From 793875bf4d54e2b614d2d6898cb2198470f80574 Mon Sep 17 00:00:00 2001
From: MooglyGuy <MooglyGuy@users.noreply.github.com>
Date: Sun, 29 Aug 2021 17:49:03 +0200
Subject: [PATCH] -bgfx: Fixed UV rounding errors in fs_blit_palette16;
 Switched back to direct texture upload rather than CPU copy. (#8505) [Ryan
 Holtz]

---
 .../dx11/chains/misc/fs_blit_palette16.bin    | Bin 672 -> 1032 bytes
 .../dx9/chains/misc/fs_blit_palette16.bin     | Bin 481 -> 805 bytes
 .../essl/chains/misc/fs_blit_palette16.bin    | Bin 445 -> 1109 bytes
 .../glsl/chains/misc/fs_blit_palette16.bin    | Bin 422 -> 1040 bytes
 .../metal/chains/misc/fs_blit_palette16.bin   | Bin 958 -> 1462 bytes
 .../spirv/chains/misc/fs_blit_palette16.bin   | Bin 1660 -> 2608 bytes
 .../shaders/chains/misc/fs_blit_palette16.sc  |  11 +++++++--
 src/osd/modules/render/bgfxutil.cpp           |  21 +-----------------
 8 files changed, 10 insertions(+), 22 deletions(-)

diff --git a/bgfx/shaders/dx11/chains/misc/fs_blit_palette16.bin b/bgfx/shaders/dx11/chains/misc/fs_blit_palette16.bin
index 91a4aaf2d297d1dcbb455efaf9cdd00cb89a9cee..fb94aeda64add997a02b0e90ae6633a2e1199dad 100644
GIT binary patch
literal 1032
zcmaJ;O>0w85IrwxUh4~(G$2TY`smt4l7&haHZjd>gVI;>k`kp58wj*gq#ss9L{b$u
zioZc|8GnTf7rONixN%+b3q12?t_>m%nY_7k&YYQhCrZ}y-mA$^&}NZ8=|3Jm?H}(w
z8<tDVc#=lpq`&v*$uv^@H&Q7fG$eN{{4xbpgXe>t{j%qspZ|;RW`Wh-?Z%z@;bP_E
z+4G(4D+`~$UHvX<g7JrI^bhnO^j5OAvCpsM=jfYsG2u&2tU%vRBmI-S-~|GMzw#NL
zvm_*2{mnRObz1QT8u8t@jZU-Ih~xGu#xZ-!45A@DL%Or+p48~n+iT4Z#^YqY*?UKE
zneP+*;T*hsz)bW$OiQKizKr~!n4VoMJ$jchg9FWXyXR-Y<UA-MJ1OdJB(5@+dkm>J
z3AJt)$Vl!cM;<_P{=D-1(~yW7@!6M#FWA?srPxXx$T@7;uqqHNS<KvsoYbtQ<}}F5
zmy8$Ke8DI*HbCB=+6rfInyy~;Ml06=Uv%W|u3lp1b5bL*^@hyWww`{*?tB`Op}X@v
z<u`%1Uw~1gdkGWMcw=?Z3$1f|k@s_)ufEntuT|6DH*k1#QSnTT4JkkEH-HOO>#>#7
r9=`(9vl*u6I&$;4FXxiG%3SXLZa40&WS4ILGK*NWsQg>$6O{2AVS#MW

delta 244
zcmeC+Siq|27VN>z*!<}s0}!w>@R!DC=9R^lq*lZiXI7;e3NbKD40M+>U}9i!iEwg0
zlk2S4<b2F1ZkwJy)4J7#j6h8WOcS^5lMMkf1Ax3xAkN}oV2os7U~%hH0P+|Z7$;9;
zOs)s1b!A}aKE%M#;2bK-V8kE*mUR;lU;rA=Fs%l{XGnn177Pp=Fg0SVP_+#TU_BrL
eD9;6xcY~V6<N%UsaE4$I8*b<1)y&aM0t^7=8z-g!

diff --git a/bgfx/shaders/dx9/chains/misc/fs_blit_palette16.bin b/bgfx/shaders/dx9/chains/misc/fs_blit_palette16.bin
index 59f3c6dce2744de8d5b2b89a5f45c8ac766816fc..8f2023599c1d70c776a46e3d840991895d412876 100644
GIT binary patch
literal 805
zcmZuuyGjF55S_b^5KLepwt`z&Xd}D$0cs+WA{H70c1zHR3(<sZB4V|{#&Yc}Z#prJ
ze_%c#g&&}hLPRXIcRh1Ai4PnYcFviZGjnG*%f*bicz+_8NoL?Rt93^R5=<~|*UlV(
z1X~@?KRIR23;gq%lbcrl$+$V8je9bkpY;=^8x8V$9Dl||D(tT3=Kw!=Bsdcv#&kvW
zicbjFp(ZFrmx$S};W}K9MiO7PJ2+3}T?cTsAzV}JH}|NxWRn46HuZ4o_YKM_h2*`=
z<FWtne-42j-q299S?_n=1)d&_nt`|CIke^Pw;Dm?ux&5yEZN0Uxn!4*st2`}U1%IP
z{W=2IS$3(vjUS<tB_X>CU&BTCOUxUJ4-1SZHjgLIsIyrS$GB&R{Zx#ah|-p1o@0R+
zd7Ls8;#yMBFiLP5xDp~EXHko*0>>emKyfaHCUp(Mqe)X(r%I1|5>9YFBJ_jnCVgDT
znzudVpqWKHQuDDN>fg9WCjBOAN5Bsf9_BGukOl5Jr1Yf6LcWfjHpUu^M?7Edsl9YH
ik2%(64N@Aw&`9R6<^uM3kviQg9o$FF#{4LbiRcS&o032P

delta 238
zcmYj}u?oUK5JYEl8U+nO3#)uYQiyh9f?}_QrL|ZHHkJV!OT#HxUL)A~BmP7nSXuc2
z#<^4u_t>4e-MzWF@}~1yOAenLl*+tt6LkgPqt3Rdviz0<B#Utz&_3)3GsR*(GT_d#
zB#L_i7c2HwTVm$~4)!*nj>IAM^i_*YIuU!jS*tR}dMEw1z&Ajekdc1MFpR5Pekv$V
cXcnekd6fb$^6v}I!05LVGz1qTSZiPPeloTvfB*mh

diff --git a/bgfx/shaders/essl/chains/misc/fs_blit_palette16.bin b/bgfx/shaders/essl/chains/misc/fs_blit_palette16.bin
index d82fc7f6fbc2f989f4fd46935724130d9492bb40..2e8dc0a04b3ea8900600816d29ccf9d900db6b9d 100644
GIT binary patch
literal 1109
zcmb7D%}&BV5H1=n_24Vau~po(rG<lf(h%=Hz@~u$O<I=pj~2qom-Ol@_yo?bY<CNo
zAR(dbH}n0>Y$xw;H@%PVF9Sb)qo30vs*Rq(zx*Vp>o8IOZ%U5SJl;mmRa<pi(s)%}
zfVisX^V%?~Fl*w~0+#V&xrQp5`cTm+OIYSSnkS%#Ac{{J%Vvm|t9Z_`6!I`#CsF3T
zg7AJGzz_=1<g_?{;GJL>)VS!J#WP65cm<Yi?o5D$ZA3f|&_`4h5iKi<g`kpqP^4>&
zOg(-qpWzuSEi~MLW|-8CtvKch3k%se)JTS$M4(p6W?)HfQV+l*j%{lsxPsP*uW|UK
z<}9ie9$G2Uen5QEU|n0*D0@0{pfLxc!y&lPHg+J?j_wqxd4-&$FGWUL--v5Rbx~#!
z-UdrthXGj9wlBul*(V<%+)J+DcNF@Lcxu@{BJ~C-coNGCM#RDS@gLhhA$H|%;<@S{
z!V|cne1}P(a~HZg+vdMI`?JFtYr`3n4T_+C=d5nF$3w&TgN`>F!)$riBB7HkT)glX
Ulp8+mdCbioUR%5jd($+203qmD3jhEB

delta 177
zcmcc0v6q?GE!cybvH8<O1|VRb$g4OpPKv33abi{5WCcc^$pMTqS|$o5xdmm3Me&B#
zTnccuUb%v;f}x%Pve@Lm%#z9)8YQU}C8b5FMlK2(#qmInjzSrTO3u$ON-@yXtE!x=
v$E;(lsaH{11=5|K6Yo}(nC_gPlV1chQv+t3f)-F;GDy@w)0(T6i;DpOeU3H1

diff --git a/bgfx/shaders/glsl/chains/misc/fs_blit_palette16.bin b/bgfx/shaders/glsl/chains/misc/fs_blit_palette16.bin
index aadd6349d8c62d9b684ce34c28f2927c77c3c718..0ec3bc4fdb856bab99f928762847931ae605a4c5 100644
GIT binary patch
literal 1040
zcmb7D%TB^j5G@)ub>T0}vQ@k(rG<sM(h&E4z)b@e(xeYk+R{Q;`3ZiZzu*@*bM0+!
z!5AeZz|5R8=ggt=*O#lo+vkUYuc0yA@?2ELz@Rr(wtN#r-S4Y{`g|Mig>%tPT^Bq|
z%0G0*9p?SgFv=jS!(<JmSa?wKMH;1<b8l*87t6SCk*3)aJ4F(%(kzB;5N{%pxz8Z=
z{@g&}T|=(W#GiF~27c|1QDGV`Ar8U>EZaPoKn`5Mg(A-dFG`LxO)-x*=!d&hP~YJZ
zEX^0(fToO9wN1mTC=GH2nP~6{!DZKq(<NAnl~p5fnPb}$E)_K`PeXgGB76seyLO2L
zDh#V}t{tn1Lg(}~PG8PC3=?Wi`f|_7AA?m5ji;EW8e~cfMU@wsz+^2M?FcNT)suJX
z<wUoFTgBynJJ(|%s|r75)eb4hgys5E=HR*L!?w?e$0~$y)!`@b0Ao3RK%3BuyT4~&
z|98*v^e!{)E;F`6eq7zFRrUV#Ug&((lVziiJ)zbSpJ&1PiJmah+T=PToy~KH;U7)Y
F_ySAwK5zg4

delta 169
zcmbQhv5cA5E!cybvH8<O1|VRb$g4OpPKqg#abi`;WDZ82$wrLgX(kFKxdmm3Me&B#
zTnccuUb%v;f}x%PidaQu6;QM!wW6f7DAmYCL8CanATdWrK|=#Y49L+@D2oSD$@%$3
mDF&K)Rh5(9G3iXc%d8-5sGtRuOU}>9FEY@y=Bnl5VgLY&$1_L(

diff --git a/bgfx/shaders/metal/chains/misc/fs_blit_palette16.bin b/bgfx/shaders/metal/chains/misc/fs_blit_palette16.bin
index 58d7bbdde95332696e440c2acb608ed00cca6451..98684b4e571125449c238c77fca8b52f2332c451 100644
GIT binary patch
delta 515
zcmdnTzKxsLE!cybvH8<O1|Z;^$Q$d-T^e7KS`lBIS(R!a#K^z^L_ok_8lRb0hAbn1
zUB*y|QDI`OI@2wdiOVv$;3ip5Vw7Ssu-?S@gRwp>CqJ>oNFm<H$Wp;pAu~@8Y;JOX
zeo=~nf|f#jZb?pjsUDIu^eQT?xfFl^tP`lw*bt}@q+hQBRl?E~ECH0)RnRcf0~rU@
zlV79()}y6hpl7Nns)?k>L?PbTzyzpivLUmyoQ4L(Z9wH9vw?=h8(Es_K&+~qoX0F#
zZvb?Rrh+!aDMlJ_$D)LeUWE?ODg#YTWUGKaHa0Sc*qxcC0JH<>S{nsJpgOSUs=*}4
zn;>;SCuxA3rdL%7@dPl)G>l9qXE7`4LCmNHt53}-PQ|CaqI_~3vo1SS%jC~YHY_0H
MC-*QLFcmNW0Hr*Jo&W#<

delta 98
zcmdnSy^o#OE!cybvH8<O1|VRc$QvunUmBm8R~BEAS`lBIS(R!i#K<sltvZts^TcZz
yn~NA<F;4bmHD}V$nEZiRS|~G553DpfKffr&KvS=(a<TxcE~Dn;7FGkM00sc*T^;-Y

diff --git a/bgfx/shaders/spirv/chains/misc/fs_blit_palette16.bin b/bgfx/shaders/spirv/chains/misc/fs_blit_palette16.bin
index 44f887fa9458dd472fd154b0fb00662d22ea3597..5ad571399394ce754139809339683bf1f73c93be 100644
GIT binary patch
literal 2608
zcmZvcUvHFE5XMit?6#$7ON$sWA&aP}kh&$H5>pW|Z7)c~R5bBVHtW(hjcu1~Thefa
zAxgN?ub~MBz4AuCi~nL`j0OZjBkJ$>_6&L5IL-9TJTqtJIdjgt({r=KeV1;0laOi}
z-e|7RZ#LJKF3dN^K>CE!$VPK%^?W%p`u~y1u~g5I>YDkD)Ng_QJqiC@Yq^ms*}pcF
zQl)?YfIYUS!8E(oD(z&4=5VSN>&&USQxj|JXD5yvZMgORtqMhKBn_o)VCqV1Y1R0r
z=vDM{^lNmKeB(62s&K}XZqlF%THaTs-xI_N+H<XybIbExT=Ht^VBoLKZ?122=L3Jb
zz&Tb={$$`i`EP=sSzTJ_bXQ(l?zG>tPK`ftr8IodfdY3HJ>$O@bGHADO~SX5-nJMb
zcjvLV+??*V7Eg9M-LqSGwV2WQX1lZ8=@xUXlI_fjYi^s_DE2$UcNZ5niwZBd)>~xH
zFga)Sh{cfb_hI{L^qpBbu5Vn<#+_X`u5Wxi$DLt0u5aA;GYhkr*?xQ_us&?D(Gq4)
zu{t_K1@{IIXI1W_Zy)Q()A<dfBiZ+D?L`&i&P+e@oma%i>0+OK^pv;u9{$%@cZ2vQ
z>*zzg<3Uz^>y#z@2k9jbbF8C(U-o^+b^HUYiuIq*zW1aUe`?DtdKuUEUX;l99+ZeX
zcjW={kFeHJIUUUY(H{R8zVa~Vn_~4`=F0mFs_f18C;w3Sy@l`DydV4Pdmm!P-^sRn
zqGlT#c~O3Tm~}nGxc)-nZSB8+Z7sRP{tt`z#gm8N<a{rqVyZcw`3ma3*z*x~N*9ms
z0{s|EZlAC^7rDt@-?t+c{?+W8e=XbgG_LPkGT-=DtiGA3r|;Vm3m(%Zr*!-oxXaIX
zpqO(9#NWCp&>HKLsBb~vJ2`}U_YrSkd-oAPjP2d)%i%Lr{`zwM0+sIxx_B?&qTZGF
zAA7#b_$y4@J>R2Zsv`Xl^4x#Gwx&6+f_X<jqT-5qVvn*(+_UfCWMYqDTg!79kKD(x
z+c|zI?>_+Meq+(+C;aFWc|U`Rn;&_<<UF8sI`jS&+gxKYi|g3V!fzvHaRXaCX7L-g
zn5tMaW^ohSy5^kVjK14jXw+Q5wytOYJ;#F2?Tq=Yy`MSW!M3Kc*TJS(Cury!@LjwC
z<{dY{#LW+$e`Va5QbB{~U2Nl*(K!3x*z&xB#@YYD7LS@9R!mha9X0!KtSk5>y!^f@
z*xrNR-EA1`z*hcb2FA9bV!nCp=u^X%x9a!7a<9(-j`1@0LHu6sLtx@jb33+}s<;IA
zVI1q4^9LP+lj|*R%Wq%?+cWxYnlInDlV`CHgZnMUog4!bcVE22o!H_^U}G5*4s1MQ
op8Yiaz2jYA?l<4KGnK!w-T21+)?Z-G{@#e)<&@7ACvBvE0d0ZuI{*Lx

literal 1660
zcmZ9L+iFum7=<T^NvlCy6%{L#)Or9DYlJEVQPE1%i-L#r;9c0JX$U6SaVIIks|4{?
zJl=S~ClK#MpTSoUFT4?4-(&_hneumLt@Y1;nw_=sdN%Rm_b;QBGBf>3qrG40)drPr
z<9Y4sF|i3GlkUP6O%jlt19yUEQFq3;kkS3*;uQXe%(VFluXR*6h>$fIyO%b~8%y2Z
z?$YwLBKnbW1gp)MgXVxZtQ9oc=q*`C){~87v+_q*BArqut^JP@q{YVXrIl|{nk~B=
zv^vdNBq+a(x#0QD+Mw5uYM#Gjd9^j+U-0~d|Caczc4IG$T6dda^(k>OstlerF-+x7
znVqp@>!7!Kv$7Ti_0=$pc9GKhqu>2XHEf2Foe>=U+lQ&*S+w_=$CLWrz(%+i^a6R8
zGzI!TD`R=Qpfj%trvJDOM(1tOnL}I$qaSxV^N8zUbiRdkRp$Gr4pvDDoR7~8;`hPa
zPsqs6T(YtmhndH`jD^lDz}|o4ur}40w0RbklX}QGr~mpq%}FtQH#ym9eQVR&!NiF#
zk54HVb>KrB@L`Af)^ftW58~f&*!PWoUObCj=wRl}^8U=1<#p!CIwAin+H*2$$~(NQ
zGwY1{S<y~xH0pdLGiJD2&^u6jSq6vA%b$0k2Ws2aPG4~32J;qh=#SkQe;?m|q-J;<
zy(yl$(76w8InHC=4j=rx?o3_iVBXT}L!Egu9}hfo+;#OsUCf<@-(NBO{%?w9{^w-O
z9?VQG%9yp+^ExwYuV2=gS%b;>Oh&z6YV6CXZ(CWI^H&)$$+IJlJKtpR_}<qGaPO52
z9;0UZdntdk*D`e8<&HRJ{6@yyc~@V{TgUsD?}XtnRy^i=VdCQBYyGY>{pDr8)*m{<
z<Nr{(r?h{N!LvN~QD-=Ec<z(q=;wiM()(u__wh&P+n`UhFEVsydP}dxJHq`_)D;_f
G*8B(mYn6lm

diff --git a/src/osd/modules/render/bgfx/shaders/chains/misc/fs_blit_palette16.sc b/src/osd/modules/render/bgfx/shaders/chains/misc/fs_blit_palette16.sc
index e87f8a76315..98f13186663 100644
--- a/src/osd/modules/render/bgfx/shaders/chains/misc/fs_blit_palette16.sc
+++ b/src/osd/modules/render/bgfx/shaders/chains/misc/fs_blit_palette16.sc
@@ -15,7 +15,14 @@ uniform vec4 u_inv_tex_size1;
 
 void main()
 {
-	vec4 srcpix = texture2D(s_tex, v_texcoord0.xy);
-	vec2 palette_uv = (srcpix.bg * vec2(256.0, 256.0)) * u_inv_tex_size1.xy;
+	vec2 original_uv = v_texcoord0.xy * u_tex_size0.xy;
+	float mod_val = mod(original_uv.x, 2.0);
+	vec2 rounded_uv = vec2(original_uv.x - mod_val, original_uv.y);
+	vec4 srcpix = texture2D(s_tex, rounded_uv * u_inv_tex_size0.xy + vec2(u_inv_tex_size0.x, 0.0));
+
+	vec2 palette_uv = (srcpix.ra * vec2(256.0, 256.0)) * u_inv_tex_size1.xy;
+	if (mod_val < 1.0)
+		palette_uv = (srcpix.bg * vec2(256.0, 256.0)) * u_inv_tex_size1.xy;
+
 	gl_FragColor = vec4(texture2D(s_pal, palette_uv).rgb, 1.0) * v_color0;
 }
diff --git a/src/osd/modules/render/bgfxutil.cpp b/src/osd/modules/render/bgfxutil.cpp
index 88d2fbf399a..a147e8c2c1c 100644
--- a/src/osd/modules/render/bgfxutil.cpp
+++ b/src/osd/modules/render/bgfxutil.cpp
@@ -22,6 +22,7 @@ const bgfx::Memory* bgfx_util::mame_texture_data_to_bgfx_texture_data(bgfx::Text
 	switch (src_format)
 	{
 		case PRIMFLAG_TEXFORMAT(TEXFORMAT_YUY16):
+		case PRIMFLAG_TEXFORMAT(TEXFORMAT_PALETTE16):
 			dst_format = bgfx::TextureFormat::BGRA8;
 			convert_stride = 2;
 			out_pitch = rowpixels * 2;
@@ -29,26 +30,6 @@ const bgfx::Memory* bgfx_util::mame_texture_data_to_bgfx_texture_data(bgfx::Text
 
 			data = bgfx::copy(base, info.storageSize);
 			break;
-		case PRIMFLAG_TEXFORMAT(TEXFORMAT_PALETTE16):
-		{
-			dst_format = bgfx::TextureFormat::BGRA8;
-			convert_stride = 1;
-			out_pitch = rowpixels * 4;
-			bgfx::calcTextureSize(info, rowpixels / convert_stride, height, 1, false, false, 1, dst_format);
-
-			uint16_t *src = (uint16_t *)base;
-			uint16_t *dst_data = new uint16_t[rowpixels * 2 * height];
-			uint16_t *dst = dst_data;
-			for (int i = 0; i < rowpixels * height; i++, src++)
-			{
-				*dst++ = *src;
-				*dst++ = 0;
-			}
-
-			data = bgfx::copy(dst_data, info.storageSize);
-			delete [] dst_data;
-			break;
-		}
 		case PRIMFLAG_TEXFORMAT(TEXFORMAT_ARGB32):
 		case PRIMFLAG_TEXFORMAT(TEXFORMAT_RGB32):
 			dst_format = bgfx::TextureFormat::BGRA8;