wH8ϫ M؋cgf7qIq'Q$P~I {ov?sV;xVW$ϯ_S]]/±T^x, (s3T*:` {Kž` Y~_tf;q}Yq=0Ҿ/wI< BUܳp`3qȥK|:ۓŽ ܛ1.]~bK-Nh߸͖~5:໓|_@mڽiz[ߤ7k흶<-8x |]؃3#?g{Ѥv'fKH/`s]jaݧu,z緅kA9V@m>2sǰۓo'NJ80 |JS>O:/\m'[ǜE8$6/J' +{jSMiIOB>5 :=q?k+ nή?:) 4heMm6ðu^\Nv>χp,Vn{l7anxh;Yz4oO]49$r ]}0z X}X.Ccm0\>6򸾻c{ռ{Cs a}!*o4??%UGcaqaq!2>ݍF0@ykk+usq&YrqF^Bj!8ݳMfMƅpʼSW73TнX 7_/W5K`vu(ZO7If3ٚ/7i❷I?_=~CI:tzu~x0N?>7%?w~gs4NL`H7C?qga38P.-(?lHqp6fՅpUWX[u6!fgrI 5z.l?V ؼ<_E^K^c8!Pm\S[8w@fɓ{;en.8s1.Z,0Qj>Z^|wfT $ii 6lo:F>:iALe$ڱtJSx}5WݧipSOO;TaG&9hn|`"s-m> !ڞ#|Y]Y#>ƍk,|CMW7VC.n.\q~ QTCpl > o}xߚ _kH]WMSW~Z}85>pY$G+J=4#'j{b')9z0=‹ui{. _n?ݝMus=6/\d@WON/N.RLl^\lvo>]\;OsͧkSM&`ɯ[ r..ק8 */bgY0sVEܜѕ>Oͅ<r pwGiթjBu9FNЇ4qt!<~~]_1Op5$75N.GŻ_JTx}(~uX~W#?$ v<\͵5}0m|l]o=>n} bfO9D<_O_ yhXQ_sG_m%d jz-%TxҤb> !j*?yE}0mnHz>u/ ww7X-F=h?]Cwl7q%5l%k!õ7CM|Sd}]E>v;Uŋ}}۞+oQi4Y;=99ؤ|?#?i?ر-f;_˧ޗ寻/_o x7OΗ\;Xy<~@q#I]Āji VwpNXy9w/ P.&@W2Ӡ- ZmT[ov@S@]S~>ݤ<0-,bP:D o|L%Pw_1xPb|N~Kjt8IB#Kݛ55<+L'H8dژ 2dGU4R 4:k3[ϴl]Q,?O H=ow& ]NV~;: <֧ !@Z`Jk#Mjm'm,yUr_dᅀjH悹'9<َTNOZM3#X_ e{~O\ge+Ho2OHS犁+b#˫A1vzϟ3㪗\#$Li+} puhɾOqB7b_mzjooy <`O=K0IU',IvU麦ZeqYǾgEZX`t/).1tqc ""r0}x -7m'/yfmݬm1['q<pD~Y'߀Y|I:;K֏ww΢>'öYS*x pή)z*k6 $oop" #  |tke{QY@2$auک*~>э D[T0DlSe4Z.D>dӈAi}׹4>0$?`R`lgXn+k A T e =Z}Keb&H!y"p-pȅ10~f #s Dϔ+FȓLѳ<:SI07V+J 4Md?\rDRc)9mQ^|E?)ҭPaEo]lKD_o*6 Ay!l((w% ݝnοl9@74D~HFxzɀt҉=76{G}m9<ɾF.2ۋ<7f lVci&pq.R[Vwf)x]4uE Dp|U>N"n/]76n tHi@Dl1B߉Iecu;y=$p@|!\n ;tg慠FO-v UK7L^ztg׮0%{]†G}z2, 3x$QNzxq2{wyh}nt˗nWUzǢT8rnztt`psf"@עGvq1\Zϒ8j&mTR_3/fa"1S2!\k Gv̋.YPPIHtT,CPCJpWETwnrxorjke(a f }n7:цmxC֓HnUAD_!("iw 6e7#@ 4)93rn}@Oٹ}w?.%ٗ@y| *jiyt~o{x{n$W62z푍'!~U1w$7M|fQ⠵;hXuڙlR 6*r$˚ݰk&HK񯮙Ikd6 %O 5B>K!Wz4 piE] 3xlX5- $CA:Ô:Hex]\j@K*nZ&$:^'Xiy9Ef{#F@S@=KmBPH/ $j) c(;ɸORfIKcz&Ny'"EFV΄[6_Wiʨcd]Q 4v}v󔺙uZ5c{..(0ĨY_nBD敭%PKUT\>+lf?mr2xaʼ8# <)8P=B)"}vQܶZʑ:5fGvea%1|`ʳ}I/Ͼ"4)\AG B FL̀fDLp5.&p Wzr̅I`L?wHFb~o>BC"*Iu7+^K6+7=ΈsefdWs. k߼ +uIі :#}(y2-TjJi`OMvm&q09g執¯yyЅQ{=396|>9 DT)TFiE49`_=2Xim\м9)idk@$Cg@'}p|O^ڐGJJٌ UF?(aP%?v)޷nfc<::{/ַQb|sN?;ω.㳟ͭ;a}\.F[;^|5w;O FgCnaFbx;۶~x<Ďͽ9|T~/Lvڶx=z}}޹qo>ݽwW/]vyC6㏷睇ës>|Z]}>\-kْm%o?Bgm>5.OFмx&ݬwwqwͨGt|'AVׇx4x9׾c+w~WO;'OM)>&j^ݝ|vpyrs%[gOכ_Ci==c<wsjkּm00`;z5|>'FC 0Ed~{ҏݭn?_oN| vOv8M8sM==1=~蟼;Dww۷[yx4b)y~Ԑ|rWo|~u5m= ]GOE"Apkx0_vҏ/緿pqx~חZwON'}2]}h ゚ݞ;w4x|{ |_~q$}>|OGOѩԝ`K~3ՅHY=FgNX#wі~O@+;/}yA,V>|OGn{nxpk>a9pnևOy} 7}Hﶿ RiN`xqލvr7q>mN>Ona<҃oͳÏ}'?>|޿^'?\Nq;Oժ)/n8OHowppx'=~[ -AO}=8x+)я7i$lO (>>юi}r/Շ添;K~J?-_gl;i7,R4_^_|uo? W{.\owwg÷Wg?vVٖ/'Ë'y>zssޖ*tzmڧƇrZN><*tWy_׻mS:XoǧKWkpzl=_ؒn8Z~\H(v{t/|x<[<^oL:`O;??vn9-ޜOϟ>O?^G:>t?:d~0~ 'lOg%v ә}}9y0OMw>QSXoo>鏣ųr}}!m|` ʅ;Vۓ$0O=;잴FoUfzq$}:+xzb?9pp7>5roOgezKg^oO$Oq{r<[W=qcX~X<<45۶1z͐u黑typ?;ӖMӼnS>|:Rޏpr^ wwX iO)iI_6J+wG}xOv+>Sf>n]no_ҳi\|;<\ .7I[ I"Y{T nswx{|~uztq~*)?&8)o)kǃ-ܝ;n}z=sW?Y>X?Nyɍ~'Rz0sr"nϢ㛳ͧ}{d޺>~x߾{u(Ǟ^ߝ .;Fß֯UT}kngg*\ 7Mݳ'ݧyxw}t{}|o0x!I֏F_j:Uv}k>ןo؝vsxM'ww%.7!tx85wq<,6>7ww;Ӈǻ&Ⱥqz}7Oí]8~~8֏޾zvh??vo{ݶI8y{}7}zkKw}+h>>O?Um)7Jt \0ʓ/vhw_ m?::<XҊw[kc!~.:?4_Yʯ+c9V΃t8sv`7FV۾[Dޏvǃr|nK~6y>O'?rׇ۟i?~+X< Nӛn9=gA.>\6Ou鯟Oӝ݋w/kzut} M\dK)_h1 $R*BZ wGk=Zn(tƿ RvIa`p'~H"wuN1$AD7[n"147>zcwڻٺ8aS *Cl${m~aIJ&aUW8YVU|e`\܅,|F0BT-[2zt]A2aO6Nf )M,rJ |*+ ]5\< †WT2B,to]@NJ@]"2͗??_w dpH 调 .ŖDt&[ Y+)2 >5 $)![Ue˚['q}9~vsN [t4h"W0&=[÷r{8An;Ķjm;7{,EA( Q֖[$Ҫ<|!ſ R %Jy(ozH=Yi _ {Kt ('{1v@  }+y28 .DRv2/+3=gv* 8wAJ?;13!87s?L#|> pO /Jn5>Q"!;0 9kcЕ?k"]!ũpΠvR|mP1Om78&΁llv{ֽ/!XO5 xnluSli.k!kx6pd ^o`Sǻm?,fU.s0_y Օ/h)B)63gCDR(Vs J=h-%' -ĽNDm@ȞcU`H X@;~O4i -/E#_Z Bpپu^[%LS{+_Qw TFز5yD<ގm\GgO hιa`CHFșh>o)-OfLʙLKama3y qEŰ}{~(w>~LOƯ*XR->Ȕ)ϳ[)SiPvV"W1t܎Zuڮ[`zp|!|Z9lԤ {6Ǎt=! D)Y[/кzQmn;w/-he&E3P<_Ueag=rzq-"h~2N|KBB%G0ך!arp+=5<кP{X3|zާ^`t!FKfwH)iOq'n!%ΑXns J 8B7o'5_:m!lC"|J ^l"p p9;X[7KGm6qÖ?ivNܾUCr+h30 >׸jg$`Hu"|1Cy sImû?zQ߆*>j[ۧn?"Pv٧Gq}0l d|oyw>}<~<4>[d]Le|u|쾕0`#>$uOb8v >n,8 X:Aӻ>a7y.ۯ36Ou;C?SK s0{u8ƨ%5FFq84r[47kg j6݆{d[Mۓmcv_GHm!Qb<ЌZN` Xxeo.tCqQ 5H:tۈtdoƠʪDa:!oQU@/y'l zQV8QQ]6¥Cgڛ(7tU`GR--(Ԅө|פ7@zݕB_>zoJQGZrzfVHpͨzp9WgԷ4-@DnX8v> > @mLڝPz<(=GxdR+bK2;|SUQD3 `s]OA6:'ڑSwsglw~!jj1](+1L}=LrUfծ`߯۹:/: W.//u5A/C'`i:23MZ|.f;Q)=uHfN0ܰ ommmdOTGYvu<"SY zU`U0q@m4 H7f'Ln&5&p FW?JcSk1x1*Afd}x-',AfǍ"uC@ Z/ |3G)Xx=W" jBZa(;EXM0KUZ#IWR w.7&]g%Ok0 c;C wf*)?w:Cѯ-p(4Ku՛oOO`ٹge %0 CV.3D:W\8nQ=l4L;~#%b51lMJk|FD#TK^G㦢YittwCY"G&^.әeV1[x <Vh7\Kw&"*j`;|Cep64rI_S[l :vEےf$;pFQ2׵qv/?Gnh-*4Kt-J} $vdKT׫`rP#X-k;  g7\E1\r@ߺਃV 1:7o׻."ӕw"? @ߏށ 1W``Yk A¶vݭZ}Q(@,5:frb5}8x9Dw:\ZeՆHg?]{Ro,g.أbbBo@ EV+TxB 6•26cۨ>Aܸ&wʚkl`,9=?$sC"ȧؓw5:租/;k_{yqԘwuעJɹqE6A8Q?2uN"Vvk> bǢ%drEEwC_~g*ʗk_i$ WČ–]A]5ffV#)5pEd~N\{ϙ,5Iu )}TTLTpQd&?C; ~`}gMg@:A+^O1X1h{IQ=J[$v#;]E7.?C1{J/cSmTMNCH 3/.+iX\lcEHf-uO\ڎaצ!dwsr7`{}ؼ.x;9~֛9FP}Gp]f{帘0.EǙ%lqրi:A.dHzD}PU3nv_7'LZ1B_D} x'EWz5q:wb)<&vHF2> USSΨpo !CύP鬧CΦwArďqyJlG\!^je7IkBTL֗,ltǀAAд r}'uzDCq4f"f[Q8 !x!8xcP\c_ըWZ!sɰv4F;\Tͩb+攝΋ QcҟWH^DaX`Vp&%3]Q1(4jqvL9}hhfν<"~9* <x+vRųc)bNgqZA:hR1OQ/HMGj'%Dipr^Z$&i5#5%%^er7%7u<@P2wsŞ5ZNfl؜e˶`d5ztA9%_-V 7%7YԔw rf'YS'b67(`-to$Ѻ=x/XhsSa Չ4!'3@jt{f9Zy],H-`ƷgUQlh"(ʦ IJ TZJ`bGdD6\/jE EW\J!Z3vR<2Wk]&m]Wle, 6$uNrf³Ly*QoPz]s/6DsI?.{ H H[=;Nwt.%|@n"vZEE3:?.|]V]E;{>,yqdV/Һ~c~fnxX봾v8F~WG"rˑ sGVL~K 4];u$+WMȶˬ-+yU3büQTU۵R53mB&s8Gjt%2*xZ*\RL'P5YTq DrOzčݬXHcz͓lCDPzZ DFitUcK+ jǡ -`S3CՊ=Jvxt8Gt#0mF]5őJ@ƪd*bj`CMpf3S3 & ύNHFz LKzn?mrJv\OAA)Y43ve 9%VzATIGcoXkYSqT-Y1U[kt%Mky\>_KU<%D#DXKMfq{"ph` fb$ش Wv io*69Q(2g@f%>ca?ᥴM}7$دoɁYAc"ƥ|83)-}m&١ڣH0.08t5kbF CПHQ؈sO7Cy2 μ$"lvLI( @X:1Q+Ѻd**eGZL88d9 /l=Rb.~6pPXAj Pf,Ӡ'v[Gpp4--")&nhՄ ^7A96I%G ,6W.j2Dih@oj2W]!p)dRۥUd .ٴF WhZ,fVgBй;Xm* PO -PE Bߓ,CV:ڎ&#!SEs;duQѓ(C!p\[ᨱcgL:e?I44٭/I,_:F7e)IMV9_nU4W~+QDVI")(+"s 4 O/=mo!HsGe^Z U)R-ݐ qG ~>7¦5jSTגTP<#'q4[l=dC_ZL$=q6'?o?"ULODnc5](C!q> l,NxNE@f9Y`BK<9_cǞ5 x9h)CN/7Zh]4vX_vLDJE؅A<=/* pque.6ݰ?ʄÙ5]4#if8X͆7ɝIu0јj,p`:}%*Ofhݟ`:=NjHYR (Ϫ1B(H:ά}vk|Xdr:"ۭQ)B4@Bքn B4ͅ(̹ ^seH֌y< U2ByG4sK4 `"io5aj(3PK.lz+ uϦio;2Ӏl~ck<}Y"&sJ c)~q|01$,(0fkɋ.nF l)2ċ. uaxazq oi qn,nu嚶ꀻW8Kff!ijttZFjV3qԼXuٻ"%]dGv D0%)m+2L͡b'Y\VYHI!8 *FbT9/S\+dc+(Oj^^܅aD %\5Ņ6V{b4s'^43HD3' 'k_>.*+<_v2`ʨg0W1`QKA:`b9%RC[j sGO'SgDni)U~o=|B2//r5]pFwnWs>I7L{zoqUvTƊhA44LRЬdJHEB6ǔU^Lץh o [E^!'; p~bЋMγ ݊@n~,A[ّ gR7Wـڒʓ-MrOQOvQ"4ڤCGOעHbT-6dGU>Aך'bz[=rb-Zx8EZPѕeY\5!酘]a쒘;>|><옌O}. #ڙ9T$ ofBlu8qV0ݎ3%5^E5F6ߕdX#ӆ^؟G2b(\[^ Dīyg@ҪhV=L? M\ʁfD?uttI(H)pXhqI5+SUPp jY~y׎4Un,{8MJok1d'Yj\jgG(BFQ:0gGr˷nBMl/C\h) $] ~c6sҾ}1t5ei$/*r&O%nabE dc5T%Adz)zԦ- 9ߔHnDZ5r(čν0kb3Xϥy2%p\~??n-6C,T)@UA\7U c&Lܩ~m1G>TwEКhbFcv\!:(4[sLOےU4g$W,DR}{XpjO]{#(;;x֩#U>Urt_u̡`ekap-|S9@s^э"u\bQTW:rH -uMy0~yBLJ2+ЁLKEdrCHb$|k ʈKIA2fta|S"F=tKylc.EIU/>H({ANn ~nUId98.eQ.M>ݕD'h\J:ZUC:a03i^j'Sa*e-B7lՆOƎޗTű$^'U`T"C{+cz^􊎒^OG7PsWzN7[A۟kiwՑ qR UK 02?bʕݸB _Ef9b9P}" 0|(oj"Q#GV4mIG 4ԷVA"Ћ!Rf(6#J}(vKN4[Z|UxWv3U3KU} pnZPfU1nBGD\\ "PWIc҂Adoeb2 NEU&$&OUKńL_Fi>YėΛ2:x0k\(e`$^f€n ̝mTR4oE⍲{z Fر"UD[ pud%>\EvwI!2W#s'& Wf9J@ʽ+x\TM)setE*5+W iëSdU1%Jir1vSr7tRǁ2sfSɟ7 JEd<{ =^0.l}*K3('G )\ϝ+Do6L0I3w8cC"~!kţ :;*IP0E yoE19̨[eHh,bJj~F(XRYlW.=1%yf̗Cn,i8 fYI %@d;czpztyʿN%Hr1>̨yT2l'ip[=T* (GybGE߶+ KkA+b]*k&*(QgGaN{9}YKydZZy2DBoN ^2TN9i)FRULN\9MoT]l-=N:-df;Xڪ~`0.}Z16sa=BcO{bǞyÜK kdK͗:]_^@k9㙊L0lhnz-J >|.(1*LA,-NOxP$h te*m#x6@UfF <;҄fE?PuVFerb&Wa7 pzah$t8꜖cyw _ ] zM8ZW5R(*(Z*U_[nĕ:fcR'He{x,+EPK&h$p=?:29bʋ5lM b9EI"i4Ebt79t*9Zy((#+x"vʤfEHО*9%cA%~# nf4#C'3Fvٰա4Xk˔f/qfi.蓚nF ^ӕY=h~ 41t0L^7ʼn+/:- VcB 9>R_,b1)ZKKR)2l+$`R+χ<{Pau {q0S{CX/.6UeXR5Lcaؙh/NhUWM &q1&*;ng2F)K)@G@VSG[#G G}{&\S 7S'z5,ƂiEEQ `"4\ҹ*Xm1=GAުO&&VV8/Ro鼜*TYg <Y(%hYd@@kA,2]F;#t:(2YUT\#*S.pɊ=6`'(Hk'IcC8_hvI}aљd-` ^e)`K-j<i.g2-߭{Dy `AHMH!Wl~Q뾡B@8o%8̓+ҫ |%t+9S$x+Y k,j<#\:=p" I\(ҋ*ş%Eb I6`w56kUDN$/I©j)'W!7_,(M^iN+EE%to2+ޗFM-<j}jJ 3H 1y f47ΐ2 (AR/,HV/ t%ǎ޶%.kV/`(#zp?LVõ~)jfNBK_( GcFSR/ sfn f yZ҂1:kЖL<3Hp:rFn'\kP;N=ߚE=cӑ}δ X.δ*ͱc.Gѥ &S wqbQ(],lmfԕ-e㵃XU !*<6- iyXIΨc+u.ģgq(a4ӗ˃p.N9y>Lƛ =1/~!I8kB9֩~u_*$Q0aRdX쯮dY'D"|dtxq5q464!ʂr(ĊW ai>؎; &=7w'I.Pbϡ#Rh/ ;΢0"SOԅvɵ륁iY ]5 z>:.[m3WlniphZ(+a: VgR],`JVVe(037,H)b>t$a Zi p6SUgKū\23pf#\cKS$JfpDñ$+M^vrN L0-Ebh3 kjh_GrjfWxуͬ6UzJc&X9 ^v&=6Oew~n]4PPc$nphYHnp'ywlv)rt\Dqf-xn/Ϥ*E;= _xn* ܭo pP)!/up9eFWJfG[9WvpP=?j6)PIm aX57օ/wXJ1B$FS驞| =I;vٓzvS#c |j~^MFAfjvf\+oyxYz;DTP|,cbPh{Vnq37K ~V_M^3j[bn;W8!$>/zeyEM%m^*ػ7!dcIMq賍(4 9Ndmz(9 hRBVhJ$-=*K2yk8DSrGʹѰcRgZ&ŒE"숈w8(u==R*DE$*yu=-_˒+y:e4Q qCN6z~o%-ݙҎ# } `nUX0Q)* 05>5Vڭͅaa/x jFkkcOOFkj` jэ2u{Uu8:(KpjIT SPt_l$̔p>hI$IQc sO\2l%$]SCǫ uZ]>3SC6V83-Fu*ꖏ96TMДW+ M2JI!c n"l^ī&mBǀɁ:ʭKpঋFEobnU-ij˳pDlu;ݙZlU:Ftn[ jK={͞s)pA2X̪YuX*hV]3UT,vze0OvC ,5%@BfXvix=qh%KKlIҎ5z:u-kԟǭmfjU˂EKj6?]( "g| dC^ܥ#| AYa37yeOQr-`T"DX13Xa7],#g nK,a< T*A.⌏7z3_Ƹ'w|W3{#֔!*,iTaI3gY}#-U1<;*5O7.U ٧PN˰D*YT26XNʩYqY;Zsj1p@^~KײHɂ/c`6u{/* k!xB6{1Z;`ZrU j2%ΒsSJLD}}&s'Gk=Cc2S?jg࿹"$ȋFpNvnn&3{T _2;kWmsE.%wZ8uسGd2XUq(?v ZF˯؆0=T!;EPa*[=6JK>b]fBdXcׇ o:>Q./:W7<7xAqXDU2ev3< gWuzWWdPۘZc#[,I[ZPcF?IgXtLG=l2-,0 r&Q2e^;(4KPה!5߸*eއ!7l?aDžl^;|Pyl9~,m*51:/$Vl M ;t¯jZTh8w㬵 u0B#Pr+WdqMfFI3"BHT֪`mIi<*^w*3"u0? }kL$s 0yV)2nB/&YvcKJeNgP}Ku'J''L@Y D,ȃ&3x,Ƙƺ/hUS*Z{Brb ًNȖW0IWsaHw -- T^x?NoUi3 AzF"P ؘ42ϣE3 U5 |x!0Y}G:Nߩ*(FcNe$ө-zE@rg6IyW*tD?NU*Hď] YN'fq$)9XWq{[t0JiGgK([հ\BwJ1םNO\hӨj8 JNz/ЃwT٘A2Ne g HEpM瘉ɝ2|c xP٢q@p$)LǣkgjѸczc:#pQiP`ݍ0HuR?EkһG؈;XK"yPSkhM|:q.'C2aA,7R]2o";m%Vs9c 0e/UQλ$q C>zR3=jĖ=Z^dIyL1GJ%a* +J9GRISUd.^S? ܮ 2hC׭ >'K>/|wL:2 n٣FGYhi^+%iGM$承 G55jaBz{ Hhz(ZC?I)eNH:Op|[~ߏXH> 1FHùMXDl#Y-53)5fP k"~L[&" 9 Qrq0t3l-"aܫAUeRvVEV/ L"MGK~1Hs4^)K#*vMt5:wZ9벊]SYzZy/EM-µ8A8-~$M̎/`U[ݭ=}ٹi_^BM$!j*R/wT-]l@:N\sH xݢ .@f_ᩎg֥ z## Lp|=#J {Tz:ȽvtXdؽًWi7Pv[1&SO3@,«qNߪ8ʗ*Di,^=Jq,E7>Nu_hnƲgQqjX;jg̀ȍFKgmU@-}EwFاjӸo1z,iј9c(= V95?&KodŤ)|Xx6.Mf0S_oe9bklC9s2OZ`]Ӵ`8sס2_rF$L\$A\\gsEZ >e̲2Mg&%V/V- Ϋ8ޮ$x0e 1!* QVBYa: fJ n_%Z2(U~|,Uo15 j(KiܷYIm&$pechs%'hb_UjE&CaGEnx7-;B.|?o7Q_Գ3^""ME^{梩|4ѠA8BˊНM<zL6k?|yl,L 6X+ _T`V/, f3]w|k 2gqTqojII(ķU9whP3ڶE4"\=o1ţa4C4}QZqg !b=ϵ&ƯKnXvd8[c)f>90JqYTVw9Vo4,sǼtu~)ab>MIpӿ n`n2u=I! Or2E( 8n,c宥N"AeHMb%o69'ajQ 0tEf, Fg\!j?ߺhR'ddhk_#Kî=ϿOXhoLӺz/hc" qIHvTg9ag0\%vcWYRD7wir %PFpcXIf98Ywԑ'HSna K&FZq2D;aB7nLjWʛrgB#37yBUT_T\^'S;7Idhe>t,gB<"O7 55쯣ؒ~Qٔu_Ph^[fEyܷ,v$0$"/)zث={(j>;`R˘Y@3eEgUs}d3(."<|@AnLtX![}Ym8llZwLiJK^ 7\:$KjNCg8]f8D^LǾd6|xnWtXԝK IU"Tӹdv}?[Lwre7\ؗ5#JJ&3Fnnʤ/gZ掲aψb쪘,UQ(# Pp7dQ*uAkP!h%gu>I-<' Rw0ߺTlwlu .Vt>:`Kod f]RdgI"iIvX%lH}?L ~p V~LG+"Zag7N gТڮQT(VvXVY(opaF 0#֘NWҖw=h0ltN'-ħ eFp0T= %ƹS(js<0^ G=W'YB(-m&/ئd;S ZLR`ɴtR^EqKQ%yfHs˒܆T~{un.x 8&xءml;E8QivҢr&Jiw4Jң:s#WE9ITlWIBrA Nj_`mI% Ґ )_@M-!\l gC@Pz26UMc:I$qf,R;E|4EHγYDMH`adVt0`OcϊDA,gu7h}׋iJ23Ak/sf ƭT`(M ٲ~cwm0!oFON\?k2c*Ӳ8< P cB,idQ,镫荒A")6ȯU=>R9u\~eio]"\y60iϝ{V_*8Z/QCzv)4(o`T΍*,J+ `UwQK[ӗ7Ą-|Sy ?2姐Jz2U3!q Ip(- WÔt !qHTfFPt҉2?*8]rm9Dv{Mt)!@Vz5^DIۨCöړX\^!-h{y]-_Lz|dp2M h^v0Ga` y4Su2?|F8[ 2]#k]d:snutߖjbn}K\-VrkϲģFgٍ[M 1OPX&lS}%Ϋ6jIdž(&si t8p3JZf-kވ| 4-?u WްkC/uIJRF]3W+gtrLN6VlB330*t5)m-d5F ]ߵ~w~ٺ 2$b Zq6y"*EfhBK3iLE`c>אE!Ћ<υ{ 7YK&iSF'HXSS\\Ӛ׍xuY5P8lED>f6yă<01ng8L{jWV7AL74>aO%kB砓\s&ꋞp8G99 w-j7{$cZU";<]R:x8/f$3,o8N<ξY(|ZLoTh~\#ډ*+EH Gq7 0³zfDSReu=AލXeWaJ>KM(q!Uk A @v؋Cey7@MٕIj8ΐA6,E5G፹ܔ|U4U(I3Tn QPOFbV!LR"TB[# }̑7"Vc즹gxbo*D+@K y.2pD±. ۊ[&TBz>ZYlLlg^=CӀ!d0L=7\wtcG,70R{ ZUذ}r \x}~I:q62^trq& ulC/Y<4ՊF%@!v%I$eגʗ%':56ZC~W) 02.M{NqFXґ4<o^Qo=̻6Yj[쥞(.V޸KAPDjs]"OGL8~L黡;Pú@|Y[Xfv`@yj\l(1ܻx,a'*!gv)I~_y~\3?+2Qett9"q`WvZ׳`0e|(ZGQvم T] kjndOSrKS1c"8qh THtgc@XX)'!,ks葑r2 Gҡ1߲12#-9lұ8:N(tcqOؘl{C& 4EAIHfD Gd[`iZt{V¼N9^Mj[R wБOlIB% R,fQ`{Ӄ̣߫#7T(CSi)K8Bu(7է^7qv.8ucrõҬaTn-%dy#dǴnḯ[t)KA Yn-}>xZ(/eeUBN?1s[pf6wٽ\`'^:,9ucG]7IucvZ4[eU+ej*]Cucǃ /}^1dT%ϬB貆E;օ~ e#WS@߽ k# p42-Ys6므;hf[6 AƗ^Yd%U PW(0sЈЎ!foGUL:WmVݬmRJAy<$ZA&9خk4|zhS%\i 58Mt#'}n\} UTR]/ K=J>Uڟ/' pT}MIQ&ֱN,P["Sn4XE&B*J|CkySa^[,)l,.(Õ$;Zq][-fuad6jAY*S<K '؈Gh$%g$Y{4ٰmV^A.@BF:\ZevVx+wOȲE؀rty~68&W[CW%SmYa'sI2n\RnPV"Z[+XgrÓ^TW}`܀H]F4sXce;]`2S*ERW<&?l]UIJ7.jz3ÜU^k/^I=K2v5ir!j`Y"_ Z1]ci;ZnC{ +} eul$`jpneƳĢye3.rGɽHgkÏ60;UGJ4Ipx?3c _̓e;RG~0w(9ԥrO3`V F9y NZXLfa&`d[؝Jv5 hU.ehT"vy.:ͮy~hM]&onoC.yh)wC `b,Hd:(P+JQnsήDABe\37.Wzg6&U} p6SUgK94-c<&pJak"qZ^]3Vi<,"WA?\@;6 #H"b֒҇$Chy[46Lcp԰P3dҢ mIQKÆ~.sa]: }G@U܊ ۽iTp ZgG"2wȷ.uhR-FO,crMwd4vFKqoV E*#(ͲeHo9:]^~M<琪£Idr |)2F(:{_?Ea &35.C1J~DTN  gJ70wqi ѥQEe`-q8hj8׌py{.7n$xOq8ڱmKMd:; N-~7%snZ)-yW %0Ȟ\ϵUms OS/zT/^yHM A^FO$*SxRbo$IIC7g1:RջA^9c:T$Jйg'$!5q ,.`tk{5B[F䭼 PO@;"/sGBkjnصD OTľ77!-})zD|ۙRTkǝr3kDK:qNWFǪT*A3m@J',adhɰ㌭DY6~ x=xVh֜ز4elJNv[#vő2u,(¦0mIs\0RdL{C;DȌ b ToJv)I:0f"^ kZ<2iUE(,-L X|&A;<{Q90WA@  -jg Ȭu#x1b!b*n:Ȫ |*²¢:HxefB\7\;^kPFJZb~BYRuCӚZ5Nei.b|3KxA\"1iD-*>8I;aL ¦O̭z U~d!ϵq4/H4x1+7()uF2jPegT*E_PT#TilJR٣ZzZqWP":謇집zوi1K CdU:\TIgP( Z eABrM`q0-䗏謹hǑGrG.t`(O&A\ۛLj5>wyyډͽ^KTpυ,npՍG$.3S@Uz)`Ǣ~} 'TrRs0*NX#i 9i澩,M *KdfHdI7auxhཛྷFbZ7qfV~U+GBm8  4E"rD.=OBKyw͍x1I/%Qd'H[4>"M P!7$"eF-  gY {`yQNh"#> X+eܱǿtqlWYE0dQ'*׎Oo JlM$yn$05/ŪΧ/qkG95ېr$CV 41Y t:T}QsN3I`ޭYnΘ,hscKF-rAGjNzi6tra ȵ\CQ?KB$T͊#.0瀃A;$DkV'C`P\q~7R \W0I~D&I Ik+5=.%r)$ꡞH^j] ט]wSqð(eLEףs4s֓ǡ6<*ղ}n)hғJJjWؤ&36dv}"I0 J',݊y"Lh}2XLT]5+ub }ag%i66ciws67}@@0!Vkuݻ2.d;hQЃx/rJ,lk(z\|dSZYGi;$[\j $]kӵqu8QS-Fdh3](bf)9sCmOlؖLls7*HlʹSۛ5|l.AVosW$xYaI.d& Z꟫/a%%\?ςrBP_$Ѽ'4~/~SKl%9|%{SRwd1QWtg-Iw 1ᰫ8F(eցd2 z131CSML1[v5UTWȁf΄:z&PjU+B$%#da"pc?og LIp;;U%bi) mbeF~[u#Yܯʦ!{SPngNP0Z8|"3W,-'ŪDN1cu=o ^x0=xʋ)z;+mTD'T0[b֞~!Kst=b>[`γ"xaӨ^/z 'ZoԣFuH)$c`f$#[+81Eq<*b[7uRFS$bKH-V69HS{`z>&[cRc56) UPҋN(;8%+W>EkZePB o /5QZriYBVT͍F%ZgԪBɈ ̉JqR]Ikv2{ND[uǞ]oi 9=HA(G)&!& KXƺ/Y/̷| h-Ȕ p`]씩%ic\SKRi\H NU@,z+(3,d7ːيey(] ¥k-a)͸jrhŠeˡdSi0.&l G4 ԋzE-,Mpza(`Ae9tKaO$L,(PrsTay7^Y6, ,3jEN4?6@aV am7֯gyvx`(/c|| U롟8n 0 'jL6t(AjvGVw5S v#,lg1H{L y$vzN\{qw&qؔ%+8!*]uuE; G80#[w4%;԰n_K=%(a$6+:/r_oWHi$Maa-)*@{Q1m蠳LR%0EDzS`S=,y%w'dSgL I(?Ȕ0]nBʚ(IԨEۛ6'ܮ>"ީxtjr9̼-_Ϡ)"9/ 12ݤѲrQE<R3ס2@xU!7AlP,Ǫ&`"k$%b foYRߗPi 46]I9qe "vM)W=:ENtBa6sUXY3'&c/ْj6I W{ c Oőɘ,%@tz0t5ˍgS9eW5u&q!s- }nhȠT13f03]9O] FU`Pd!Mc N"^{L ԩD|+3vs 7y1 e0)+5\A/*M>/ ;ґֈDXuӭAt\n3u}g81&>4Aꪸ"1UU}}E ],.87Kbe|(C8Fi,Bva1 itj2'*e,,m.EK lBp ~[0&I$:Re,D`1E=i߰UpE;YPQcNۃL`4%3t/IZfS8 et"tcH;JQ{9V9BH˲c^j=Q'L6+ Kb2a0|{gFPɴJkc/dCECHZNM/X׋Eutoh Ri}(iR-IVd0Xzg"/VvIC!D#nЗ<@)R%K.|s-SA^RXcqWP^`#v- O iAI6E3JOV{J[N9$E$&re*Q#-եtE} (EeLP{Vdql{ l4՟cKx VRLXp*[qa@i \0Й(·"Nb {q]a@9x-Q8/4Qt1^ZC]}aP;Ssj*,[ ?C1dFCbWT/wX? $L\nxrnpX-@Mu9?q@}?j,euZt zU9a,j*ÊPBƔml 9(Z7?\"1 m4.Ui_&S~PU>jy %E ' ;CZ" I@)ֹ8k* d)Gf ]g< TbgF/AhaNMtKA6?j+8\7ֳ1gÅƃ  ;(N|Qe+fGyQ0Qy8]c-KTO0O"E HEcy 9aff$J,s9v|8vQYKnUxaz+脲ے5IRe[lTDESf*p)1+jQ Cy բ*OgMw1N|tX =*X#*m/̬$f>N&sUM[TWGeB{31ZW9l'U^ >kThv+F2覘 hi9@߿_v}qzk8כ˾íN [SSAR-2324kOֱUyɅH`!pfUό~s/K0 U ` Vafalelʙo62eulDiT.;v԰3żݓUI(Μ$-giN[l9ĕNY݁! 9w}Յ`X:?ZE\[m3[`DdPxHuf䊢T U,>_xqzXxkq؛\M5?EᕌUIGH|뱖f/*/< D#f> >“BLKqֈf͑oW@xPٛ _pL;n%;:.^ΕVvO(4VX9H:`vp>)DkE+닱 P! ]]2ǩwqcJP<{d1k%.P*Z ̐2V'Od/J^~ͬX@t:bi7˩r.zXn QpDntGN!E(Ul\9Ob -Z*UrjBE0݁iRX7n,G&kƭ/;$:n$.>f9G`mVtu%yxa*5Q\pYEK e؆y?wBP@nE fd7IXXM}jaZgb=?JiEyT+eּ,)mhZ-(X-˭X^8fUbw' ٢kEw% fS|*Z.Nΰ3w>*nQ$l>K gJy ZFי9h{A[8%:P9y$΂YM{[rѡ;y<ՔARRQa%Y22 w5Zk;66ajzY2 ˴~ 0_No!̢]~Z!`Γ_ itS;#2FL5%'Ӳ ˄j :K, C] Qqe,3s^+VXY`*:vZ>EbKGV"CEc#MQ:Q 2#ܘ.H6v 7pZs/ gPA/?rYֆ)|B vӉqp\_}(TG+¸vMؑl`r sDBG! 6KۨHNbpEθT"0:4RM)7)-kܗEcZ 7^ϭ_awex(؏ź/э1+*`"TOF)Zgm|!La0W-rdZM+TaO*R/l0j㞵ϩ5vIP$Z44q_&M%"p\hT-*~SÜһl☍$jCoқ&OP`Kvߓ$f!KDA9pnǟUA-}.+OtؔLAG6I<,wjMO"/쩩;r ffe7^"ѻc&vc6I2TIbF_Wn%Eޑ1;6z̤٤wӀUR=?W#7N]"MFSp(Y /+O.!rQX ’<5XEbL-Y,+0lqee$`p%KcACJ>Ej2m#'Q0e}i,O5Q{JuIS2lɺ{SD Ax(E.>R[d?q/mlMfR/jVр4ԟeIڞU}}79D,b"hYX6YF\cmL6Y#( vt%t$N3($Yno68Mo2ҳH-xH\!qOBq: rAR/< ̍ dF43,MRR ˙NI,=4ܙCtE H"p+zӹ9NV?~{Ձ~kw'#å[wmSe7%zP25n!RSDRR!TU J19R.ȧ BEY iO˿|CG@l4(ZڥwF NA Ezn-7DrN=T 0H޷xO&Й7n8%NOP?ORN1?:Q\8<0(kC\+b,a{!< Pj$ZF,/#*㪖pL󸀅Nftfnb.UlBLf * j]7IJŸ33yהRCBJTu*zØTXz\ : *6wo=~g\ZS\:LNDTOY݌/m%Tc7VuY/Wt@UL9{CD07FM'-|;bJ>kqf0 y%}v MIQ(I}7? k$F|Х*4k4AJؾ0gs^T-/='Om|'2 diѕMU1fxq9:+ĩՇ$1W;du- l@a]7ZRb0zl4զj:GFy^N*ѢN%|b+gnxn2%Ȕ#"76%zuK|Ux${._uv/2c2*qN+. XCUJm..{r,<3 ;yYR8q'ibf@o\UKh:֑^GnhzV"-p,uh~^[,O5d:8ަYyi%@'7m]tC?mi"ޒ 8і8QSP&vj'St`,';EJt)!Y|Oӫ,yn±k}oxQsRfrBp1aq^#-yZUe߹b R{kS-\qZuS\,OUYqU~r9"ȼa*uz~< ('=4g~^ oYkv-:yOnB1&ѴB/e rFkTW^ y2=w֕a̓l# b0$6&pgW i/; T-/l2![!uRi75[莑ވ8DO{gxK~+|L͍ $z!JC{)okfFU53v艡SDvt 9#ױ=ˇKTpEH]o!pڊ Wt6P"B壨{l $a#yw Pt%Չ-qӭ@f$tpDsmW.l%e=.&0H GYGX1._ bM[}P*.ޕ(5zM!l6 E.Wf3=rbz-b G\d#\Ub֡j) lf)Gx ol7Wæ<١Zvɀsv6ҜeL%?:&{=8l]Tn.x~Ώ؛_O vIkjK P<$="W\I1cd&4"= NYy4;z=De4^{a՛/D(8|F Ɇal^FײW1mؚy/`Y"ɀӀ5 $]LE.İ8An:Z&Y|&$qYx[ (٣yrFf}Ww-kN1 ؐ+ҝ{p:7tx^-̋4|TB8{UM*-֞yA¯ !R5٢Z.P(t*I:inFZ#!t2;pP¦ 9hJq)^{p`eX.ZSQW\&y1ߊץk5џNX9y[0)e b6y~_HWb)z=3z |CSTj_ߛ& a]tt 9ڷEkI ↓_n'k_4:?W;^eF@H= q D~OVG۟[u< ZV֘NVKb%InHO|kd,/=:B]:"%AIuV?L|gU P7;'Zڥ#S8^ Xo +6b+z0AWWb4ځ):BRy\D!\!޺SU+lfA?u0!X"r5Fb8OrxXc_yrGPkv$K4OT}6Si^rcPS韩teib\7# a$5!ߢj:Ʀнᶱ\߸]K|+ym6%Gd?(,%=>e8Qq WzHa2v{"icZJw9JCe")n#Õa6۵A[iϟ\5E3ҸHJ"qX$໯)uzqc <(Yf(3QQC[f4`D=DZm8(,(oKZWA f[ν#ʫ2o%S7O0agxHMQ_ahaFNdOU(d*|'=w)1QȞ0(Flggg$XvNAZXR4QUV Z|ɄR_y4F=.MHhH5(ϚdB yn:eЎGTqГTĊBSE* _S1IK0fa<'ƠI(¸h=}7lbSmyHM]GSkP0ߊ.k2 wSH}kqWi~֤4FmWǶ̟z,ғФK+HY>}0)#QU5)tRF ㇶ"0dp|xv/YP' ǸHA+kLXQvNl#[F<[9ڥ!qU7a_"8۩ǙuoXm_o3]/eٳInoKܞݍC[t@J'Ϟjm?7j|Y}>K F=-Zﵿw[[׷OmsSwvЇ?c=1_N:SV:eK%\x%DDxF<`ZB:ЗS MR\OrrU&$ Kޠg9uQ#ѥlOk7LxDۛjl RZ68&ᲕM =DFܼ>cG%4 c dnKbT0|4BiED{;"_ ;kR-ly~GA5\ ug6kckMTHbflv ")d,df=bT9(|d׳.bi] )zP=eml&ZFU_*ͶRQDFˋYW 7j> | &dk7D 怍c-yQ,THddg^B]ѝ{Vq8];Dk9'Xz.{D;,0-gsف?):E]̈́B4#,/L# ɋttHzl8oҳm;"b Y1-WxT~IclB λ3o:g_U"_uEr eA&Q2G^^02XA+yRkQ )3cɕڬMm#`U,M2oʢomνtv t6} Z.l>;o.a?כޟK 5h{ x\* 3ԫ$NjuWGu{#/-ݠ{pAa&+dtɱ2mӒpb(Qb.鍅ՇT:݁zHY{4{f_54.y&7WU&՛ͮ6py$(XE#%J^mӊ _-q҇II.O**RŅ]JWSQA 8Dݜf R=zD BoɅHwEU685w~ 履~ tZ$NW+B|Vg];VJX%M;vaM.%uqx'ox9,@7v]UOQfx\lղ4t dG}ß?Ԯu{E-kWnظ8;i/,rrdgj`:>V`]gUMJ ($ a:s^*H!dhjiص觡.)fiaz\WBC246ְCkN8 ~epD ygEz0w~Z uKh(N JosrlmA/W*Bi=/bvy#g {P+T 8@ZHZho[싓[, T4a &^fØ7]x["yljvDԻE%щy2=U w \W $|##:I?jg·zhQlloS6[޷0{I6^촥!]G4}*\/NGϲ9&'7j0gǩ[L Ys-G\49؏-ʊ8?$iXq ӣ9is47(F~AHJ@T7"]޿n<kyv-ȟuux<l "H]IR|rI5MdB#OŞ]^Fh&0HAfMXpsܸjaa}}!̴yΩ::YkElW}8=q8w&'m6lkh~&;hd6J.C"S$|N<28/}(\>%ak۝ j/R#7\ۣn@ <(9NDU}oK}.o󸸜M3[9>@aMb#?u4Rv  c프JOPQ ~}kSC?K)鉰;W͎+'H֬ `<::9k3;57(9+{5~/gT `aUVYǪnz +6CݛTzG|7DEdzI372ǣ#rGABim:y隅՜ͥ՜m3_J;*cRQ]MJ(7DD+{n+_,S* ƣHwkG{oVJTW8&TDsycOaG.XRض_9/zOڝ'-8ъW-M!#/A&OxTh /;$+*W*Qmz+$& ׉nђ*a^sXemdžAʏU_216it; 3CtRBSnSMAx7u,8LQԄo V1Uy[_Jx,*]P0_Gzy#jew)#}P)XVie(37^y]sXe4 ;1iM}fBz:/vJէ٢ιmrR(oLrg>/g*?v^>/^z(G124v_0{pޣ0@Dvzf!fRݚ^.+fM|C_by hHWj)^VS B:QTW~DX0vEvTR2g6Jv9ۋ)Rwu3+eol&5P>+&c=s#;yNp hLi&Hy >. `:sFqx |4I.3շ+C|gJH pCf)y#.U7Z9H%sPBT}E\6嚵\O qd( 27΃B!ŚHԖg@k FCU-é%|vĮJUԹ|66&a"S)1\H &ŇmJr\G t#5yzuLoX7ň4qz4nIS_> Txzљ>!xSRcJ#s4eH=Q4&NJ7#Y 3Z!IF(O5> AsCx*fMxa: eF b_rV`ְH6 k)OСwO(>}ԃS=nh8 ǫq&,Pyc jC Zƻfn,E |p"R?Z 鐤%I?l ~Aоv1C)Y#Y6tDXI [.-sS"-Lύ3t[D3OE|Pi^ x7s`sk .ů2v0gQZqDKVv|<95RVeM4'|W-O Usum'} S)&G m,/;`H5.DIؒyKn|16\9a=ʷ~yx͠v3X0{ul YPx0!zW2d{UL8^ػfN ſPDSaM r-ƹVDvwߺnt`eaҌhӸV`uۻOXִCHʀ'ck݅{ktH*MS?;p`oȫ<h@*ǙLC+hWЂ6aqf Km  1yXe,48+c74P]*-m  8,=Ǘvɞ)" /nr݅^en3DYei~>q¿KIj T..?xB:u)mDyyئESCi:ڧ(sz/o B@k@]\غWfkp؂|牙u*s'Pp/`Aw'$:?1쩱2Mf[پaym621 Ezۜ8y$+) u1u}G#'on]֏IxUї*4Tg@`ؗIRFr nO.5TsbX;eZW9IiVp5EZר]iTuJV_r#-Je\u372KUQ/ $psM4}QS:@^Lä3a$6)νZ*4^QU4'20BR 痨ϗĻ1>{MxAnb`q(64bB(9ьxjpQmIp|u?5Y[ ŶrpWlzK5 BmJTU>Z5b mQqM!!zƿ`m}vHkhxI,GZ$GXj+ͅX]sRYS7/fvp=傭uvD`k C ߕ?Ƀ|h|"vV>@q%NCɑ7,u˦dU-QZ f [oKA@~:: "Ui,Pkg4Gyp"; 8eeNl %܊a` xlʗgjng lu~WaH=InPIW(R!( 'ca1ЌY0ƣ/}{H3ZOE \bUQ\(q fU$Ms;̙B\&JVs3V"]BE>FO.R> }а_0lO2֩\@ 4 *BV~ww4"a h\-Zm\[*^?z`ݰҙB9A@$p҆TG'yI$JQ<@d82 wH5J%(QͭҨjiԌM&hh,"(r3$@ANYqfD;MhA 9ng<r#vԤtK9# GƋiq^r.⺯Ax.qH+bYU8g&Ǝ?2$*Ŝ' (1 u5ڷPqX2$\'y)՚p eIG) `IQG%0.$:` R'5-Mna_h 4e8+3`&'D k 1˜4 Fm9^:1*cbg *PZnAbf_ ;/ϙò 1# T\1Iڊ'eHc+ 2RmfG>4}t9?>:N. *,[?Wa7 x/Wf^*:`G觡'^"cOsvYUA;|#7<oM;>HTs/DSYLQ X\pqס+ S="h嗑** OpWQD) w.fguA"\_FvA?٭:|^]#a$uG_01#ϳ/w>=: ݹcf٤w?ivZ!b2$K0, BROLZnwOw,U5Ĕf4ѶC.*4_nɫʹ內e;Um4sqts_R;X_Bh"fimNhgs{ h}Nu=^*+Oe[C9` v=5?N\M^o|$gIay]p{9ꔰ76{9p DMֵgVUTk{>lU2ӷ|l3FbQ*?Ӥ^JG7$N D Kp.#kkjUĵ-{RǾM6MKf]KV: ´Yx Jeh~rXk[?VR[Z3_  `>qX~jXk\(hpl$1mPXn`"@ѧNd܌H?--isQPuRȋ8ꬬ,eq$FkaS17C|9rWxSH8]6xgh^EAק_~]s&a^ Fn(IQ FR͛zd hY Mcb Tb8+  R?ڭ>9pzT< ⌄J=6r5 n"EB5taV  }`3͐v3O sy -]QJS:\KҢXM< RylX頍Q$*"@Y(PX=󱜇iȂ|fgFIZ)a$c#ALC湓I7ZVgvZe2q |ī%hӚ>pk\7U~6셺{{ V&ޠ'\ EI9MLHq2X:QwY2a,ЏU{5eZyڐU'`[6u" S7':7lTl:>XhI>h'S5482jLNc}_SxT^}Un7bL(.>RCz)>XEX4^qيEcF3[,{@dk4# r~ ԇU| INx$yysN i*8,} ɊCR YKq3\x(7` UղÒ te|,` oxX$or~e]"i!>% nhGlf纵MqF;ͿK &w)*Rfo:]{ɏ1m}Ki]7'I}$ AK?M}Mt'ś=pnCa+4MNy׫jP߻.[H Q{(]ʻoqv4ymHюih]U욶ijYaTpr }?ק0Vț+F'?cZ!xt?"*+ Content creation Archives - TRUSTED REVIEWZ https://trustedreviewz.com/tag/content-creation/ Elevate Your Standards... Thu, 22 Feb 2024 09:32:37 +0000 en-US hourly 1 https://wordpress.org/?v=6.5.9 https://trustedreviewz.com/wp-content/uploads/2024/02/wp-1708575895810-150x150.png Content creation Archives - TRUSTED REVIEWZ https://trustedreviewz.com/tag/content-creation/ 32 32 Introducing SORA by OpenAI: The Future of Text-to-Video Transformation https://trustedreviewz.com/sora-openai-text-to-video/ https://trustedreviewz.com/sora-openai-text-to-video/#respond Thu, 22 Feb 2024 09:32:36 +0000 https://trustedreviewz.com/?p=176 The Future of Text-to-Video: SORA by Open AI OpenAI’s Sora is a testament to the rapid advancement of artificial intelligence. This remarkable text-to-video model seamlessly translates descriptive text prompts into captivating videos, forever altering the landscape of visual content creation. Let’s explore Sora’s workings, its potential impact, and how it compares to other image-generation powerhouses. […]

The post Introducing SORA by OpenAI: The Future of Text-to-Video Transformation appeared first on TRUSTED REVIEWZ.

]]>
The Future of Text-to-Video: SORA by Open AI

OpenAI’s Sora is a testament to the rapid advancement of artificial intelligence. This remarkable text-to-video model seamlessly translates descriptive text prompts into captivating videos, forever altering the landscape of visual content creation. Let’s explore Sora’s workings, its potential impact, and how it compares to other image-generation powerhouses.

Sora Text-to-Video Demos

Prompt: A stylish woman walks down a Tokyo street filled with warm glowing neon and animated city signage. She wears a black leather jacket, a long red dress, and black boots, and carries a black purse. She wears sunglasses and red lipstick. She walks confidently and casually. The street is damp and reflective, creating a mirror effect of the colorful lights. Many pedestrians walk about.

Prompt: A movie trailer featuring the adventures of the 30 year old space man wearing a red wool knitted motorcycle helmet, blue sky, salt desert, cinematic style, shot on 35mm film, vivid colors.

Prompt: Animated scene features a close-up of a short fluffy monster kneeling beside a melting red candle. The art style is 3D and realistic, with a focus on lighting and texture. The mood of the painting is one of wonder and curiosity, as the monster gazes at the flame with wide eyes and open mouth. Its pose and expression convey a sense of innocence and playfulness, as if it is exploring the world around it for the first time. The use of warm colors and dramatic lighting further enhances the cozy atmosphere of the image.

What is OpenAI’s Sora Model?

Sora is a groundbreaking AI model developed by OpenAI that specializes in generating videos from simple text descriptions. Here’s a breakdown of the key concepts:

1. The Power of Diffusion Models

  • Noise to Image: At its core, Sora is a diffusion model. It starts with a frame full of random noise (similar to the static on an old TV screen) and progressively removes that noise. Each step is guided by your text description until the noise transforms into your desired video scene.
  • Video Expertise: Sora excels because it can generate entire videos all at once, or even extend existing videos to make them longer. This gives it exceptional control over continuity and movement.

2. Capabilities Beyond Still Images

  • Movement and Dynamics: Sora understands the physics of the real world, allowing it to mimic realistic motion with objects and characters moving naturally.
  • Cinematic Flair: Imagine Sora as your cameraperson – it can perform camera movements like zooming, panning, and orbiting, adding a dynamic element to your videos.
  • Character Interactions: Sora can manage multiple characters and create believable interactions between them within a scene.
  • Expressing Emotions: Sora goes beyond simple visuals by adding emotive depth. Characters can express various emotions through their facial expressions and body language.

3. Sora’s Unique Strengths

  • Handling Longer Videos: Sora can create videos up to a minute long while retaining visual quality. This allows for more complex stories and narratives, which was a challenge with earlier video AI models.
  • Maintaining Consistency: Even when a subject briefly leaves the frame of a video, Sora ensures it looks the same when it pops back into view.
  • Future-Proof: Sora uses a transformer architecture known for its superior scaling. This means Sora will likely continue to improve as OpenAI provides it with more training data.

Why Does Sora Matter?

Sora is redefining how we create visual content. It offers:

  • Accessible Creativity: Soon, you won’t need specialized video editing skills to bring your ideas to life.
  • Boundless Possibilities: From entertainment to education, to marketing, Sora has the potential to transform how we communicate and tell stories visually.
  • Pushing AI Boundaries: Sora is a testament to the rapidly advancing field of artificial intelligence and hints at even more remarkable capabilities to come.

The Capabilities of Sora

Sora’s Key Capabilities

  1. Turning Text into Moving Images
  • Your Words, Your Vision: Sora’s primary strength lies in its ability to translate simple text prompts into vibrant videos. Whether you envision a playful scene with animals, a breathtaking landscape, or a futuristic cityscape, Sora can bring it to life.
  • Detail-Oriented: Sora isn’t limited to just broad actions or scenes. It can pick up on subtle details in your descriptions to create nuanced and visually rich videos.
  1. Cinematic Storytelling Potential
  • Camera Control: Sora understands the language of filmmaking, allowing it to incorporate camera movements like zooming, panning, tilting, and orbiting around a subject. This adds a dynamic flair to your videos.
  • Longer Videos: Unlike earlier AI video models, Sora is capable of generating longer, more cohesive video sequences (up to a minute long), paving the way for more complex storytelling.
  1. Complex Interactions and Dynamics
  • Multiple Characters: Sora can choreograph videos with multiple characters interacting in believable ways, opening the door to character-driven mini-narratives.
  • Realistic Movement: Understanding real-world physics is essential for natural-looking videos. Sora excels at simulating fluid, realistic motion for objects and characters.
  1. Emotional Expression
  • Going Beyond the Visual: Sora can portray a range of emotions through facial expressions and body language, deepening the emotional impact of your videos. This helps create more compelling and engaging characters.

What Makes Sora Stand Out

  • Consistency: Sora is skilled at maintaining visual consistency throughout longer video sequences. This means characters and objects retain their appearance, even if they briefly move out of the frame.
  • Adaptability: Sora’s flexibility allows it to handle prompts ranging from the mundane to the fantastical, giving you immense creative freedom.

Sora’s Potential Impact

Sora’s unique capabilities have the power to revolutionize various fields:

  • Entertainment: Imagine easily creating short films, personalized trailers, or visually striking music videos.
  • Marketing and Advertising: Instantly produce eye-catching ads, interactive product demos, and social media content that stands out from the crowd.
  • Education and Training: Complex topics come alive with engaging visual aids that enhance comprehension.

How to Get Access to and Use Sora

Current Status of Sora Access

  • Limited Access Phase: Sora is not yet widely available to the public. Currently, access is granted through a limited, invite-only system to select developers, researchers, and creatives.
  • No Public API: There isn’t a publicly available Sora API that developers could use to integrate Sora into their applications.

How to Potentially Get Access to Sora

  1. Request Access: Visit the official OpenAI Sora page (https://openai.com/sora) and look for an option to request access. You’ll likely need to complete a form detailing your intended use case and potentially your affiliation with a company or institution.
  2. Waitlist: Due to Sora being in high demand, you should be prepared to join a waitlist. OpenAI prioritizes use cases that align with its mission of safe and beneficial AI development.
  3. Evaluation: OpenAI may review your request based on criteria including:
    • The purpose of your project or research
    • The potential benefits and societal impact
    • Alignment with OpenAI’s safety and ethics guidelines
  4. Potential Approval: If your request fits their priorities and there’s availability, you might receive an invitation to use Sora.

Sora’s Content Limitations

OpenAI recognizes the enormous potential of AI tools like Sora, but also the potential for misuse if left unchecked. They implement strict content limitations to promote responsible use and minimize the risk of harm. These limitations are meant to prevent Sora from being used to generate:

  • Harmful or Misleading Content: This includes:
    • Violence or graphic content
    • Content that promotes hate, discrimination, or harassment.
    • Deepfakes or other content designed to deceive or impersonate.
  • Biased or Stereotypical Output: Sora should be used to create diverse and inclusive content, not reinforce harmful stereotypes.
  • Content Infringing on Intellectual Property: Sora should respect the existing copyright and intellectual property rights of others.

Specific Examples of Restricted Content

  • Violence: Videos depicting acts of physical harm or cruelty towards people or animals.
  • Sexual Content: Explicit videos or those that exploit, abuse, or endanger children.
  • Hate Speech: Content targeting groups based on race, religion, gender identity, sexual orientation, etc.
  • Celebrity Likeness: Generating videos of real people without their consent.
  • Misinformation: Videos intended to promote false and fabricated information.

OpenAI Sora API

The SORA API offers developers the flexibility to incorporate SORA’s text-to-video capabilities into their applications, enabling a wide range of custom uses and integrations. The API documentation provides comprehensive guidelines on how to effectively use SORA within various development environments.

Sora Pricing & Sora API’s Pricing

Pricing for SORA and its API is structured to accommodate different usage levels, from casual users to enterprise-scale applications. While specific pricing details may vary, OpenAI typically offers a tiered pricing model that includes free trial periods, pay-as-you-go options, and subscription packages.

Can I Use Sora AI on ChatGPT?

Integrating SORA with ChatGPT opens up fascinating possibilities for generating video content based on conversational inputs. While direct integration specifics can depend on the current OpenAI offerings, the combination of SORA’s video creation capabilities with ChatGPT’s conversational AI could redefine interactive media.

Are There Any Sora GPTs on the GPT Store?

The GPT Store, a marketplace for AI-driven applications, may feature SORA-based applications or tools. These offerings leverage SORA’s capabilities, providing users with innovative ways to generate video content directly from textual descriptions.

Sora Within the AI Ecosystem

  • Sora vs. ChatGPT: Sora and ChatGPT are distinct AI models from OpenAI. ChatGPT focuses on text generation and conversation, while Sora specializes in video creation.
  • Sora GPT Alternatives: As of now, there aren’t any GPT (Generative Pre-trained Transformer) models that precisely replicate Sora’s video generation abilities.

Sora VS Diffusion, Midjourney, and DALL·E 3

Let’s break down the distinctions between Sora, Diffusion models, Midjourney, and DALL-E 3 to understand their unique strengths and where they overlap:

Key Differentiators

  • Output Type:
    • Sora: The star here is its focus on video generation from text descriptions. This is its key difference from the others.
    • Midjourney, DALL-E 3, Diffusion Models: Primarily designed to generate still images. While some diffusion models can generate limited sequential images, they don’t offer Sora’s level of video generation control.
  • Generative Process:
    • Sora: Built on diffusion model principles, similar to its image-generating counterparts. It iteratively denoises images, guided by your text descriptions. However, it’s uniquely trained to create videos, understanding temporal flow and transitions.
    • Midjourney, DALL-E 3, Diffusion models (general): Also leverage diffusion models for their image generation prowess.

Capabilities and Use Cases

ToolStrengthsBest Suited For
SoraGenerating dynamic videos from text; potential for complex animations and scene transitions.Storytelling, educational videos, marketing content, and artistic expression.
MidjourneyArtistic styles, abstract concepts, highly detailed and imaginative imagery.Concept art, illustrations, unique visuals, surreal artwork.
DALL-E 3Photorealism, understanding nuanced details in text prompts, versatility.Product design visualizations, realistic images for advertising, concept mockups.
Diffusion Models (general)Open-source flexibility, ability to be fine-tuned for specific tasks, continuous development.Researchers, developers, projects requiring customization.

Where Things Get Interesting

  • Convergence: The lines between these models may blur in the future. Developers are already experimenting with video generation using diffusion models, though this is still in its early stages.
  • Hybrid AI Systems: Combining Sora with tools like ChatGPT could open up fascinating possibilities for dynamic storytelling where both visuals and text narratives are AI-generated.

Potential Alternatives to Sora

While Sora is leading the way, other notable AI video-generation tools exist:

  • Imagen Video (Google): A strong contender with impressive video generation capabilities.
  • Phenaki (Google): Allows for long-form video generation with text-guided editing.
  • Make-A-Video (Meta): Another promising player in the field of AI video generation.

Despite its advanced capabilities, SORA has limitations, including the resolution and length of generated videos, and the complexity of prompts it can accurately interpret. OpenAI continues to refine the model, aiming to expand its versatility and performance.

Safety Measures

OpenAI implements robust safety measures in SORA to prevent misuse and ensure responsible AI usage. These include content moderation, usage monitoring, and user education on ethical practices.

Future Plans

Potential Future Plans for Sora

  1. Enhanced Capabilities:
  • Higher Resolution and Quality: Expect Sora to generate increasingly sharper and more visually detailed videos as it continues to learn and refine its techniques.
  • Longer Videos: Sora might handle longer, more complex narratives with multiple scenes and intricate transitions.
  • Finer Control: Providing users with more granular tools to customize camera movements, character interactions, and other visual elements could become possible.
  1. Accessibility & Integrations
  • Public API Release: A Sora API would open the doors for developers to integrate its video generation power into their own applications and products. This could revolutionize many industries.
  • User-Friendly Interface: Sora might evolve to include a more approachable, streamlined interface, even for those without technical backgrounds.
  1. Advanced Features:
  • Audio Integration: Generating videos with accompanying soundtracks or narration synchronized to the visuals.
  • Text-Guided Video Editing: Imagine modifying existing videos using textual descriptions (e.g., “add a sunset in the background”).
  • Style Transfer: Applying artistic styles to videos, similar to how image-based style transfer works.
  1. Responsible AI Development:
  • Bias Mitigation: Continued refinement of techniques to combat biases and ensure Sora generates diverse and inclusive content.
  • Improved Safety Tools: Developing more robust tools to detect and flag potentially harmful uses of Sora.
  • Transparency: OpenAI maintains open communication about Sora’s capabilities, limitations, and safety measures.

The Path Ahead

Sora’s future is incredibly promising. It’s likely to be shaped by a combination of:

  • Technical Advancement: Progress in diffusion models and video generation techniques will unlock new possibilities.
  • User Feedback: As users get access to Sora, their experiences and needs will guide the model’s evolution.
  • Ethical Considerations: OpenAI’s commitment to safety and responsible use will determine how Sora is made available and the safeguards put in place.

Where to Stay Informed

  • OpenAI Blog: Keep an eye on the official OpenAI blog for announcements about Sora, new research, and API releases.
  • AI Research Communities: Follow AI research conferences and publications to stay updated on breakthroughs in video generation.

Conclusion

Sora marks a turning point in the democratization of visual storytelling. By turning simple text descriptions into captivating videos, it unlocks the power of creativity for everyone, regardless of traditional video editing expertise. This could have far-reaching impacts on entertainment, education, marketing, and beyond. However, Sora’s potential comes with a weighty responsibility. OpenAI’s proactive approach to content restrictions, combating misinformation, and addressing biases highlights the importance of ethical AI development.

As Sora continues to evolve, we can expect even higher-quality videos, greater control over visual elements, and potentially a publicly available API. One thing is certain: Sora represents a leap forward in artificial intelligence. It will shape the way we create, interact with, and understand video content. The future, where the line between imagination and reality continues to blur, is unfolding right before our eyes.

The post Introducing SORA by OpenAI: The Future of Text-to-Video Transformation appeared first on TRUSTED REVIEWZ.

]]>
https://trustedreviewz.com/sora-openai-text-to-video/feed/ 0