In　the　past　years,　a　novel　face　modeling　method,　the　morphable　model　is　proposed　to　reconstruct　realistic　3D　face　from　images.　The　model　uses　the　combination　of　prototypic　3D　faces　to　represent　new　faces.　Matching　the　model　to　the　given　facial　image,　the　3D　face　of　the　image　will　be　reconstructed　automatically.　Although　the　model　has　good　result　in　face　modeling,　there　are　some　disadvantages.　The　construction　of　the　model　depends　on　the　unstable　algorithm,　the　optical　flow　algorithm,　which　does　the　pixel-wise　correspondences　among　the　prototypes.　The　model　is　not　fit　for　the　complex　illuminations　and　requires　too　much　computation　in　the　model　matching.　This　paper　proposes　a　mesh　resampling　method　to　do　the　correspondences　and　construct　a　multi-resolution　model　to　reduce　the　computation.　A　multi-light　lighting　model　is　devised　to　overcome　the　model　matching　problems　in　complex　illumination.　The　experimental　results　show　above　methods　have　impressive　effectiveness　on　the　improvements　of　the　model　matching　accuracy,　efficiency,　and　illumination　adaptability.