, while the eye is estimated to be around 14 f-stops. Also an eye in good conditions can detect more color tones than a camera. All of this means that even with the best parameter, we will usually not achieve a picture that shows the reality as it truly was when we took it. Even for people defending a realistic point of view of photography, the use of some corrections is something necessary to increase the quality of the picture.

Tone map is one of the filters/processes that help to achieve more natural pictures. It derives from a technique called HDR (High Dynamic Range), which combines two or more pictures with different expositions to achieve a final picture that captures a bigger dynamic range than any of the original ones. In this case, tone map can simulate this effect to a point, simulating an extended dynamic range similar to the eye, but only using one picture. What the filter does is to average luminosity along all the histogram (i.e. avoiding vast zones with similar luminosity while some luminosities doesn’t appear on the picture), increasing the contrast in the darker and clearer areas (shadows and highlights) while reducing contrast in midtones areas. This translates into a more natural picture where everything seems to have a more uniform illumination.

This technique not always gives good results, as it is very dependent of the illumination conditions of the place and the quality of the picture. It usually is very useful in photography under natural light or when working in interiors with artificial light (not controlled by the photographer, like fluorescent lights or similar illuminations), although sometimes it can also increase quality in pictures taken with a frontal flash. In high contrast situation, when this difference on illumination is important, or in pictures where the blacks suppose an important factor for the character of the image, usually this filter doesn’t perform well, though sometimes used in a small amount can provide an interesting effect.


Original B
Original picture: The shadows obscure the image. The poor girl can’t be sure if it’s her mother or somebody else impersonating her.

To show how this filter is applied, we’ll start with the picture on the right. It is a nocturnal picture of a statue in my city, which represents a woman and her child, who is talking to her. As it can be seen, the picture has a wide tonal range, from almost black shadows in the sky to almost white highlights in the street lights. Also, the woman’s face is shaded and her eyes can be barely seen. In general, the picture has an illumination pattern good for an artistic picture at night, but it is not the best illumination for an informative picture (e.g. for an use in a tourism catalogue). In this case we don’t want to play with the shadows but to obtain a clear and representative picture of the statue. This supposes a perfect example for applying this process.

Layer order to begin.

Step 1: Duplicate the layer where the image is twice. If the image is just in one layer, a standard duplication is enough. If working with many layers inside a group, duplicate the whole group and merge it after. This will condense all group information into a new layer, while conserving the original group. After this duplicate the new layer again. Put both layers inside a new group and, if working with a group, put this new group inside the original one. The situation after the process should be similar to the one in the picture.

Step 2: Leave the lower image as it is. It will provide the base for all the modifications. Desaturate the upper layer, and invert it. The effect is to discard all color information (in on the lower layer) and retain only brightness information, which is inverted (midtones are mildly affected while highlights and shadows are exchanged).

Before the gaussian blur the effect is too strong. Promediating the brightness allows to obtain a more natural effect.

Step 3: Apply an opacity value of 75% to the upper layer, so the lower layer can provide the color information and contribute with a 25% of the original illumination. We will obtain an almost grey image that we will call the map. Now change the group blending mode to “Soft light”. Soft light will brighten the original image if the map is clearer than 50% grey, and will darken it if the modified is lower than 50% grey. At this point we should obtain a very “washed” picture similar to the one depicted on the right. This is considered a hard mapping, as every pixel is changed only considering its own information. In order to obtain a better effect, we have to average every map’s pixel with the surrounding ones, so every pixel will have some information about the brightness of the pixels around and produce a more natural and appealing effect.

Step 4: Select the upper layer and apply a Gaussian blur filter. We can adjust the radius of the effect, measured in pixels. The more blur we add, the more information we give to any pixel about the surroundings, and more dim the effect will be. Values close to 0 will yield the hard mapping effect, while too long values (255, maybe upper in some software) will yield an image similar to the original. In my experience, a value lower than 100 pixels is never satisfactory for a good quality picture, while a range between 100 and 200 pixels tends to provide good results. The way to decide is just by trial and error, but with a little bit of practice is really easy to determine the best amount of blur. The effect shall result appealing and natural to the eye.

Step 5: After this process, reduce opacity of the group to 90 %. Higher opacities should not be used, as sometimes can merge bad with the original image and produce some artifacts. 90% opacity offers almost the same effect and merges better with the picture. Also, after applying the filter, an increase in color saturation can be noticed. It is not always undesirable, but it depends on the situation. If it is not wanted, just add a saturation adjustment layer over the mapping group that reduces overall saturation by a 10%. This will make the colors similar to the original.

The picture at the end should look like the one in the comparison behind this paragraph, with a more natural and clear illumination. If in some pictures the effect seems good in “shape” but not in “amount”, you can reduce the opacity of the group in order to adjust the quantity of mapping (in this case I reduced to 66%). Also, sometimes, this effect doesn’t blend well with the picture, washing it too much or destroying critical contrast for the composition. In this cases don’t force it and just avoid using the filter. In my workflow I have an action with this filter recorded, and I apply to all pictures I take. Sometimes the effect is good, sometimes it requires some adjustments and sometimes is awful and I just remove it. For me, it just takes pushing one button and waiting a couple of seconds and, many times, it can increase the quality of the picture noticeably. In this case, some more adjustments could be done, as re-adjusting the black point so we recover a little bit of the contrast on the bodies, but this goes out of the scope of this example.

Now, the girl can recognize her mother and feels safe. Also will do the tourist who saw the guide when they pass around. Good job photographer!

And now is your turn… Have you ever used this filter? Do you know any variations of it? If so, comment and share your opinion, so this information can be more complete and useful for everybody.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s