宁静致远

Here’s a small animation: night.avi

and here’s the final scene:

Goals:
1. Fix refraction

2. Add spotlights. Add soft shadows.

3. Motion blur.

I’m behind…finally I put things together…

improved material

Goals:
1. fix bump mapping
2. model more ladles and chopsticks…

Goal:
Depth of field.

Goals:
1. Render images for primitives with texture and bump mapping.
2. Clean up code and tag.
3. Model other parts of the scene (spoons, walls, etc) and put them together.
4. Adaptive sampling to improve performance. — DONE

1. Model the table and walls
2. Apply UV texture correctly to those objects
3. Apply UV texture to one of the bowls

first piece. no bump mapping.

add bump mapping (I’m ahead of schedule! :D)

Of course bump mapping needs adjusting.

Very useful plugin for creating normal maps from texture: GIMP normalmap plugin

Goal:
1. Simple texture mapping (dropped)
2. Blender model
3. Find and adjust texture for the wooden table
4. (change) Write a new .obj file loader which can load UV texture coordinates and vertex normals correctly.

a fantastic place to get good textures: Mayang’s Free Texture Library

Blender took me much longer than I thought…

The bowl is far from perfect, but I think finally I got the shape right. Some normals are weird, but I hope that will disappear after I add Phong shading.

normals and texture coords can be loaded now.
Phong shading is mostly done (some normals are still weird, but I blame the modeling tool for that, haha)

Goals:
1. build octree.
2. build kd-tree if there is time
3. render the complex scene and compare performance
4. compare octree with kd-tree if possible

Current Status:
1. didn’t build octree. used kdtree directly.
2. use tree only for mesh.
3. 1m20s for 1792 triangles (without tree), 17s after adding tree; 6m38s for the cow (5804 triangles), 19.7s with tree. (512×512, 1 sample per pixel)

This takes 17m54s. (47956 triangles, 4 samples per pixel, shadow on)

Goals:
1. build Cornell box
2. Reflection of metal surfaces
3. Refraction

Detailed Cornell box data can be found at Cornell Box Data. Simple reflection. Point light source.

Not quite a correct reflectance model. Fresnel part is missing.

(verified that refraction doesn’t quite work…)