18h45:
try to retranscrire en code les spec
understand all au fur et a mesure
understand that is it a kind of forth (said in it but pas percute)
 that helps
20h10

22h
reread
try just a basic interpretor

en regardant la tele en meme tps
 at 23h30 = interpreter with few ops (addi addf)
 at 0h30  = lexer+parser
 at 2h20  = fix subi pb (inverse the argument) so that fact now work

------------------------------------------------------------------------
lendemain
14h45
 try find doc on the web
 not really useful
15h30

17h
 found book (cgpp) not really useful
 reread task
 understand principle of ray tracing
17h20

18h
reread reread, understand more ray tracing algo
remake little geometry to understand all the formule in the task
 (2 * tan 1/2 fov = width, ....)
make generator ppm
20h

22h
 pass 3 hour to find a bug in the spec task.ps, not +d but -d in the place*ray intersection
 i need to remake math to understand, reformulate the pb of intersection
 of a ray to a plane, then i can calculate an image with one obj (a plane)
1h
 allow multiple obj with different color
 try circle
 fix pb of calculus tca, the vector Rd must be normalised (cos
  dotproduct is a projection only if the vector is normalised
  => a ray must be normalised !!
1h51
 support for sphere, redo math to understand
3h51
 make the code more propice to add the illimunation component
4h10
 try incorpore lightings
at 6h work quite well
 make test until 7 oclock
7h


------------------------------------------------------------------------
lendemain
14h
 try incorpore my partial renderer with GML
 at 15h35 limited support in interaction (dont handle transformation, and
  surface_fonction are not called
 at 15h51 handel surface fonction, dont handle transfo and union 
16h

16h50
 try union,
 at 17h10 done (time cos bug in my sphere return handling of surface_fonction
 (now render: obj list -> ...)
17h10
 try movment of obj
 at 1735, limiter support for movment (only translation)
 fix bug of not same size my sphere.gml rendered than theirs
  cos i forget translate fov from deg to rad
 try handle height 
 at 18h done
 18h10 try translate plane, lourd => try fix pb of light HARD
  comprenais pas comment la lumiere pouvait assombir,
  c'est parce que le factor de C depare les 1 => 
  on arrive vers des 477 => kan on revient vers 255 ca fait bizarre
  c'est parce qu'il faut clampfer !! avec la transfo en 255
  c'est pour ca que la sphere de l'autre cote etait pas belle (car plus lumineuse)
   car le module 255 devait etre a l'interieur alors que kan lumiere
   ca depassait 
 pb des ombres qui marchait pas avec des boutons partout (sur plane
  et surface) venait du surface acne pb )
 (* if do "if lloc = r then None", pb cos precision error => des fois un peu
 au dessus, des fois un peu a l'interieur => soluce = arrange so that
 a ray origin is not on the sphere by offset it a big *)
 fixed at 19h49
 pb shere pas illumine, parce que j'avais oublie le test tca < 0 =>
  no intersection if outside the sphere
 fixed at 20h11
 done at 21h things to allow fib.pm and found bug again in task.ps, when eval surf_fonc
  we must keep the environment if this surf_fonc, it was a closure =>
  not eval (surf apply, ..) but eval2 
  and bug in interpreter GML for array, must do a reverse when get the values
 pb directionnal light, je comprenais pas (gaetan confirme mon intuition
  de faire que lj = vector oppose a dir du projo, par contre je me disais
  que si projo dans mauvais direction, alors dotproduct < 0 => ca ferait baisser
  sa lumiere => ?? mais en fait non car si projo dans le mauvais sens, de toute
  facon le shadow ray rencontrerait la sphere (puis que direction oppose) => 
  cool
 at 22h little support for directionnal, pb no more shadow
  cos i make a comparaison with distance of light and here distance is infinite =>
  must make an extra fonc for handle that
  at 22h05
  pictures sphere.ppm and fib.ppm are almost identical to what you must have :)))))))
    (for tier1 i need: 
    recursive call in good direction (S), 
    specular
    translation/rotation/... (i cheat on shere.gml by hard coding the plane value)
    procedural texturing converion x y z -> u v face, 
   it is just geometry koi, sinon toute l'architecture est la, just a few hole to fill
23h    

-----
to see what code i have at precise time, i just
 add some stuff au fur et a mesure, alors just suppr les trucs pas fait
 et t'auras le truc avant


