Motion graphics often involves the use of photographs – which is a bit of a problem because photographs are rarely as engaging as moving images when you mix the two. In this post I’ll discuss 4 methods of making stills look 3D, and give a quick tutorial on my favourite one, which isn’t that well known.
For years, the usual technique for making stills more interesting was the so-called Ken Morse technique, which involves slowly zooming in (or out) to create a bit of movement. This was so popular a feature of BBC TV programmes that the credit “Rostrum Camera: Ken Morse” was almost a cliche, despite there being a whole industry of other rostrum camera operators out there (including my old friend Herman, who ran a rostrum camera business in Soho in the 1980s).
Fast forward to 2013, and audiences expect a bit more. In an ideal world of unlimited budgets, the perfect solution would be to accurately 3D model every single object in your photograph (or painting), and then paint the photograph onto that 3D model, filing in the backs of objects with new data. This is the technique used to great effect in the opening sequence of the 2011 big budget Hollywood film “Captain America: The First Avenger”. Hop over to Art of the Title to watch this sequence and read a really informative interview with the people who made it, and to get a feel for why this technique is way too expensive to consider for anything less than a blockbuster film.
So, we don’t have a Hollywood budget, how can we cut corners and still get good results? Well, there are 3 other techniques that we can use. Recent versions of Adobe Photoshop allow you to specify a 3D grid, which you can (with a bit of tweaking) bring into After Effects. Projecting your image onto this 3D object can give impressive results, but only if your source material is composed of solid planes, preferably at 90 degree angles. Video Copilot has a nice tutorial on this, showing how well it works on a photo of an alleyway, but unfortunately Photoshop isn’t a 3D modelling program (and we’d be back to square one if it was) so more complex shapes such as human faces will prove impossible to model with it.
So, we have 2 techniques remaining, both of which can handle more complex shapes, with different levels of accuracy. Firstly there’s the better known technique of displacement mapping. This is a quick and dirty method that can give acceptable results, if you don’t look too closely. I won’t give a full tutorial here (there are already plenty online), but basically the idea is to draw a grayscale image that represents how far away each part of the scene is. Here, I’ve taken a portion of Grant Wood’s famous “American Gothic” (as I can now safely do, since the copyright on his work expired 2 days ago), and airbrushed a rough guess at where the closer parts of the image are.
Dropping this and the original image into After Effects, I’ve used the “Displacement Map” effect to tell the software to move the source pixels to the left or right depending on how bright that part of the displacement map is. The results are flawed (I’m no good at drawing pitchforks freehand, but even with a lot of work, thin elements like this, and the glasses cause problems), but give you an idea of what’s possible (and what isn’t). Note that there is no real difference in depth between the background and the edges of the characters, and I’ve pushed the movement further than it really wants to go, to show the problems caused by the fact that elements can’t be made to move past each other using this technique.
So, on to the last method – one I’ve not seen documented on the web, so I’ll give a quick tutorial. Here’s the image I’m using, a painting (possibly a self portrait) by Jan Van Eyck, which is also out of copyright, having been painted over 550 years ago.
In Photoshop, I’ve painted out the furthest away parts of the image, then duplicated the remaining parts and repeated the process, to make a total of 14 layers. The last one only has the tip of the nose, while this is one of the middle ones:
Once this process is done, I brought the .psd file with it’s transparent layers into After Effects, and pushed each of the layers back into 3D space. Adding a camera (with a bit of movement) brings the results to life – with the added benefit of being able to add depth of field effects, 3D particle systems, and in this case a bit of 3D text, all of which are not possible with displacement mapping.
…which is a rather convincing three-dimensional result for just a few hours work, if I do say so myself.
This is just a low-quality animated gif, but I’ll be putting the whole finished sequence (which includes the firey pentagrams from the last post) on this blog in glorious HD quality soon, and it looks really nice.When I put it up, see if you can spot where I’ve used the model and project technique, where I’ve used layers, and where I’ve been able to get away with displacement maps.
Of course, this layered slices method does have it’s drawbacks (look at the top of the turban near the end of the shot, and you’ll see obvious repeats that need fixing), but it can support much wider angles than displacement maps (but not the full 360 degrees that accurate 3D mdels allow), and if more layers are used, the quality goes up proportionally.