_plan.txt

0705
rebuttal周五中午截止，4天

3.13-5.13 schedule

1.MS 1week;3.20
2.SS 0.5week:3.23
3.MS 2pass 2week:3.30,->4.6
4.SS 2pass 0.5week:4.10
5.SS ani 2pas 1week:4.17

6.MS+SS, tone, Raytrace, specular, texture, RMSE 1week:4.24
7.finish fig&video 1week:4.30
8.shader optimize, e.g. mem pool, 1week:5.6

TOO LATE!!!

2. SS 0.5week
))SSIso
#ComputeSaveInvis2[nv*16][6k]
test: 可视化，按一下键，显示一层p的Invis2[nv][p][6k]'s visRatio

#SSIso T1, 同时逐一小波至T1Coefs[nv][256]（也取256，同MS，只是巧合，够用就行了）
test: 小心点。日后可变化p数，对比效果。

#修改TransportRadiance，并入a*Td + b*T1
test: MS+SSIso一起画出来。

))SSAniso
a)SVD(HG)为g[4][6*32*32],h[4][6k]，g,h均为4*6k的float*。
b)将isoT1中b)vpT1改为： float vpT1k[k][vert][side,offset],即，有k(4)个T1。
每个里面都用vpT1再乘内存里load进来的gk(wi')(像Fwi一样)。
c)小波4个vpT1k为nv*256的数组: T1kCoef[nv][256]
d)TR = SIGMA(a * TdCoefs(xo,wj) + b * SIGMA(h[k][wo]*T1Coef(xo,wj)) * L(wj))


1.MS 1week
to do : Td 对称可利用
矩阵T(xo,wi)=sigma r(xi-xo) * E(xi,wi) d(xi)
结果为：每一元素为
T(xo,wi) = R(xi,xo)×E(xi,wi)。
3个矩阵维度可表示为 
xo*wi = (xo*xi) × (xi*wi)
))先把map预采样到32*32，实时根据w或side,i,j找点（都很简单）。以后再实时采，并补边。
)) V*n 4-term wavelet, as Td, L->l 4 term, to render diffuse day4.3
))test 1day
))figure
4(a): bunny72k MS
4(d): bunny diffuse
7(a): buddha56k MS
7(d): buddha diffuse
//since these meshes are tooo big, can't do them. Left to milestone3