GPU_SetScissorTest: function prototype misrepresents usage. about libctru HOT 11 CLOSED

So it should be left, top, right, bottom then (assuming Y points down).

from libctru.

@fincs I'm not sure left,top,right,bottom is right, especially since we're mostly working on sideway buffers.

from libctru.

The command is loosely based on glScissor. x,y specify the lower left corner of the box so 0,0 is the very bottom-right of the screen because of the sideways framebuffer then x1 and y1 specify the upper-right corner (unlike glScissor) of the box. Arguably, you could make it left, bottom, right, top, however, I advise against this as it implies a specific interpretation of how the command is used considering some developers will rotate their scene to accommodate the sideways framebuffer while others may choose to make their game sideways oriented.

from libctru.

Please remember that the physical screens are sideways, not the framebuffer. The physical scanlines are vertical instead of horizontal. Note that it is perfectly possible to use rendered output as a texture - the output should not be tilted.

from libctru.

Ah I see, then I'd agree to left, bottom, right, top as the parameters.

from libctru.

Is it confirmed that Y points up, as in the actual behavior of this PICA command?

from libctru.

I can confirm this. Transferring the GPU output directly to the screen gives these results:
GPU_SetScissorTest(GPU_SCISSOR_NORMAL, 120, 200, 240, 400);
Top-left quarter of the screen is visible. top-right, bottom-right, bottom-left are excluded.

GPU_SetScissorTest(GPU_SCISSOR_NORMAL, 0, 0, 120, 200);
Bottom-right quarter of the screen is visible. Bottom-left, top-left, top-right are excluded.

from libctru.

Those are my notes for the 3DS axes :

LCD :

240
x
^
|
|
|
|
|
+------------------>y 400(320)

(0,0)

displays only with z in [-1.0,0.0]

gpu with setViewPort(0,0,240,400)

(1.0,1.0)             x 
   +   -   -   -   -  ^ 1.0
                      |
   |                  |
                      |
   |                  |
  y<------------------+
 1.0            (-1.0,-1.0)     

Textures :

   (0,0)                1.0
     +------------------>u
     |
     |
     |
     |
     |  
 1.0 V
   v

from libctru.

@ic3water : this issue has nothing to do with it. If you need help for the IES parser open an issue on the IES parser repo.

from libctru.

Is this method actually working? I tried to use it in the simple_tri example and it has no effect.
I modified the main loop like this:

while (aptMainLoop())
    {
        gspWaitForVBlank();  // Synchronize with the start of VBlank
        gfxSwapBuffersGpu(); // Swap the framebuffers so that the frame that we rendered last frame is now visible
        hidScanInput();      // Read the user input

        // Respond to user input
        u32 kDown = hidKeysDown();
        if (kDown & KEY_START)
            break; // break in order to return to hbmenu

        // Render the scene
        gpuFrameBegin();
        GPU_SetScissorTest(GPU_SCISSOR_NORMAL, 0, 0, 20, 20);
        sceneRender();
        gpuFrameEnd();
        gpuClearBuffers(CLEAR_COLOR);

        // Flush the framebuffers out of the data cache (not necessary with pure GPU rendering)
        //gfxFlushBuffers();
    }

I expected that only a part of the triangle should be drawn. But in the result the method call has no effect.

from libctru.

Ignore my last comment. It works in real Hardware. Scissortest is mit yet implemented properly in citra.

from libctru.