Vanishing Point

Vanishing point is an innovative feature which allows you to select, paint, clone, and transform in perspective. Cloning away an open window in the example image we used on the previous page will not work with the regular Clone Stamp tool because the perspective makes each row of windows smaller until it "vanishes" into a "point" at infinity. So you would have to copy the row below it, paste it into a new layer, rescale it, adjust perspective, adjust the effect of the different lighting, etc. The Filter -> Vanishing Point dialog box makes this much easier. It allows you to draw a mesh that matches the perspective so that the Clone Stamp tool takes into account the perspective differences between the source and the destination rows. Of course there is more to it than geometrical accuracy because lighting and reflections on buildings typically vary with each floor as the angle of light differs. So the Vanishing Point stamp tool inherited some characteristics of the Healing Brush which ensures the natural lighting and reflection gradients are preserved.


Window cloned away in perspective using the Vanishing Point Stamp tool with Healing "OFF"

Window cloned away in perspective using the Vanishing Point Stamp tool with Healing "ON"


Perhaps less impressive than Vanishing Point, Edit -> Transform -> Warp is more useful and I am glad I can finally remove it from my personal wish list. In earlier versions of Photoshop, the bounding box of the Edit -> Transform command was rectangular which posed very serious editing limitations. Warp allows for "non-linear" custom distortions and distortions like the ones you can apply to fonts.

This is for instance useful in manual stitching or correction of panoramas. In this example it allowed me to "bend" the highway of the second frame so it fits smoothly with the first frame. In earlier versions of Photoshop, this was only achievable via the Liquify filter. But it was more difficult because the Liquify filter is brush based and therefore more suitable for distorting people shots, and less suitable for creating non-linear distortions on rectangular areas.


High Dynamic Range 32bit Floating Point Images

-1.0 EV
0.0 EV
+1.0 EV
+2.0 EV
Converted HDR

Although this feature targets the Computer Graphics industry (gaming and animation), it is useful for digital photographers to capture scenes with a high dynamic range. Because I often shoot in tropical conditions (dark shadows, very bright highlights), I developed an early interest in high dynamic range images. Four years ago, in my Nikon Coolpix 5000 review on this site, I illustrated how to capture a large dynamic range by combining exposure bracketed images. HDR is a leap forward in this area. In this example, the intense tropical sunlight carved its way through the dark shadows of this gallery at Angkor Thom, Cambodia. The dynamic range of the scene was several stops larger than the dynamic range of the sensor. Due to the linear nature of a digital camera sensor, it is impossible to capture the shadow detail, without clipping the highlight detail. Alternatively, if you wanted to capture the highlight detail, you will not have enough detail in the shadows.

The new File -> Automate -> Merge to HDR command automatically combines the bracketed images with the following benefits:
- more extended dynamic range
- there are no halos around the edges
- reduced noise in the shadows
- any tonal curves that were applied, e.g. by the RAW converter, are eliminated. You can apply a tonal curve to linear data with a phenomenal dynamic range.
- the data are stored in 32 bit floating point format

These HDR images are linear and look very flat and dull and of course, no monitor or printer can display them. So why do we want HDR? First of all, the linearity is useful in computer graphics applications to create realistic blending of images from different sources based on exposure information of the actual scene (e.g. subjects against backgrounds) and apply realistic lighting and motion blur via software.
By applying a proper tonal curve, photographers can compress the dynamic range and tonal range so the image represents a large dynamic range that looks pleasing to the eye on a monitor or as a pring by allocating the detail where they want it, and without causing posterization by working in a high bit environment. This allows images with spectacular detail in both the shadows and highlights as shown in the above example.

Needless to say that HDR only works well in static scenes. Of course, with a camera like the Nikon D2X, which was used for the above shots, you can shoot in auto-bracketing mode at 7 frames per second, which means it only takes 0.57 seconds to cover all 4 shots. However, this may still be too long for fast moving subjects. There is an option to automatically compensate for any camera or tripod movement that occurred in between shots. Needless to say that this will not work if for instance leaves of trees moved in between shots because of the wind.

This article is written by Vincent Bockaert, author of
The 123 of digital imaging Interactive learning Suite featuring
Adobe Photoshop Elements 2 & 3 and Photoshop 7, CS & CS2
Click here to visit