-> Changed to the S2DEX microcode (not compatible)
gspSprite2D, gspSprite2D.dram, gspSprite2D.fifo
-> High-quality 2D sprite geometry microcode
this is the optimized, high-quality, full-featured 2D sprite geometry microcode. It supports automatic subdivision and loads any size of all of the texture format sizes and types supported in the gSPTextureRectangle command, and sends it directly to the RDP. Additionally, images can be scaled up or inverted in the X or Y directions.
The sprite microcode is accessed through a combination of four new commands. The first command, guSprite2DInit initializes a specified a sprite structure. gSPSprite2DBase command sends the sprite structure to the sprite microcode to start actual processing. guSprite2DInit command is a useful function that the application doesn't need to initialize the sprite structure.
gSPSprite2DBase initializes or the common sprite parameters. It does not perform actual screen drawing.
gSPSprite2DScaleFlip is used to specify the X/Y scaling and/or flipping parameters for a sprite. It does not perform actual screen drawing.
gSPSprite2DDraw specifies the screen coordinates where the sprite is to be drawn, and starts actual screen drawing using the parameters specified by gSPSprite2DBase and gSPSprite2DScaleFlip.
Simple code for displaying a Sprite:
#include "gu.h"
#include "gbi.h"
uSprite MySprite;
guSprite2DInit(&MySprite, ImagePointer, TlutPointer,
ImageWidth, RectangleWidth, RectangleHeight,
ImageType, ImageSize,
TextureStartS, TextureStartT);
gSPSprite2DBase(glistp++, OS_K0_TO_PHYSICAL(&MySprite));
gSPSprite2DScaleFlip(glistp++, ScaleX, ScaleY,
FlipTextureX, FlipTextureY);
gSPSprite2DDraw(glistp++, PScreenX, PScreenY)
Each of the arguments is explained below.
typedef struct {
void *SourceImagePointer;
void *TlutPointer;
short Stride;
short SubImageWidth;
short SubImageHeight;
char SourceImageType;
char SourceImageBitSize;
short SourceImageOffsetS;
short SourceImageOffsetT;
/* 20 bytes for above */
/* padding to bring structure size to 64 bit allignment */;
char dummy[4];
} uSprite_t;
typedef union {
uSprite_t s;
/* Need to make sure this is 64 bit aligned */;
long long int force_structure_allignment[3];
} uSprite;
void guSprite2DInit(uSprite *SpritePointer,
void *SourceImagePointer,
void *TlutPointer,
int Stride,
int SubImageWidth,
int SubImageHeight,
int SourceImageType,
int SourceImageBitSize,
int SourceImageOffsetS,
int SourceImageOffsetT);
The following GBI commands are not supported by this microcode:
The sprite microcode does not directly support Z-Buffering. This is unnecessary as Z-Buffering can be accomplished outside of the sprite microcode by setting up the proper rendering mode and making use of the hardware primitive depth registers. Following is a code fragment that does Z-Buffering.
gDPSetRenderMode(glistp++, G_RM_AA_ZB_OPA_SURF, G_RM_AA_ZB_OPA_SURF2); gDPSetDepthSource(glistp++, G_ZS_PRIM); gDPSetCombineMode(glistp++, G_CC_DECALRGB, G_CC_DECALRGB); gDPSetPrimDepth(glistp++, ZBufferValue, 0);
guSprite2DInit(&MySprite, ImagePointer, TlutPointer,
ImageWidth, RectangleWidth, RectangleHeight,
ImageType, ImageSize,
TextureStartS, TextureStartT);
gSPSprite2DBase(glistp++, OS_K0_TO_PHYSICAL(&MySprite));
gSPSprite2DScaleFlip(glistp++, ScaleX, ScaleY,
FlipTextureX, FlipTextureY);
gSPSprite2DDraw(glistp++, PScreenX, PScreenY);
Images that have been non-unit scaled and flipped around the Y axis may not be smoothly converted in the vertical direction, depending on the quantity of sub-pixels. Jumping will occur at a certain quantity. The solution is to convert non-unit scaling to unit amounts in the vertical direction.
The Sprite Microcode was designed to be able to scale up images by any amount. Images can also be scaled down together with some attendant artifacts. Please note that, while the TextureScaleX and TextureScaleY parameters are s 5.10 fixed-point numbers, they are restricted to being positive. Consequently, the largest usable scale value is 32767, which corresponds to a texel to pixel ratio of 31.999.
Texture images that are either scaled in the Y axis or placed on a subpixel scanline boundary require filtering by the hardware texture filter unit. This filtering requires that at least one extra line in the screen image be loaded in the texture memory so that the filtering can occur. The texture memory is limited to 4K bytes, so there are some restrictions:
32 bit subrectangles that are scaled in the Y direction are clamped by the microcode to a maximum of 512 texels wide.
32 bit subrectangles that are placed at the beginning of a subpixel boundary in the Y direction and that are larger than 512 texels in width are clamped by the microcode to being on an integral scan line boundary.
These filtering conditions also mean that, when displaying a Y-scaled image, you will need to add 1 to the T value to prevent filtering of texels outside the range.
Y flipped images that have been scaled vertically by a non-squared value will be slightly different in height than their non-flipped versions. If you are doing an animation involving scaled flipped and non-flipped textures, restrict the Y scale value to a squared value.
gspLine3D, gspTurbo3D, gspFast3D, osSpTaskStart, gSPSprite2DBase, gSPSprite2DScaleFlip, gSPSprite2DDraw