The much talked-about integration with OpenGL accelerated graphics is, to be honest, a bit of a let down. I had hoped that it would use the high-speed RAM and extra processing power to speed up rendering operations such as the Liquefy filter and large photo-merges, but in fact unless you’re using CS4 Extended and working with textured 3D models, the effects are largely cosmetic. They’re very pretty to be sure, but they don’t do a huge amount to improve the operation of the program for day-to-day photo editing.
The most frequent application of OpenGL integration is in zooming images on the screen. The zoom action is now animated using OpenGL, giving the impression of a smooth continuous zoom in and out, but only when using the Zoom tool. While this does look very pretty in product demonstrations, in practice it’s not a lot of actual use, because there is a slight lag as the zoom motion accelerates and then decelerates smoothly, making it impossible to zoom in to a precise magnification factor. In practice most people will continue to use the CTRL+ and CTRL- keyboard shortcuts to zoom images, as they always have. One useful addition though is the pixel grid which appears at the highest zoom settings. It helps to distinguish between adjacent pixels, handy if you have to make very fine adjustments to your pictures.
The other OpenGL function is a little more useful, but only under certain circumstances. If you are drawing or painting on paper on a desktop, it’s natural to rotate the paper as you work to give a more comfortable drawing angle. It is now possible to do the same thing in Photoshop. Simply press the R button, and then use the mouse to drag the canvas round to the desired angle. This rotation is entirely non-destructive, and has no effect on the final image. I have found this feature useful when using a pen tablet for photo editing and drawing, but to be honest you can get much the same effect by rotating the tablet.
There is another immensely clever graphical trick that is incorporated into Photoshop CS4, a feature called Content-Aware Scaling. I first saw this technology demonstrated in 2007 in a YouTube video by the two guys that invented it, which has since received over a million views. I remember thinking at the time “If Adobe sees this they’re going to hire those guys on the spot.” It seems I was right, because their invention is incorporated into CS4.
If you resize a picture in one direction using conventional methods, people and objects in the image will become distorted in proportion to the scale of the change in size. What Content-Aware Scaling does is to look for lines of pixels that don’t change much across the width of the frame, since these contain the smallest amount of useful image date. As the image is resized it then removes or adds pixels so as to minimise the change in the amount of information. The practical upshot of this is that you can shrink pictures by a surprisingly large amount in any one direction without causing figures or objects in the scene to become distorted. Of course the effectiveness of Content-Aware Scaling varies from one image to the next, but it is still a very impressive technology. See the sample pictures at the end of this review for an example of Content-Aware Scaling.