Implementation ofSpeakingSoftMan inGame

Unlike traditional NPCs(Nonplayer Characters) whicharejust"smart"programs, Thisthesis considers SoftMans asNPCmodel, whichlive inasimulated worldand haveasynthetic bodywithautonomous behaviors. Basedon thisnovelidea, we research text-driven lipmotionsynthesis systemforSoftManin game.First, we divideChinese phonemesinto11basic static visemes inwhichwhole3D virtual faceisconsidered. Secondly, textscripts aredivided intophonemesequences. Thirdly, properintermediate animation frames arecreated bymorphing technique. Finally, those frames arerendered byvertex shader technique. I INTRODUCTION Thegameindustry evolves veryfast. Thequality of commercial computer gamesisdirectly related totheir entertainment value. Historically ,traditional Nonplayer Characters(NPCs) are just "smart" programs, they aregenerally given direct access tothegamedata, freetoextract whatever theyneed; Logically theymayknoweverything andneverforget whatever known. Butitisunfair toPlayer Characters(PCs). NPCsthemselves mayexplore thesameworld aswellas PCswhile obtaining knowledge andachieving their goals. Intelligent Animation basedon artificial intelligence, artificial life anddigital technology should beconsidered as Al NPCs11). SoftMans(SMs) and Generalized SoftMans(GSMs) areakindofvirtual robot, whichare subject toconstraints oftheir environmentE21131. GSMsmay haveavirtual body. Itisnatural toconsider GSMsasNPCs. Theaddedfeeling ofreality just increases thegamer's potential enjoyment ofeachgame.Withemotions, all nonplayer character behaviors wouldseemmorerealistic andgenerally increase theimmersiveness ofthegame environment. Inaddition tohelping display emotion, the mouthisgreat forcommunicating. Thelips arepowerful little muscle machines thatcancontort themselves into manyshapes. Themouthchanges shape tohelp create the sounds that compose thespeech. Thepaper will propose amethod toimplement speaking SMs(generally, they areGSMs). Theflowisdepicted inFig 1.First, textscripts aredivided into phoneme sequences, then eachphoneme isplayed according toits recorded voice andatthesametimetheproper facial animation ofthe phoneme isrendered. Obviously, creating visemes, building phoneme sequences andrealizing animations arethree main problems which will bediscussed inflowing sections. text bword syllable