Render-to-Texture with a texture larger than the screen

OpenGL OpenGL: Basic Coding
1

Hello !

For an OpenGL application I need to be able to copy an image onto another using Render-to-Texture. Here is the section of code in question (note that I’m using JOGL, a Java OpenGL wrapper, but OpenGL should work the same as in C) :

Code section 
this.context.gl.glBindFramebuffer(GL.GL_FRAMEBUFFER, context.getTextureRenderer());
this.context.gl.glFramebufferTexture2D(GL.GL_FRAMEBUFFER, GL2.GL_COLOR_ATTACHMENT0, GL.GL_TEXTURE_2D, texture.get(0), 0);
this.context.gl.glDrawBuffer(GL2.GL_COLOR_ATTACHMENT0);

this.context.gl.glBindTexture(GL.GL_TEXTURE_2D, src.getTexture());

float sx0 = getXTex(src, sx);
float sy0 = getYTex(src, sy);
float sx1 = getXTex(src, sx + sw);
float sy1 = getYTex(src, sy + sh);

float dx0 = getXScreen(dx);
float dy0 = getYScreen(dy);
float dx1 = getXScreen(dx + dw);
float dy1 = getYScreen(dy + dh);

float actual_sx0 = max(sx0, 0);
float actual_sy0 = max(sy0, 0);
float actual_sx1 = min(sx1, 1);
float actual_sy1 = min(sy1, 1);
		
float actual_dx0 = map(map(actual_sx0, sx0, sx1, 0, 1), 0, 1, dx0, dx1);
float actual_dy0 = map(map(actual_sy0, sy0, sy1, 0, 1), 0, 1, dy0, dy1);
float actual_dx1 = map(map(actual_sx1, sx0, sx1, 0, 1), 0, 1, dx0, dx1);
float actual_dy1 = map(map(actual_sy1, sy0, sy1, 0, 1), 0, 1, dy0, dy1);

this.context.gl.glColor4f(1f, 1f, 1f, 1f);

this.context.gl.glBegin(GL2.GL_QUADS);
this.context.gl.glTexCoord2f(actual_sx0, actual_sy0);
this.context.gl.glVertex2f(actual_dx0, actual_dy0);
this.context.gl.glTexCoord2f(actual_sx0, actual_sy1);
this.context.gl.glVertex2f(actual_dx0, actual_dy1);
this.context.gl.glTexCoord2f(actual_sx1, actual_sy1);
this.context.gl.glVertex2f(actual_dx1, actual_dy1);
this.context.gl.glTexCoord2f(actual_sx1, actual_sy0);
this.context.gl.glVertex2f(actual_dx1, actual_dy0);
this.context.gl.glEnd();
this.context.gl.glFlush();

this.context.gl.glBindTexture(GL2.GL_TEXTURE_2D, 0);

this.context.gl.glFramebufferTexture2D(GL.GL_FRAMEBUFFER, GL2.GL_COLOR_ATTACHMENT0, GL.GL_TEXTURE_2D, 0, 0);
this.context.gl.glBindFramebuffer(GL.GL_FRAMEBUFFER, 0);
this.context.gl.glDrawBuffer(GL.GL_FRONT_AND_BACK);

When calling this specific function with a destination size larger than the screen size, the image seems to be “cut off”, as if the pixels were simply not rendered “outside of the screen”, even though it is on an image that itself is larger than the screen.

Resulting image

The top image is the input, and the bottom one is a resized version of the input, to double the screen width and height, and then rendered onto the screen at the same size as the input. I would in this case want both images to be rendered as the same (although in memory the bottom one would be stored onto a higher resolution image).

frame
frame1280×720 118 KB

I looked un both the glScissor and the glViewport functions, but as far as I understand the glViewport function only modifies the pixel coordinates transform and shouldn’t cause this issue, but disabling the GL_SCISSOR_TEST should act as if the scissor box encapsulates the whole window and disables rendering outside of it.

This is why I also tried setting the scissor box to a larger size than the screen size and enabling the scissor test, but this seems to change nothing.

Is there anything that may cause rendering to fail in this case ? Perhaps if the coordinates are outside of the range [-1, 1] nothing is rendered and a glViewport call is mandatory ?

Thanks !

2

Geometry is clipped to the unit cube [-1,1]3 in normalised device coordinated (NDC), i.e. after the projection transform but before the viewport transform. The effect is that primitives are clipped to the viewport. So you need to call glViewport with the dimensions of the texture.

The first time you bind a context to a surface (this will probably be done automatically by the GUI toolkit), the viewport is automatically set to the bounds of that surface. But after that, the viewport won’t change unless you call glViewport. So you need to do that if the window dimensions change or if you render to a texture or renderbuffer with dimensions different from the window.

Note that the “unit cube” clipping is applied to geometry, while the scissor test is applied to the results of rasterisation. If the scissor test is enabled, nothing will be rendered outside the scissor box. For filled triangles, nothing will be rendered outside the viewport, but wide lines and points larger than a pixel extend outside their defining geometry, so rasterisation can cause pixels outside the viewport to be rendered.

3

Thanks for the reply ! I’ll try changing the viewport then :slight_smile:

Update : yes it worked, thanks a lot for the quick reply :3