AntonioND/nitro-engine
1
Fork 0
mirror of https://github.com/AntonioND/nitro-engine.git synced 2025-06-19 00:55:38 -04:00
Code Issues Actions Packages Projects Releases Wiki Activity

library: Fix 2D projection

Browse Source 
The current factor of 12 is too big, and it causes significant Y axis
distortion in the 2D projection. It has been lowered to 2.

Check the comment in the code for more information.
This commit is contained in:
Antonio Niño Díaz 2023-07-16 13:46:52 +01:00
parent 8441e9d05e
commit 193a50c939
1 changed files with 28 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
source/NE2D.c
View File

@ -269,22 +269,44 @@ void NE_2DViewInit(void)
GFX_VIEWPORT = 0 | (0 << 8) | (255 << 16) | (191 << 24); GFX_VIEWPORT = 0 | (0 << 8) | (255 << 16) | (191 << 24);
// The projection matrix actually thinks that the size of the DS is // The projection matrix actually thinks that the size of the DS is
// (256 << 12) x (192 << 12). After this, we scale the MODELVIEW matrix to // (256 << factor) x (192 << factor). After this, we scale the MODELVIEW
// match this scale factor. // matrix to match this scale factor.
// //
// This way, it is possible to draw on the screen by using numbers up to 256 // This way, it is possible to draw on the screen by using numbers up to 256
// x 192, but internally the DS has more digits when it does transformations // x 192, but internally the DS has more digits when it does transformations
// like a rotation. // like a rotation. Not having this factor results in noticeable flickering,
// specially in some emulators.
//
// Unfortunately, applying this factor reduces the accuracy of the Y
// coordinate a lot (nothing is noticeable in the X coordinate). Any factor
// over 4 starts showing a noticeable accuracy loss: some sprites start
// being slightly distorted, with missing some horizontal lines as the
// height is reduced. When the number is higher, like 12, the Y coordinate
// is significantly compressed. When the number is even higher, like 18, the
// polygons disappear because too much accuracy has been lost.
//
// The current solution is to compromise, and use a factor of 2, which
// doesn't cause any distortion, and solves most of the flickering. Ideally
// we would use 0 to simplify the calculations, but we want to reduce the
// flickering.
//
// On hardware, the difference in flickering between 0 and 2 isn't too
// noticeable, but it is noticeable. In DeSmuMe it is very noticeable.
// In my tests, Y axis distortion starts to happen with a factor of 4, so a
// factor of 2 should be safe and reduce enough flickering.
MATRIX_CONTROL = GL_PROJECTION; MATRIX_CONTROL = GL_PROJECTION;
MATRIX_IDENTITY = 0; MATRIX_IDENTITY = 0;
glOrthof32(0, 256 << 12, 192 << 12, 0, inttof32(1), inttof32(-1));
int factor = 2;
glOrthof32(0, 256 << factor, 192 << factor, 0, inttof32(1), inttof32(-1));
MATRIX_CONTROL = GL_MODELVIEW; MATRIX_CONTROL = GL_MODELVIEW;
MATRIX_IDENTITY = 0; MATRIX_IDENTITY = 0;
MATRIX_SCALE = inttof32(1 << 12); MATRIX_SCALE = inttof32(1 << factor);
MATRIX_SCALE = inttof32(1 << 12); MATRIX_SCALE = inttof32(1 << factor);
MATRIX_SCALE = inttof32(1); MATRIX_SCALE = inttof32(1);
NE_PolyFormat(31, 0, 0, NE_CULL_NONE, 0); NE_PolyFormat(31, 0, 0, NE_CULL_NONE, 0);