Links and Guides to Custom Shaders for Pete's OpenGL2 plugin

[SimoneT]

New 5xBR v3.7a GLSL and ARB (for NVIDIA cards only)...
Big speedup... and no code obfuscation (so Guest.r can copy without problems)!
Remember to disable "Screen filtering" and set "Internal X/Y resolution to 0:low"
Ciao.

Attached Files

5xBR v3.7a NVIDIA.zip (4.6 KB, 259 views)

__________________
CPU: Intel CORE i5 M 560 @ 2.67 GHz
GPU: NVIDIA NVS 3100M

[Hyllian]

Are you planning "b" and "c" versions?

[SimoneT]

Originally Posted by Hyllian

Are you planning "b" and "c" versions?

Yes, but I think the "a" version is enough good for most PSX games (and gamers). I'm working on a version that permit the use of the internal screen filtering. When I think is ready I sent you a PM.
Ciao.
P.S.: have you tested my "vertex calculation" trick?

__________________
CPU: Intel CORE i5 M 560 @ 2.67 GHz
GPU: NVIDIA NVS 3100M

[Hyllian]

Originally Posted by SimoneT

Yes, but I think the "a" version is enough good for most PSX games (and gamers). I'm working on a version that permit the use of the internal screen filtering. When I think is ready I sent you a PM.
Ciao.

Ok. Does the 'internal screen filtering' work over textures?

Originally Posted by SimoneT

P.S.: have you tested my "vertex calculation" trick?

Not yet. But I've passed my eyes a bit over it. It seems you know some clever shader tricks.

I'll try to translate to Cg later and see what happens.

[Hyllian]

That I4 texture lookup should be "xw" instead "xy", or Am I mistaken?

[SimoneT]

Originally Posted by Hyllian

That I4 texture lookup should be "xw" instead "xy", or Am I mistaken?

You are right!
Ciao.

__________________
CPU: Intel CORE i5 M 560 @ 2.67 GHz
GPU: NVIDIA NVS 3100M

[Hyllian]

SimoneT, why don't you use variables instead assuming resolution of 1024x512?

I'll use something like this in my code (not tested yet):

	half2 ps = half2(1.0/IN.texture_size.x, 1.0/IN.texture_size.y);
	half dx  = ps.x;
	half dy  = ps.y;

//    A1 B1 C1
// A0  A  B  C C4
// D0  D  E  F F4
// G0  G  H  I I4
//    G5 H5 I5

	OUT.texCoord = texCoord;
	OUT.t1 = texCoord.xxxy + half4( -dx, 0, dx,-2.0*dy); // A1 B1 C1
	OUT.t2 = texCoord.xxxy + half4( -dx, 0, dx,    -dy); //  A  B  C 
	OUT.t3 = texCoord.xxxy + half4( -dx, 0, dx,      0); //  D  E  F 
	OUT.t4 = texCoord.xxxy + half4( -dx, 0, dx,     dy); //  G  H  I 
	OUT.t5 = texCoord.xxxy + half4( -dx, 0, dx, 2.0*dy); // G5 H5 I5
	OUT.t6 = texCoord.xyyy + half4(-2.0*dx,-dy, 0,  dy); // A0 D0 G0
	OUT.t7 = texCoord.xyyy + half4( 2.0*dx,-dy, 0,  dy); // C4 F4 I4

[SimoneT]

Originally Posted by Hyllian

SimoneT, why don't you use variables instead assuming resolution of 1024x512?

I'll use something like this in my code (not tested yet):

	half2 ps = half2(1.0/IN.texture_size.x, 1.0/IN.texture_size.y);
	half dx  = ps.x;
	half dy  = ps.y;

//    A1 B1 C1
// A0  A  B  C C4
// D0  D  E  F F4
// G0  G  H  I I4
//    G5 H5 I5

	OUT.texCoord = texCoord;
	OUT.t1 = texCoord.xxxy + half4( -dx, 0, dx,-2.0*dy); // A1 B1 C1
	OUT.t2 = texCoord.xxxy + half4( -dx, 0, dx,    -dy); //  A  B  C 
	OUT.t3 = texCoord.xxxy + half4( -dx, 0, dx,      0); //  D  E  F 
	OUT.t4 = texCoord.xxxy + half4( -dx, 0, dx,     dy); //  G  H  I 
	OUT.t5 = texCoord.xxxy + half4( -dx, 0, dx, 2.0*dy); // G5 H5 I5
	OUT.t6 = texCoord.xyyy + half4(-2.0*dx,-dy, 0,  dy); // A0 D0 G0
	OUT.t7 = texCoord.xyyy + half4( 2.0*dx,-dy, 0,  dy); // C4 F4 I4

Your code is right. 1024x512 is the psx buffer resolution at low X/Y, used as a input texture (pbuffer) by the Pete OGL2 plugin. If you have a fixed input resolution, is better that you force it into your vertex program.
...and x+x is little bit faster than 2.0*x...

__________________
CPU: Intel CORE i5 M 560 @ 2.67 GHz
GPU: NVIDIA NVS 3100M

[Hyllian]

Originally Posted by SimoneT

...and x+x is little bit faster than 2.0*x...

Ok, tks. And I've seen you used x*x instead abs(x) on the d function. Is it always faster?

[SimoneT]

Originally Posted by Hyllian

Ok, tks. And I've seen you used x*x instead abs(x) on the d function. Is it always faster?

Yes, is a little bit faster. Please use PM if you can, is better to no fill this tread with personal conversations...
Ciao.

__________________
CPU: Intel CORE i5 M 560 @ 2.67 GHz
GPU: NVIDIA NVS 3100M

[RyuRanX]

Can someone port bsnes' CRT shaders to the openGL2 plugin format? I love the way it looks with the curvature and all.

__________________
Avatar! Saiba que Britannia entrou em uma nova era de iluminismo! Saiba que a hora finalmente chegou para que o verdadeiro Senhor de Britannia clame Seu lugar à frente do Seu povo! Sobre minha guia, Britannia florescerá, e todos festejarão e prestarão homenagens ao seu novo... Guardião! Saiba que você também deverá se ajoelhar diante de mim, Avatar. Você também brevemente tomará consciência da minha autoridade - pois eu serei seu companheiro... seu provisor... e seu mestre!


Core2Duo E4600 - 2,4ghz - 2GB DDR2 800mhz - Zogis GEforce 9800GT 512mb - Windows XP SP3

[KrossX]

I tried, but the curve didn't work. It kept flickering. =S

__________________

[SimoneT]

Small update. In my system, the ARB(fp/vp) version is a little bit faster than the GLSL(slf/slv). Ciao.

Attached Files

5xBR v3.7a NVIDIA rev2.zip (4.7 KB, 213 views)

__________________
CPU: Intel CORE i5 M 560 @ 2.67 GHz
GPU: NVIDIA NVS 3100M

[Hyllian]

SimoneT, I've seen you changed this code:

	half3 res = nc.x ? px.x ? F : H : nc.y ? px.y ? B : F : nc.z ? px.z ? D : B : nc.w ? px.w ? H : D : E;

By this:

if ((nc.x && px.x) || (nc.y && !px.y)) 
{
	E = F;
} 
else
if ((nc.y && px.y) || (nc.z && !px.z)) 
{
	E = B;
} 
else
if ((nc.z && px.z) || (nc.w && !px.w)) 
{
	E = D;
} 
else
if ((nc.w && px.w) || (nc.x && !px.x)) 
{
	E = H;
}

This is wrong, because you changed priorities. I expect some small artifacts to appear now.

[SimoneT]

Originally Posted by Hyllian

...this is wrong, because you changed priorities. I expect some small artifacts to appear now.

Here your code in c++ notation:

    if (nc.x) { // if nc.x begin
        if (px.x) { 
            TMP2 = F;
        } else {
            TMP2 = H;
        } 
        TMP1 = TMP2;
    } else {
        if (nc.y) { // if nc.y begin
            if (px.y) { 
                TMP4 = B;
            } else {
                TMP4 = F;
            } 
            TMP3 = TMP4;
        } else {
            if (nc.z) { // if nc.z begin
                if (px.z) { 
                    TMP6 = D;
                } else {
                    TMP6 = B;
                } 
                TMP5 = TMP6;
            } else {
                if (nc.w) { // if nc.w begin
                    if (px.w) { 
                        TMP8 = H;
                    } else {
                        TMP8 = D;
                    } 
                    TMP7 = TMP8;
                } else {
                    TMP7 = E;
                } // end if nc.w
                TMP5 = TMP7;
            } // end if nc.z
            TMP3 = TMP5;
        } // end if nc.y 
        TMP1 = TMP3;
    } // end if nc.x 
    gl_FragColor = TMP1;

My code is only an optimization, try all 16 cases of nc and px...
Ciao.

__________________
CPU: Intel CORE i5 M 560 @ 2.67 GHz
GPU: NVIDIA NVS 3100M

---

Last edited by SimoneT; April 4th, 2012 at 17:20..

[Hyllian]

In my code, if (nc.x==1 && px.x==0) then it ALWAYS return H.

In your code, if (nc.x==1 && px.x==0), then it MAY return F, for example.

[SimoneT]

Originally Posted by Hyllian

In my code, if (nc.x==1 && px.x==0) then it ALWAYS return H.

In your code, if (nc.x==1 && px.x==0), then it MAY return F, for example.

I have tested your assumption and... I discovered that my code, compared to yours, is able to better discriminate the ambiguities of the edges (see the screenshots). But, if you wants, I will rewrite the shader code as the original.
Ciao
P.S.: the original.jpg was at 128x64 scaled to 1280x640 (10x)

Attached Images

	Original.jpg (95.2 KB, 117 views)
	yourcode.jpg (136.2 KB, 118 views)
	mycode.jpg (136.2 KB, 117 views)
	Difference.jpg (5.7 KB, 68 views)

__________________
CPU: Intel CORE i5 M 560 @ 2.67 GHz
GPU: NVIDIA NVS 3100M

---

Last edited by SimoneT; April 5th, 2012 at 10:18..

[Hyllian]

I don't know if we can assume it always happen based only in one example picture.

The fact I was trying to demonstrate is that those pieces of codes weren't equivalent anymore. We can test more and if it becomes clear your optimizations fix some defects, then I will incorporate to the cg codes. Improvements are always welcomed!

[SimoneT]

Originally Posted by Hyllian

I don't know if we can assume it always happen based only in one example picture.

The fact I was trying to demonstrate is that those pieces of codes weren't equivalent anymore. We can test more and if it becomes clear your optimizations fix some defects, then I will incorporate to the cg codes. Improvements are always welcomed!

Ok. I will test with more games and emulators (I have ported yours shader to bsnes but I haven't released it...) and i will post some more screenshots.

__________________
CPU: Intel CORE i5 M 560 @ 2.67 GHz
GPU: NVIDIA NVS 3100M

[Hyllian]

Originally Posted by SimoneT

Ok. I will test with more games and emulators (I have ported yours shader to bsnes but I haven't released it...) and i will post some more screenshots.

Great. This code is so tight I always expect artifacts when something is changed.


Bsnes? Someone else already made it in this thread: xml shader

Maybe you wanna contribute there.

EDIT: Your implementation seems a bit different from mine. This is what I'm getting with my Cg shader on PS3:

Attached Images

SNES9x PS3_6.JPG (207.0 KB, 195 views)

---

Last edited by Hyllian; April 5th, 2012 at 13:13..