diff --git a/AUTHORS b/AUTHORS index ab733de..ba49079 100644 --- a/AUTHORS +++ b/AUTHORS @@ -30,7 +30,8 @@ Glenn Mabutt, Paul Kreiner, Erik Kunze, Mike Frysinger, Martin Waitz, Mark McLoughlin, Paul Fox, Juan Jose Costello, Andre Leiadella, Alberto Lusiani, Malvina Mazin, Dave Stuart, Rohit Kumar, Donald Dugger, Steven Carr, Uwe Völker, Charles Coffing, Guillaume Rousse, -Alessandro Praduroux, Brad Hards, Timo Ketola, and Christian Ehrlicher. +Alessandro Praduroux, Brad Hards, Timo Ketola, Christian Ehrlicher, and +Noriaki Yamazaki. Probably I forgot quite a few people sending a patch here and there, which really made a difference. Without those, some obscure bugs still would diff --git a/ChangeLog b/ChangeLog index 3cadb15..81951cf 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,9 @@ +2008-01-28 Noriaki Yamazaki + * rfb/rfbproto.h, libvncserver/rfbserver.c, libvncserver/scale.c, + libvncserver/zrle.c, libvncserver/zrleencodetemplate.c, + libvncserver/zywrletemplate.c: add (server-side) ZYWRLE support, + and fix a few endian/scale errors + 2008-01-27 Christian Ehrlicher * CMakeLists, rfb/rfbconfig.h.cmake, rfb/rfbint.h.cmake: support CMake diff --git a/libvncserver/rfbserver.c b/libvncserver/rfbserver.c index b7507bd..66eac19 100644 --- a/libvncserver/rfbserver.c +++ b/libvncserver/rfbserver.c @@ -962,6 +962,7 @@ rfbSendSupportedEncodings(rfbClientPtr cl) #endif #ifdef LIBVNCSERVER_HAVE_LIBZ rfbSendSupporteddEncodings_SendEncoding(cl, rfbEncodingZRLE); + rfbSendSupporteddEncodings_SendEncoding(cl, rfbEncodingZYWRLE); #endif rfbSendSupporteddEncodings_SendEncoding(cl, rfbEncodingUltra); rfbSendSupporteddEncodings_SendEncoding(cl, rfbEncodingUltraZip); @@ -1908,6 +1909,7 @@ rfbProcessClientNormalMessage(rfbClientPtr cl) #ifdef LIBVNCSERVER_HAVE_LIBZ case rfbEncodingZlib: case rfbEncodingZRLE: + case rfbEncodingZYWRLE: #ifdef LIBVNCSERVER_HAVE_LIBJPEG case rfbEncodingTight: #endif @@ -2806,6 +2808,7 @@ rfbSendFramebufferUpdate(rfbClientPtr cl, #endif #ifdef LIBVNCSERVER_HAVE_LIBZ case rfbEncodingZRLE: + case rfbEncodingZYWRLE: if (!rfbSendRectEncodingZRLE(cl, x, y, w, h)) goto updateFailed; break; diff --git a/libvncserver/scale.c b/libvncserver/scale.c index 67de35f..55598fa 100644 --- a/libvncserver/scale.c +++ b/libvncserver/scale.c @@ -151,6 +151,10 @@ void rfbScaledScreenUpdateRect(rfbScreenInfoPtr screen, rfbScreenInfoPtr ptr, in h1 = h0; rfbScaledCorrection(screen, ptr, &x1, &y1, &w1, &h1, "rfbScaledScreenUpdateRect"); + x0 = ScaleX(ptr, screen, x1); + y0 = ScaleY(ptr, screen, y1); + w0 = ScaleX(ptr, screen, w1); + h0 = ScaleY(ptr, screen, h1); bitsPerPixel = screen->bitsPerPixel; bytesPerPixel = bitsPerPixel / 8; @@ -197,8 +201,8 @@ void rfbScaledScreenUpdateRect(rfbScreenInfoPtr screen, rfbScreenInfoPtr ptr, in /* Get the totals for rgb from the source grid... */ for (w = 0; w < areaX; w++) { for (v = 0; v < areaY; v++) { - srcptr2 = &srcptr[(((x * areaX) + v) * bytesPerPixel) + - (w * screen->paddedWidthInBytes)]; + srcptr2 = &srcptr[(((x * areaX) + w) * bytesPerPixel) + + (v * screen->paddedWidthInBytes)]; pixel_value = 0; diff --git a/libvncserver/zrle.c b/libvncserver/zrle.c index d72993e..5481893 100644 --- a/libvncserver/zrle.c +++ b/libvncserver/zrle.c @@ -40,18 +40,50 @@ #define EXTRA_ARGS , rfbClientPtr cl +#define ENDIAN_LITTLE 0 +#define ENDIAN_BIG 1 +#define ENDIAN_NO 2 #define BPP 8 +#define ZYWRLE_ENDIAN ENDIAN_NO +#include +#undef BPP +#define BPP 15 +#undef ZYWRLE_ENDIAN +#define ZYWRLE_ENDIAN ENDIAN_LITTLE +#include +#undef ZYWRLE_ENDIAN +#define ZYWRLE_ENDIAN ENDIAN_BIG #include #undef BPP #define BPP 16 +#undef ZYWRLE_ENDIAN +#define ZYWRLE_ENDIAN ENDIAN_LITTLE +#include +#undef ZYWRLE_ENDIAN +#define ZYWRLE_ENDIAN ENDIAN_BIG #include #undef BPP #define BPP 32 +#undef ZYWRLE_ENDIAN +#define ZYWRLE_ENDIAN ENDIAN_LITTLE +#include +#undef ZYWRLE_ENDIAN +#define ZYWRLE_ENDIAN ENDIAN_BIG #include #define CPIXEL 24A +#undef ZYWRLE_ENDIAN +#define ZYWRLE_ENDIAN ENDIAN_LITTLE +#include +#undef ZYWRLE_ENDIAN +#define ZYWRLE_ENDIAN ENDIAN_BIG #include #undef CPIXEL #define CPIXEL 24B +#undef ZYWRLE_ENDIAN +#define ZYWRLE_ENDIAN ENDIAN_LITTLE +#include +#undef ZYWRLE_ENDIAN +#define ZYWRLE_ENDIAN ENDIAN_BIG #include #undef CPIXEL #undef BPP @@ -80,6 +112,20 @@ rfbBool rfbSendRectEncodingZRLE(rfbClientPtr cl, int x, int y, int w, int h) rfbZRLEHeader hdr; int i; + if (cl->preferredEncoding == rfbEncodingZYWRLE) { + if (cl->tightQualityLevel < 0) { + zywrle_level = 1; + } else if (cl->tightQualityLevel < 3) { + zywrle_level = 2; + } else if (cl->tightQualityLevel < 6) { + zywrle_level = 3; + } else { + zywrle_level = 1; + } + } else { + zywrle_level = 0; + } + if (!cl->zrleData) cl->zrleData = zrleOutStreamNew(); zos = cl->zrleData; @@ -89,11 +135,21 @@ rfbBool rfbSendRectEncodingZRLE(rfbClientPtr cl, int x, int y, int w, int h) switch (cl->format.bitsPerPixel) { case 8: - zrleEncode8( x, y, w, h, zos, zrleBeforeBuf, cl); + zrleEncode8NE(x, y, w, h, zos, zrleBeforeBuf, cl); break; case 16: - zrleEncode16(x, y, w, h, zos, zrleBeforeBuf, cl); + if (cl->format.greenMax > 0x1F) { + if (cl->format.bigEndian) + zrleEncode16BE(x, y, w, h, zos, zrleBeforeBuf, cl); + else + zrleEncode16LE(x, y, w, h, zos, zrleBeforeBuf, cl); + } else { + if (cl->format.bigEndian) + zrleEncode15BE(x, y, w, h, zos, zrleBeforeBuf, cl); + else + zrleEncode15LE(x, y, w, h, zos, zrleBeforeBuf, cl); + } break; case 32: { @@ -107,18 +163,24 @@ rfbBool rfbSendRectEncodingZRLE(rfbClientPtr cl, int x, int y, int w, int h) cl->format.blueShift > 7); if ((fitsInLS3Bytes && !cl->format.bigEndian) || - (fitsInMS3Bytes && cl->format.bigEndian)) - { - zrleEncode24A(x, y, w, h, zos, zrleBeforeBuf, cl); + (fitsInMS3Bytes && cl->format.bigEndian)) { + if (cl->format.bigEndian) + zrleEncode24ABE(x, y, w, h, zos, zrleBeforeBuf, cl); + else + zrleEncode24ALE(x, y, w, h, zos, zrleBeforeBuf, cl); } else if ((fitsInLS3Bytes && cl->format.bigEndian) || - (fitsInMS3Bytes && !cl->format.bigEndian)) - { - zrleEncode24B(x, y, w, h, zos, zrleBeforeBuf, cl); + (fitsInMS3Bytes && !cl->format.bigEndian)) { + if (cl->format.bigEndian) + zrleEncode24BBE(x, y, w, h, zos, zrleBeforeBuf, cl); + else + zrleEncode24BLE(x, y, w, h, zos, zrleBeforeBuf, cl); } - else - { - zrleEncode32(x, y, w, h, zos, zrleBeforeBuf, cl); + else { + if (cl->format.bigEndian) + zrleEncode32BE(x, y, w, h, zos, zrleBeforeBuf, cl); + else + zrleEncode32LE(x, y, w, h, zos, zrleBeforeBuf, cl); } } break; @@ -138,7 +200,7 @@ rfbBool rfbSendRectEncodingZRLE(rfbClientPtr cl, int x, int y, int w, int h) rect.r.y = Swap16IfLE(y); rect.r.w = Swap16IfLE(w); rect.r.h = Swap16IfLE(h); - rect.encoding = Swap32IfLE(rfbEncodingZRLE); + rect.encoding = Swap32IfLE(cl->preferredEncoding); memcpy(cl->updateBuf+cl->ublen, (char *)&rect, sz_rfbFramebufferUpdateRectHeader); diff --git a/libvncserver/zrleencodetemplate.c b/libvncserver/zrleencodetemplate.c index 7c89c40..7d19766 100644 --- a/libvncserver/zrleencodetemplate.c +++ b/libvncserver/zrleencodetemplate.c @@ -43,17 +43,37 @@ #define __RFB_CONCAT2E(a,b) __RFB_CONCAT2(a,b) #endif +#ifndef __RFB_CONCAT3E +#define __RFB_CONCAT3(a,b,c) a##b##c +#define __RFB_CONCAT3E(a,b,c) __RFB_CONCAT3(a,b,c) +#endif + +#undef END_FIX +#if ZYWRLE_ENDIAN == ENDIAN_LITTLE +# define END_FIX LE +#elif ZYWRLE_ENDIAN == ENDIAN_BIG +# define END_FIX BE +#else +# define END_FIX NE +#endif + #ifdef CPIXEL #define PIXEL_T __RFB_CONCAT2E(zrle_U,BPP) #define zrleOutStreamWRITE_PIXEL __RFB_CONCAT2E(zrleOutStreamWriteOpaque,CPIXEL) -#define ZRLE_ENCODE __RFB_CONCAT2E(zrleEncode,CPIXEL) -#define ZRLE_ENCODE_TILE __RFB_CONCAT2E(zrleEncodeTile,CPIXEL) +#define ZRLE_ENCODE __RFB_CONCAT3E(zrleEncode,CPIXEL,END_FIX) +#define ZRLE_ENCODE_TILE __RFB_CONCAT3E(zrleEncodeTile,CPIXEL,END_FIX) #define BPPOUT 24 +#elif BPP==15 +#define PIXEL_T __RFB_CONCAT2E(zrle_U,16) +#define zrleOutStreamWRITE_PIXEL __RFB_CONCAT2E(zrleOutStreamWriteOpaque,16) +#define ZRLE_ENCODE __RFB_CONCAT3E(zrleEncode,BPP,END_FIX) +#define ZRLE_ENCODE_TILE __RFB_CONCAT3E(zrleEncodeTile,BPP,END_FIX) +#define BPPOUT 16 #else #define PIXEL_T __RFB_CONCAT2E(zrle_U,BPP) #define zrleOutStreamWRITE_PIXEL __RFB_CONCAT2E(zrleOutStreamWriteOpaque,BPP) -#define ZRLE_ENCODE __RFB_CONCAT2E(zrleEncode,BPP) -#define ZRLE_ENCODE_TILE __RFB_CONCAT2E(zrleEncodeTile,BPP) +#define ZRLE_ENCODE __RFB_CONCAT3E(zrleEncode,BPP,END_FIX) +#define ZRLE_ENCODE_TILE __RFB_CONCAT3E(zrleEncodeTile,BPP,END_FIX) #define BPPOUT BPP #endif @@ -64,12 +84,20 @@ static const int bitsPerPackedPixel[] = { 0, 1, 2, 2, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4 }; +int zywrle_level; +int zywrleBuf[rfbZRLETileWidth*rfbZRLETileHeight]; + static zrlePaletteHelper paletteHelper; #endif /* ZRLE_ONCE */ void ZRLE_ENCODE_TILE (PIXEL_T* data, int w, int h, zrleOutStream* os); +#if BPP!=8 +#define ZYWRLE_ENCODE +#include "zywrletemplate.c" +#endif + static void ZRLE_ENCODE (int x, int y, int w, int h, zrleOutStream* os, void* buf EXTRA_ARGS @@ -92,7 +120,7 @@ static void ZRLE_ENCODE (int x, int y, int w, int h, } -void ZRLE_ENCODE_TILE (PIXEL_T* data, int w, int h, zrleOutStream* os) +void ZRLE_ENCODE_TILE(PIXEL_T* data, int w, int h, zrleOutStream* os) { /* First find the palette and the number of runs */ @@ -144,6 +172,11 @@ void ZRLE_ENCODE_TILE (PIXEL_T* data, int w, int h, zrleOutStream* os) estimatedBytes = w * h * (BPPOUT/8); /* start assuming raw */ +#if BPP!=8 + if (zywrle_level > 0 && !(zywrle_level & 0x80)) + estimatedBytes >>= zywrle_level; +#endif + plainRleBytes = ((BPPOUT/8)+1) * (runs + singlePixels); if (plainRleBytes < estimatedBytes) { @@ -253,14 +286,24 @@ void ZRLE_ENCODE_TILE (PIXEL_T* data, int w, int h, zrleOutStream* os) /* raw */ -#ifdef CPIXEL - PIXEL_T *ptr; - for (ptr = data; ptr < data+w*h; ptr++) { - zrleOutStreamWRITE_PIXEL(os, *ptr); +#if BPP!=8 + if (zywrle_level > 0 && !(zywrle_level & 0x80)) { + ZYWRLE_ANALYZE( data, data, w, h, w, zywrle_level, zywrleBuf ); + zywrle_level |= 0x80; + ZRLE_ENCODE_TILE( data, w, h, os ); + zywrle_level &= 0x7F; } + else +#endif + { +#ifdef CPIXEL + PIXEL_T *ptr; + for (ptr = data; ptr < data+w*h; ptr++) + zrleOutStreamWRITE_PIXEL(os, *ptr); #else - zrleOutStreamWriteBytes(os, (zrle_U8 *)data, w*h*(BPP/8)); + zrleOutStreamWriteBytes(os, (zrle_U8 *)data, w*h*(BPP/8)); #endif + } } } } @@ -269,4 +312,5 @@ void ZRLE_ENCODE_TILE (PIXEL_T* data, int w, int h, zrleOutStream* os) #undef zrleOutStreamWRITE_PIXEL #undef ZRLE_ENCODE #undef ZRLE_ENCODE_TILE +#undef ZYWRLE_ENCODE_TILE #undef BPPOUT diff --git a/libvncserver/zywrletemplate.c b/libvncserver/zywrletemplate.c new file mode 100644 index 0000000..2a096da --- /dev/null +++ b/libvncserver/zywrletemplate.c @@ -0,0 +1,815 @@ + +/******************************************************************** + * * + * THIS FILE IS PART OF THE 'ZYWRLE' VNC CODEC SOURCE CODE. * + * * + * USE, DISTRIBUTION AND REPRODUCTION OF THIS LIBRARY SOURCE IS * + * GOVERNED BY A FOLLOWING BSD-STYLE SOURCE LICENSE. * + * PLEASE READ THESE TERMS BEFORE DISTRIBUTING. * + * * + * THE 'ZYWRLE' VNC CODEC SOURCE CODE IS (C) COPYRIGHT 2006 * + * BY Hitachi Systems & Services, Ltd. * + * (Noriaki Yamazaki, Research & Developement Center) * * + * * + ******************************************************************** +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions +are met: + +- Redistributions of source code must retain the above copyright +notice, this list of conditions and the following disclaimer. + +- Redistributions in binary form must reproduce the above copyright +notice, this list of conditions and the following disclaimer in the +documentation and/or other materials provided with the distribution. + +- Neither the name of the Hitachi Systems & Services, Ltd. nor +the names of its contributors may be used to endorse or promote +products derived from this software without specific prior written +permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION +OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + ********************************************************************/ + +/* #define ZYWRLE_ENCODE */ +/* #define ZYWRLE_DECODE */ +#define ZYWRLE_QUANTIZE + +/* + [References] + PLHarr: + Senecal, J. G., P. Lindstrom, M. A. Duchaineau, and K. I. Joy, "An Improved N-Bit to N-Bit Reversible Haar-Like Transform," Pacific Graphics 2004, October 2004, pp. 371-380. + EZW: + Shapiro, JM: Embedded Image Coding Using Zerotrees of Wavelet Coefficients, IEEE Trans. Signal. Process., Vol.41, pp.3445-3462 (1993). +*/ + + +/* Template Macro stuffs. */ +#undef ZYWRLE_ANALYZE +#undef ZYWRLE_SYNTHESIZE +#define ZYWRLE_ANALYZE __RFB_CONCAT3E(zywrleAnalyze,BPP,END_FIX) +#define ZYWRLE_SYNTHESIZE __RFB_CONCAT3E(zywrleSynthesize,BPP,END_FIX) + +#define ZYWRLE_RGBYUV __RFB_CONCAT3E(zywrleRGBYUV,BPP,END_FIX) +#define ZYWRLE_YUVRGB __RFB_CONCAT3E(zywrleYUVRGB,BPP,END_FIX) +#define ZYWRLE_YMASK __RFB_CONCAT2E(ZYWRLE_YMASK,BPP) +#define ZYWRLE_UVMASK __RFB_CONCAT2E(ZYWRLE_UVMASK,BPP) +#define ZYWRLE_LOAD_PIXEL __RFB_CONCAT2E(ZYWRLE_LOAD_PIXEL,BPP) +#define ZYWRLE_SAVE_PIXEL __RFB_CONCAT2E(ZYWRLE_SAVE_PIXEL,BPP) + +/* Packing/Unpacking pixel stuffs. */ +/* Endian conversion stuffs. */ +#undef S_0 +#undef S_1 +#undef L_0 +#undef L_1 +#undef L_2 +#if ZYWRLE_ENDIAN == ENDIAN_BIG +# define S_0 1 +# define S_1 0 +# define L_0 3 +# define L_1 2 +# define L_2 1 +#else +# define S_0 0 +# define S_1 1 +# define L_0 0 +# define L_1 1 +# define L_2 2 +#endif + +/* Load/Save pixel stuffs. */ +#define ZYWRLE_YMASK15 0xFFFFFFF8 +#define ZYWRLE_UVMASK15 0xFFFFFFF8 +#define ZYWRLE_LOAD_PIXEL15(pSrc,R,G,B) { \ + R = (((unsigned char*)pSrc)[S_1]<< 1)& 0xF8; \ + G = ((((unsigned char*)pSrc)[S_1]<< 6)|(((unsigned char*)pSrc)[S_0]>> 2))& 0xF8; \ + B = (((unsigned char*)pSrc)[S_0]<< 3)& 0xF8; \ +} +#define ZYWRLE_SAVE_PIXEL15(pDst,R,G,B) { \ + R &= 0xF8; \ + G &= 0xF8; \ + B &= 0xF8; \ + ((unsigned char*)pDst)[S_1] = (unsigned char)( (R>>1)|(G>>6) ); \ + ((unsigned char*)pDst)[S_0] = (unsigned char)(((B>>3)|(G<<2))& 0xFF); \ +} +#define ZYWRLE_YMASK16 0xFFFFFFFC +#define ZYWRLE_UVMASK16 0xFFFFFFF8 +#define ZYWRLE_LOAD_PIXEL16(pSrc,R,G,B) { \ + R = ((unsigned char*)pSrc)[S_1] & 0xF8; \ + G = ((((unsigned char*)pSrc)[S_1]<< 5)|(((unsigned char*)pSrc)[S_0]>> 3))& 0xFC; \ + B = (((unsigned char*)pSrc)[S_0]<< 3)& 0xF8; \ +} +#define ZYWRLE_SAVE_PIXEL16(pDst,R,G,B) { \ + R &= 0xF8; \ + G &= 0xFC; \ + B &= 0xF8; \ + ((unsigned char*)pDst)[S_1] = (unsigned char)( R |(G>>5) ); \ + ((unsigned char*)pDst)[S_0] = (unsigned char)(((B>>3)|(G<<3))& 0xFF); \ +} +#define ZYWRLE_YMASK32 0xFFFFFFFF +#define ZYWRLE_UVMASK32 0xFFFFFFFF +#define ZYWRLE_LOAD_PIXEL32(pSrc,R,G,B) { \ + R = ((unsigned char*)pSrc)[L_2]; \ + G = ((unsigned char*)pSrc)[L_1]; \ + B = ((unsigned char*)pSrc)[L_0]; \ +} +#define ZYWRLE_SAVE_PIXEL32(pDst,R,G,B) { \ + ((unsigned char*)pDst)[L_2] = (unsigned char)R; \ + ((unsigned char*)pDst)[L_1] = (unsigned char)G; \ + ((unsigned char*)pDst)[L_0] = (unsigned char)B; \ +} + +#ifndef ZYWRLE_ONCE +#define ZYWRLE_ONCE + +#ifdef WIN32 +#define InlineX __inline +#else +#define InlineX inline +#endif + +#ifdef ZYWRLE_ENCODE +/* Tables for Coefficients filtering. */ +# ifndef ZYWRLE_QUANTIZE +/* Type A:lower bit omitting of EZW style. */ +const static unsigned int zywrleParam[3][3]={ + {0x0000F000,0x00000000,0x00000000}, + {0x0000C000,0x00F0F0F0,0x00000000}, + {0x0000C000,0x00C0C0C0,0x00F0F0F0}, +/* + {0x0000FF00,0x00000000,0x00000000}, + {0x0000FF00,0x00FFFFFF,0x00000000}, + {0x0000FF00,0x00FFFFFF,0x00FFFFFF}, +*/ +}; +# else +/* Type B:Non liner quantization filter. */ +static const signed char zywrleConv[4][256]={ +{ /* bi=5, bo=5 r=0.0:PSNR=24.849 */ + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, +}, +{ /* bi=5, bo=5 r=2.0:PSNR=74.031 */ + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 32, + 32, 32, 32, 32, 32, 32, 32, 32, + 32, 32, 32, 32, 32, 32, 32, 32, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 56, 56, 56, 56, 56, + 56, 56, 56, 56, 64, 64, 64, 64, + 64, 64, 64, 64, 72, 72, 72, 72, + 72, 72, 72, 72, 80, 80, 80, 80, + 80, 80, 88, 88, 88, 88, 88, 88, + 88, 88, 88, 88, 88, 88, 96, 96, + 96, 96, 96, 104, 104, 104, 104, 104, + 104, 104, 104, 104, 104, 112, 112, 112, + 112, 112, 112, 112, 112, 112, 120, 120, + 120, 120, 120, 120, 120, 120, 120, 120, + 0, -120, -120, -120, -120, -120, -120, -120, + -120, -120, -120, -112, -112, -112, -112, -112, + -112, -112, -112, -112, -104, -104, -104, -104, + -104, -104, -104, -104, -104, -104, -96, -96, + -96, -96, -96, -88, -88, -88, -88, -88, + -88, -88, -88, -88, -88, -88, -88, -80, + -80, -80, -80, -80, -80, -72, -72, -72, + -72, -72, -72, -72, -72, -64, -64, -64, + -64, -64, -64, -64, -64, -56, -56, -56, + -56, -56, -56, -56, -56, -56, -48, -48, + -48, -48, -48, -48, -48, -48, -48, -48, + -48, -32, -32, -32, -32, -32, -32, -32, + -32, -32, -32, -32, -32, -32, -32, -32, + -32, -32, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, +}, +{ /* bi=5, bo=4 r=2.0:PSNR=64.441 */ + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 64, 64, 64, 64, 64, 64, 64, 64, + 64, 64, 64, 64, 64, 64, 64, 64, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 88, 88, 88, + 88, 88, 88, 88, 88, 88, 88, 88, + 104, 104, 104, 104, 104, 104, 104, 104, + 104, 104, 104, 112, 112, 112, 112, 112, + 112, 112, 112, 112, 120, 120, 120, 120, + 120, 120, 120, 120, 120, 120, 120, 120, + 0, -120, -120, -120, -120, -120, -120, -120, + -120, -120, -120, -120, -120, -112, -112, -112, + -112, -112, -112, -112, -112, -112, -104, -104, + -104, -104, -104, -104, -104, -104, -104, -104, + -104, -88, -88, -88, -88, -88, -88, -88, + -88, -88, -88, -88, -80, -80, -80, -80, + -80, -80, -80, -80, -80, -80, -80, -80, + -80, -64, -64, -64, -64, -64, -64, -64, + -64, -64, -64, -64, -64, -64, -64, -64, + -64, -48, -48, -48, -48, -48, -48, -48, + -48, -48, -48, -48, -48, -48, -48, -48, + -48, -48, -48, -48, -48, -48, -48, -48, + -48, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, +}, +{ /* bi=5, bo=2 r=2.0:PSNR=43.175 */ + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 88, 88, 88, 88, 88, 88, 88, 88, + 88, 88, 88, 88, 88, 88, 88, 88, + 88, 88, 88, 88, 88, 88, 88, 88, + 88, 88, 88, 88, 88, 88, 88, 88, + 88, 88, 88, 88, 88, 88, 88, 88, + 88, 88, 88, 88, 88, 88, 88, 88, + 88, 88, 88, 88, 88, 88, 88, 88, + 88, 88, 88, 88, 88, 88, 88, 88, + 0, -88, -88, -88, -88, -88, -88, -88, + -88, -88, -88, -88, -88, -88, -88, -88, + -88, -88, -88, -88, -88, -88, -88, -88, + -88, -88, -88, -88, -88, -88, -88, -88, + -88, -88, -88, -88, -88, -88, -88, -88, + -88, -88, -88, -88, -88, -88, -88, -88, + -88, -88, -88, -88, -88, -88, -88, -88, + -88, -88, -88, -88, -88, -88, -88, -88, + -88, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, +} +}; +const static signed char* zywrleParam[3][3][3]={ + {{zywrleConv[0],zywrleConv[2],zywrleConv[0]},{zywrleConv[0],zywrleConv[0],zywrleConv[0]},{zywrleConv[0],zywrleConv[0],zywrleConv[0]}}, + {{zywrleConv[0],zywrleConv[3],zywrleConv[0]},{zywrleConv[1],zywrleConv[1],zywrleConv[1]},{zywrleConv[0],zywrleConv[0],zywrleConv[0]}}, + {{zywrleConv[0],zywrleConv[3],zywrleConv[0]},{zywrleConv[2],zywrleConv[2],zywrleConv[2]},{zywrleConv[1],zywrleConv[1],zywrleConv[1]}}, +}; +# endif +#endif + +static InlineX void Harr(signed char* pX0, signed char* pX1) +{ + /* Piecewise-Linear Harr(PLHarr) */ + int X0 = (int)*pX0, X1 = (int)*pX1; + int orgX0=X0, orgX1=X1; + if ((X0^X1) & 0x80) { + /* differ sign */ + X1 += X0; + if (((X1^orgX1)&0x80)==0) { + /* |X1| > |X0| */ + X0 -= X1; /* H = -B */ + } + } else { + /* same sign */ + X0 -= X1; + if (((X0^orgX0) & 0x80) == 0) { + /* |X0| > |X1| */ + X1 += X0; /* L = A */ + } + } + *pX0 = (signed char)X1; + *pX1 = (signed char)X0; +} +/* + 1D-Wavelet transform. + + In coefficients array, the famous 'pyramid' decomposition is well used. + + 1D Model: + |L0L0L0L0|L0L0L0L0|H0H0H0H0|H0H0H0H0| : level 0 + |L1L1L1L1|H1H1H1H1|H0H0H0H0|H0H0H0H0| : level 1 + + But this method needs line buffer because H/L is different position from X0/X1. + So, I used 'interleave' decomposition instead of it. + + 1D Model: + |L0H0L0H0|L0H0L0H0|L0H0L0H0|L0H0L0H0| : level 0 + |L1H0H1H0|L1H0H1H0|L1H0H1H0|L1H0H1H0| : level 1 + + In this method, H/L and X0/X1 is always same position. + This lead us to more speed and less memory. + Of cause, the result of both method is quite same + because it's only difference that coefficient position. +*/ + +static InlineX void WaveletLevel(int* data, int size, int l, int SkipPixel) +{ + int s, ofs; + signed char* pX0; + signed char* end; + + pX0 = (signed char*)data; + s = (8<>(l+1))*s; + s -= 2; + ofs = (4<>1; + if (r & 0x02) + pH += (s>>1)*width; + for (y = 0; y < height / s; y++) { + for (x = 0; x < width / s; x++) { + /* + these are same following code. + pH[x] = pH[x] / (~pM[x]+1) * (~pM[x]+1); + ( round pH[x] with pM[x] bit ) + '&' operator isn't 'round' but is 'floor'. + So, we must offset when pH[x] is negative. + */ + if (((signed char*)pH)[0] & 0x80) + ((signed char*)pH)[0] += ~((signed char*)pM)[0]; + if (((signed char*)pH)[1] & 0x80) + ((signed char*)pH)[1] += ~((signed char*)pM)[1]; + if (((signed char*)pH)[2] & 0x80) + ((signed char*)pH)[2] += ~((signed char*)pM)[2]; + *pH &= *pM; + pH += s; + } + pH += (s-1)*width; + } + } +} +# else +/* + Type B:Non liner quantization filter. + + Coefficients have Gaussian curve and smaller value which is + large part of coefficients isn't more important than larger value. + So, I use filter of Non liner quantize/dequantize table. + In general, Non liner quantize formula is explained as following. + + y=f(x) = sign(x)*round( ((abs(x)/(2^7))^ r )* 2^(bo-1) )*2^(8-bo) + x=f-1(y) = sign(y)*round( ((abs(y)/(2^7))^(1/r))* 2^(bi-1) )*2^(8-bi) + ( r:power coefficient bi:effective MSB in input bo:effective MSB in output ) + + r < 1.0 : Smaller value is more important than larger value. + r > 1.0 : Larger value is more important than smaller value. + r = 1.0 : Liner quantization which is same with EZW style. + + r = 0.75 is famous non liner quantization used in MP3 audio codec. + In contrast to audio data, larger value is important in wavelet coefficients. + So, I select r = 2.0 table( quantize is x^2, dequantize sqrt(x) ). + + As compared with EZW style liner quantization, this filter tended to be + more sharp edge and be more compression rate but be more blocking noise and be less quality. + Especially, the surface of graphic objects has distinguishable noise in middle quality mode. + + We need only quantized-dequantized(filtered) value rather than quantized value itself + because all values are packed or palette-lized in later ZRLE section. + This lead us not to need to modify client decoder when we change + the filtering procedure in future. + Client only decodes coefficients given by encoder. +*/ + +static InlineX void FilterWaveletSquare(int* pBuf, int width, int height, int level, int l) +{ + int r, s; + int x, y; + int* pH; + const signed char** pM; + + pM = zywrleParam[level-1][l]; + s = 2<>1; + if (r & 0x02) + pH += (s>>1)*width; + for (y = 0; y < height / s; y++) { + for (x = 0; x < width / s; x++) { + ((signed char*)pH)[0] = pM[0][((unsigned char*)pH)[0]]; + ((signed char*)pH)[1] = pM[1][((unsigned char*)pH)[1]]; + ((signed char*)pH)[2] = pM[2][((unsigned char*)pH)[2]]; + pH += s; + } + pH += (s-1)*width; + } + } +} +# endif + +static InlineX void Wavelet(int* pBuf, int width, int height, int level) +{ + int l, s; + int* pTop; + int* pEnd; + + for (l = 0; l < level; l++) { + pTop = pBuf; + pEnd = pBuf+height*width; + s = width<= 0; l--) { + pTop = pBuf; + pEnd = pBuf+width; + s = 1< YUV conversion stuffs. + YUV coversion is explained as following formula in strict meaning: + Y = 0.299R + 0.587G + 0.114B ( 0<=Y<=255) + U = -0.169R - 0.331G + 0.500B (-128<=U<=127) + V = 0.500R - 0.419G - 0.081B (-128<=V<=127) + + I use simple conversion RCT(reversible color transform) which is described + in JPEG-2000 specification. + Y = (R + 2G + B)/4 ( 0<=Y<=255) + U = B-G (-256<=U<=255) + V = R-G (-256<=V<=255) +*/ + +#define ROUND(x) (((x)<0)?0:(((x)>255)?255:(x))) + /* + RCT is N-bit RGB to N-bit Y and N+1-bit UV. + For make Same N-bit, UV is lossy. + More exact PLHarr, we reduce to odd range(-127<=x<=127). + */ +#define ZYWRLE_RGBYUV1(R,G,B,Y,U,V,ymask,uvmask) { \ + Y = (R+(G<<1)+B)>>2; \ + U = B-G; \ + V = R-G; \ + Y -= 128; \ + U >>= 1; \ + V >>= 1; \ + Y &= ymask; \ + U &= uvmask; \ + V &= uvmask; \ + if (Y == -128) \ + Y += (0xFFFFFFFF-ymask+1); \ + if (U == -128) \ + U += (0xFFFFFFFF-uvmask+1); \ + if (V == -128) \ + V += (0xFFFFFFFF-uvmask+1); \ +} +#define ZYWRLE_YUVRGB1(R,G,B,Y,U,V) { \ + Y += 128; \ + U <<= 1; \ + V <<= 1; \ + G = Y-((U+V)>>2); \ + B = U+G; \ + R = V+G; \ + G = ROUND(G); \ + B = ROUND(B); \ + R = ROUND(R); \ +} + +/* + coefficient packing/unpacking stuffs. + Wavelet transform makes 4 sub coefficient image from 1 original image. + + model with pyramid decomposition: + +------+------+ + | | | + | L | Hx | + | | | + +------+------+ + | | | + | H | Hxy | + | | | + +------+------+ + + So, we must transfer each sub images individually in strict meaning. + But at least ZRLE meaning, following one decompositon image is same as + avobe individual sub image. I use this format. + (Strictly saying, transfer order is reverse(Hxy->Hy->Hx->L) + for simplified procedure for any wavelet level.) + + +------+------+ + | L | + +------+------+ + | Hx | + +------+------+ + | Hy | + +------+------+ + | Hxy | + +------+------+ +*/ + +#define ZYWRLE_TRANSFER_COEFF(pBuf,data,r,width,height,level,TRANS) \ + pH = pBuf; \ + s = 2<>1; \ + if (r & 0x02) \ + pH += (s>>1)*width; \ + pEnd = pH+height*width; \ + while (pH < pEnd) { \ + pLine = pH+width; \ + while (pH < pLine) { \ + TRANS \ + data++; \ + pH += s; \ + } \ + pH += (s-1)*width; \ + } + +#define ZYWRLE_PACK_COEFF(pBuf,data,r,width,height,level) \ + ZYWRLE_TRANSFER_COEFF(pBuf,data,r,width,height,level,ZYWRLE_LOAD_COEFF(pH,R,G,B);ZYWRLE_SAVE_PIXEL(data,R,G,B);) + +#define ZYWRLE_UNPACK_COEFF(pBuf,data,r,width,height,level) \ + ZYWRLE_TRANSFER_COEFF(pBuf,data,r,width,height,level,ZYWRLE_LOAD_PIXEL(data,R,G,B);ZYWRLE_SAVE_COEFF(pH,R,G,B);) + +#define ZYWRLE_SAVE_UNALIGN(data,TRANS) \ + pTop = pBuf+w*h; \ + pEnd = pTop + (w+uw)*(h+uh)-w*h; \ + while (pTop < pEnd) { \ + TRANS \ + data++; \ + pTop++; \ + } + +#define ZYWRLE_LOAD_UNALIGN(data,pData,TRANS) \ + pTop = pBuf+w*h; \ + if (uw) { \ + pData= data + w; \ + pEnd = (int*)(pData+ h*scanline); \ + while (pData < (PIXEL_T*)pEnd) { \ + pLine = (int*)(pData + uw); \ + while (pData < (PIXEL_T*)pLine) { \ + TRANS \ + pData++; \ + pTop++; \ + } \ + pData += scanline-uw; \ + } \ + } \ + if (uh) { \ + pData= data + h*scanline; \ + pEnd = (int*)(pData+ uh*scanline); \ + while (pData < (PIXEL_T*)pEnd) { \ + pLine = (int*)(pData + w); \ + while (pData < (PIXEL_T*)pLine) { \ + TRANS \ + pData++; \ + pTop++; \ + } \ + pData += scanline-w; \ + } \ + } \ + if (uw && uh) { \ + pData= data + w+ h*scanline; \ + pEnd = (int*)(pData+ uh*scanline); \ + while (pData < (PIXEL_T*)pEnd) { \ + pLine = (int*)(pData + uw); \ + while (pData < (PIXEL_T*)pLine) { \ + TRANS \ + pData++; \ + pTop++; \ + } \ + pData += scanline-uw; \ + } \ + } + +static InlineX void zywrleCalcSize(int* pW, int* pH, int level) +{ + *pW &= ~((1<