Permalink
Switch branches/tags
Nothing to show
Commits on Nov 5, 2011
  1. SDL noodling

    committed Nov 5, 2011
  2. TODO update

    committed Nov 5, 2011
  3. add a bunch of papers

    committed Nov 5, 2011
  4. add NOTES

    committed Nov 5, 2011
  5. fix COM-CLEAR-CANVAS

    committed Nov 5, 2011
  6. TODO update

    committed Nov 5, 2011
Commits on Nov 17, 2009
  1. get rid of NORMALIZED-VEC

     SB-CGA now does a better job with (NORMALIZE (VEC ...))
    committed Nov 15, 2009
Commits on Nov 14, 2009
  1. branchless UNPACK-SINGLE

     Thanks to Paul Khuong.
    committed Nov 14, 2009
  2. get rid of stale FP constants

     * Use SB-EXT:SINGLE-FLOAT-POSITIVE-INFINITY &co directly.
    committed Nov 14, 2009
  3. fix KD-tree construction for x86

     * Thanks to wonders of the FPU, the surface area heuristic
       missed its sanity checks. Use 1-epsilon instead of 1
       for cull out bogus splits.
    committed Nov 14, 2009
  4. 'make release' for wrapping up

     * Oh boy.
    committed Nov 14, 2009
Commits on Nov 13, 2009
  1. cleaning up

    committed Nov 13, 2009
Commits on Nov 12, 2009
  1. dependency ordering

     kernel.lisp needs the scene type
    committed Nov 12, 2009
Commits on Jul 28, 2009
  1. nuke raylisp.html

     * Outdated, and a waste of time.
    committed Jul 28, 2009
Commits on Jul 27, 2009
  1. better repaint handling for CANVAS-PANE

     * Repaint only the requested area -- if it's easy to do.  Most
       importantly, don't copy from pixmap if repainting the dirty area
       already did everything.
    
     * Expose the dirty area as CANVAS-DIRTY-REGION, so users can request
       repainting just it.
    committed Jul 27, 2009
  2. gui refactoring

     * Move canvas logic outside the main GUI file.
    
     * As a matter of fact, put it in its own package, 'cos I have a
       feeling I may want to package this up -- and if nothing else it
       easier to ask comments on it...
    committed Jul 27, 2009
  3. gui hacking

     * Areas of canvas can be nicely selected for rerendering.
    
     * Wastly improved repainting.
    
     * Misc. minor things.
    committed Jul 27, 2009
Commits on Jul 26, 2009
  1. deal with window minimization in the gui

     * Repaint the canvas.
    committed Jul 26, 2009
  2. correct(er) output redirection in the gui

     * Output from commands to the repl-pane.
    
     * By default *STANDARD-OUTPUT* is the Emacs repl.
    committed Jul 26, 2009
  3. rendering subregions from the gui

     * Also patch in FINISH-OUTPUT for panes.
    committed Jul 26, 2009
Commits on Jul 25, 2009
  1. color picker for the GUI

    committed Jul 25, 2009
  2. add reflections to TEXTURE-SHADER, delete RAYTRACE-SHADER

     ...and fix some .asd, etc minor issues from last commit.
    committed Jul 25, 2009
  3. simple wrinkled and rippled normals

     * Also enable fresnel effect in non-metallic texture patterns.
    committed Jul 25, 2009
  4. non-consing KD-tree traversal

     * Split the stack into a pointer-stack (for nodes and stackpointers that
       are fixnums) and a float-stack (for distances and points.)
    
       This way we don't need to cons up points to store them on the stack.
    committed Jul 25, 2009
  5. add ambient term to TEXTURE-SHADER

     * Nuke PHONG-SHADER, replace it in tests with TEXTURE-SHADER.
    
     * Ditto for SOLID-SHADER.
    
     * Use +IDENTITY-MATRIX+ where appropriate.
    committed Jul 25, 2009
  6. noise stuff

     * Fix noise vector generation: Noise vectors need to repeat the
       permutation twice in the same order for noise to be continuous.
    
     * Cleanup implementation of noise3 a bit.
    
     * Use globals for noise vectors for now.
    
     * Rename VECTOR-NOISE NOISE, and VECTOR-DNOISE NOISE-VEC.
    committed Jul 25, 2009
Commits on Jul 24, 2009
  1. initial sketch for adding normal perturbation to TEXTURE-SHADER

     * Splitting of normals into their own directory seems excessive,
       though.
    
     * Noted buglet in noise: I get a sharp discontinuiety at 0.0.
       Need to figure out what to do about it.
    committed Jul 24, 2009
  2. refactoring

      * Instead of defining N FOO-LAMBDA macros, add a macro-defining
        macro for it.
    committed Jul 24, 2009
  3. better sorting for KD-tree

     * Mergesort FTW! Now dragon at ~20s, and buddha at ~30.
    
       ...and the merge-loop could still be optimized. :)
    committed Jul 24, 2009
  4. optimize KD-tree, _still_

     * Reduce memory traffic during sorting a bit. Sorting still
       dominates the performance, though -- maybe I should try
       a different algorithm?
    
     * Make SURFACE-AREA-HEURISTIC non-consing, and remove generic
       artithmetic from there.
    
       ...now bunny is on target (under 2s), dragon is at 35s, and buddha
       at 46s. Go me!
    committed Jul 24, 2009