@icedquinn @xianc78 afaik all GPUs use square textures with power of two sizes, but most modern game engines and even modern GPUs handle NPOT textures using a "fake" square, nearest power of 2 texture by placing it on a bigger texture or scaling it if necessary. This of course comes with a performance and memory usage penalty.

Not the case for old consoles or GPUs.

@enigmatico @xianc78 the PC ones were powers of two but they had max sizes. for iphones that was around 1024x1024.

i was told the NES had fixed size sprites and you just called it to draw that sprite at some <x,y> and it could only draw a precious few unless you resorted to flickering between frames.

i dunno how SEGA managed

@enigmatico @xianc78 there really were not "textures" per say if we're talking about the 2D consoles. everything was bitblt and they kinda had bitblt accelerators but the triangle accelerator would have still been locked up in SGI workstations

@enigmatico @xianc78 a speccy *only* had 8x8 character memory because that's what could render words. so all the graphics were either loaded in to character memory and called to draw by index or had to be done much more slowly with pokes and manual line drawing.

i think up through the gameboy advance they were still reliant on bitblt with CPU transforms. it really wasn't until the N64 that we had the fixed function GPUs that even had texture memory that you could sample on to triangles

anyway i'm sure console enthusiasts would know more i never had a SEGA

@icedquinn @xianc78 Those 8-16 bit consoles are different. Sprites and textures are also different concepts. 2D sprites use pixels. Textures use texels. 2D sprites normally use indexed color palettes (could also be 24-bit color bitmaps, it depends). 3D textures might use 24-32 bit color (depending on wether it's RGB or RGBA or the likes of, the order might change). Sprites might be rendered in 3D space, but that's a "trick" pulled out by the console. i.e: willboards do this.

Normally sprites and 3D objects are rendered in separated layers. Specially in the Saturn, since both 2D and 3D functionalities are separated. However, at least on the Saturn, textures are stored at the same place the sprites are, and handled by the VDP.

Sprites are also handled differently from textures since they are not meant to be drawn in 3D space. You can notice in games like Tomb Raider how sprites rendered in 3D space always 'look' at the camera, and all the hardware does is some trickery with the scaling to make it look like it's further or closer from the camera (plus the Z-buffer to determine the order of the elements being drawn).

Bit blitting is a thing used when drawing sprites with certain effects such as transparency or when copying portions of another bitmap into another. Normally you would use a mask (often called "transparent color") for the transparent parts and then blit the image using an OR for the mask and an AND for the sprite, which is faster than just copying bytes. Most game consoles handled this for you through hardware.