You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Everything seems to work fine, build and run is ok but the writings are a horrible rendering artifact made by retorts lines in columns. What could be the issue? Maybe something about compatibility? I'm testing on the latest Ubuntu with a Radeon GPU.
This sounds like the problem I also encountered, and I found that the implementation of Texture::CreateFromSurface is not completely correct - it works when I boot into Windows, but not in Ubuntu.
The problem is here: glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, mWidth, mHeight, 0, GL_BGRA, GL_UNSIGNED_BYTE, surface->pixels);
where: mWidth = surface->w;
The pixel data in the rendered surface is in fact NOTsurface->w * surface->h pixels, as might be expected, but instead you should use:
mWidth = surface->pitch/4;
The explanation (as I understand it) is that the rendered surface is padded by some bytes in order to reach a "nice" width (seems to be a factor of 16 or something like that), so each row in the pixel data may in fact be longer than "surface->w". And since surface->pitch is defined as number of bytes per row, it should be divided by the number of bytes per pixel (4 channels, i.e. RGBA).
Activity
Sioniras commentedon Sep 6, 2023
This sounds like the problem I also encountered, and I found that the implementation of
Texture::CreateFromSurfaceis not completely correct - it works when I boot into Windows, but not in Ubuntu.The problem is here:
glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, mWidth, mHeight, 0, GL_BGRA, GL_UNSIGNED_BYTE, surface->pixels);where:
mWidth = surface->w;The pixel data in the rendered surface is in fact NOT
surface->w * surface->hpixels, as might be expected, but instead you should use:mWidth = surface->pitch/4;The explanation (as I understand it) is that the rendered surface is padded by some bytes in order to reach a "nice" width (seems to be a factor of 16 or something like that), so each row in the pixel data may in fact be longer than "surface->w". And since surface->pitch is defined as number of bytes per row, it should be divided by the number of bytes per pixel (4 channels, i.e. RGBA).