TITLE: A Mystery NAME: Michael Hunter COUNTRY: USA EMAIL: intertek@one.net WEBPAGE: http://www.interactivetechnologies.net http://www.interactivetechnologies.net/making_of_Mystery/index.htm TOPIC: Mystery COPYRIGHT: I SUBMIT TO THE STANDARD RAYTRACING COMPETITION COPYRIGHT. JPGFILE: a_myster.jpg RENDERER USED: 3D Studio Max Version 5.1 TOOLS USED: 3D Studio Max, PhotoShop (for texture maps) RENDER TIME: 9 Minutes @ 2600 x 1440 HARDWARE USED: Pentium 4 1.8 GHz 261 MB RAM IMPORTAINT NOTE: Let me begin by saying I created a web site about the making of my image. It shows the evolution of the idea, detailed views of the image, technical discussion with illustrations, and links to websites I found helpful: http://www.interactivetechnologies.net/making_of_Mystery/index.htm IMAGE DESCRIPTION: My inspiration is not a book or a movie or some unexplained event it is the art movement Surrealism. I have enjoyed Surrealistic art for many years, especially De Chirico_s and Paul Delvaux_s paintings. From Paul Klee to Dali Surrealism is mysterious and disturbing like an anxious dream. I wanted to get the same dream like feeling. I looked at every Surrealistic image I could find on the internet and in books. Over time, I started to see how different artists portray dream worlds and what aspects of the images made them seem not of this world. There are obvious indicators such as limp clocks in Dali_s Persistence of Memory or the floating castle in Magritte_s Castle of the Pyrenees but I was more interested in subtle things that give Surrealism its hallmark, slightly disturbed, delusional view. For example De Chirico freely distorts perspective to suit his compositional needs. Not only does it make his seemingly primitive images dynamic but also slightly disorientating. As if looking though someone else_s broken glasses, De Chirico_s damaged perspective let_s you know this world is not your world. Is it possible to create a Surrealistic image that has no obviously illogical elements? Yes. There are a small number of paintings that are mysterious but don_t break the laws of physics (at least not overtly). Unfortunately Surrealism has become synonymous for any old silly picture. There seems to be a misunderstanding. You can make a picture of a flying elephant and certainly that is not realistic but that doesn_t make it Surrealistic by default. I wanted to make a truly Surrealistic image, free of gimmicks, free of flying elephants. This seemed like a worthwhile challenge. I leave it to you to decide if I succeeded. DESCRIPTION OF HOW THIS IMAGE WAS CREATED: I worked mostly in 3D Studio Max while developing the idea. I can hear people shaking their fingers at me already. Hear me out before you scorn me. Since I was dealing in dreams I didn_t want to do too much planning. I wanted instead to let the image develop on its own. My approach was to place simple placeholders for items, move them about, move the camera around and delete away anything that didn_t inspire me. I kept it very general for as long as I could but in the end some ideas stuck and the image started to be more defined. When I THOUGHT I was sure about a model, a light, a texture I would substitute more refined versions. I will admit that this is a painful and slow process because you can never let yourself keep something in the image simply because you_ve invested too much time in it to through it away. After all it_s not the models that are important it_s the image as a whole. THINGS I LEARNED: I had enormous difficulties with the sky. I put a procedural map on a _sky sphere_ but the curvature of the sphere made the clouds at the top compressed vertically and the clouds on the horizon were as tall as they were wide - due to perspective. What was worse, I wanted to soften the horizon using a little bit of fog but being near the center of the sky sphere the entire sky was evenly affected. I came to the conclusion that sky spheres only work in the real world and used a large parabolic curve instead. So instead of looking at a concave surface we are now looking at a convex one. (You can see an illustration of this on the Making of a Mystery web site mentioned above) All of the models in the image were created for this competition. Some of them _ namely the Victorian sofa _ were challenging. With the sofa I started thinking about the part that looked the most difficult _ the buttons on the back of the sofa. Making one button and one bump of upholstery didn_t seem near as difficult though. So I did that. After that I copied it several times forming a wall of buttons and bumps. I adjusted the height of each to make the curve at the top of the sofa. Finally I bent the back so it looked like a _C_ from above. It was much easier than I thought it would be, maybe because I thought about it before jumping in and doing it. I think it is a natural tendency to create overly complex models because you know they will look good even if they are too small to see. I tried hard to keep bitmaps small and geometry simple. The train has only 834 faces and two 200x200 bitmaps. The sofa by comparisons has 18,000 faces and no maps. The man and woman have about 7,000 faces each (including the baby and trumpet). Sometimes I had to revise a model because it was to rough but I gained several significant advantages by keeping it simple. First it_s easier to model something with less geometry and it takes fewer resources in general so files are smaller, less RAM is used etc. But most importantly it renders faster so you can do more rendering at higher resolutions. Even at the end I could make a good size final rendering in three minutes. The woman with the baby has a hierarchical, jointed skeleton in side her mesh. Its sole purpose is to allow easy repositioning. It is a very powerful tool built into 3D Studio Max however it is extremely complicated and takes a long time to set up properly. The man with the trumpet was an addition made just hours before the deadline (I hope I made the right decision adding him). He had to be modeled in position. If I need to move the man_s arm it_s nearly as time consuming as building a new arm. For the woman, I only need to bend the skeleton arm and she follows. When time permits, setting up better models save time on revisions so it_s easier to get the end result you are looking for. The grass accounts for 85% of the geometry in the image (about 430,000 faces out of 506,000). It would have been considerably more if all of the ground was covered with grass. I doubt that my PC would be able to render that. So clearly the trick was to switch from physical geometry to a map without being to obvious. I made the grass in rectangular patches then covered the ground with these. As the patches move further away from the camera I lowered the opacity of the tiles fading out the grass. Though I did most of the brainstorming in my 3D modeler, I found it very helpful for testing out ideas to take a test rendering into PhotoShop and paint over it. Of course that is not permitted in the final image but it is a very fast way to see what something would look like. I look forward to hearing your comments!