"Resurface"

The Idea

Sometimes, you put aside some feelings, got over them or whatever. Still, they're part of you, and even if you've forgotten about them (which I didn't in my case, but that's a whole different story), something may happen which sparks your memories. It may happen then that something may disrupt that - again.

The Objects

Modelling the scene took a lot less time (a few hours, maybe 4) than actually rendering it (32 hours).

The frame is just a bundle of cylinders and sphere, along with a box mapped with an older image I don't have on display here anymore. The image is actually post- processed for more contrast, as some of the red would have been lost when just applying it in a scene.

The cardboard-box is a set of boxes and prisms. The boxes create the basic shape, and the prisms are used to model the lashes which are supposed to hold a box together. If you analyze the box closely, you'll notice that the lashes don't actually (when taken apart) resemble a typical box. I had given thought to that, but wanted a more artistic approach to things. The boxes you're mind uses never are like the ones you actually go out and buy.

The broken glass is made of two parts. The first part is modelled with two prisms, each fit into the corners of the frame to have some "intact" glass-framing. The edge is modelled on random-basis. For the shards, I've conceived a simple algorithm, which does the following:

  1. Randomly generate 3 corners in a spherical hull
  2. Move the 3 corners to a random location in the scene
  3. Shoot rays from the corners down onto the scene
  4. Take the three placed points, calculate a normal for the resulting triangle and model a triangular shard with the final data

In order to do that, I've used POV-Rays trace()-function and shot rays at the frame, both boxes and the ground-plane. Everytime a new shard was generated, that shard was bundled along with the former objects into a union, and each successive shard would thus be placed on top of older shards. It doesn't take intersection of surfaces into account, so you might notice some shards interfering with the frame or others shards, but doing extensive intersection test would have been overkill for this kind of image.

As final note on that: most CG-images screw up on close scrutiny, but I think this one does a fair job.

Finally, as an object, the plane is left, which is just the ground-plane taken from an older image of mine.

The Lighting

I've used Radiosity and Photon Lightmapping for the image.

Radiosity (as I always do since I know of that technique) is always generated in a two-step process: Generate very detailed radiosity, which gives blotchy results most of the time, especially in corners and in detailed areas with lots of obscuring objects around. The second step is to just load that data with a slightly larger error_bound, which leads POV-Ray to averaging some detailed samples into less-detailed one. Though losing some tiny (and, most of the time, unneeded) details, I get a smooth radiosity-scene overall.

For the Photon Lightmapping I've first used a two-step procedure as well: shooting the photons with a much brighter light-source, and loading those in the final image with a realistic lightsource. Thus, I would get brighter photons in the final image than realistically possible with the given lightsource. I've later decided against it, cause some side-effects (like refracted photons underneath a shard being too bright as well) were interfering with the realistic look of the image. I didn't want glowing glass!

Personal Note

I've written most of the personal motivation in the idea for the image. Perhaps I'd have to add that I did have some more details in mind when first beginning modelling, but scratched that because I didn't want to lose a sense of universatility. In this image, it is still possible to wonder who has dropped the box, how it has happened, and why the frame broke. It wouldn't have been possible with tell-tale details.

Data Protection Act