ObjReader Community

Discussion => Post Your Questions & Comments Here => Topic started by: Michael Lobko-Lobanovsky on January 10, 2017, 02:47:53 am

Title: PNG Optimization
Post by: Michael Lobko-Lobanovsky on January 10, 2017, 02:47:53 am
Quote from: Patrice
The gem stone animation is one of the biggest i have, 30000×500 pixels in 32-bit, go figure.

Patrice,

Are you aware of quite a number of SW packages around that could optimize your PNGs efficiently while preserving their alphas where available? And in many cases you might even resort to lossy recompression (https://pngquant.org/).
Title: Re: PNG Optimization
Post by: Patrice Terrier on January 10, 2017, 08:53:21 am
Mike--

Thank you very much for the link, i shall try it.

While very interresting to produce smaller disk file, that won't change anything when loaded in memory, because then everything must be turned into 32-bit DIB or standard BITMAP.
Title: Re: PNG Optimization
Post by: Michael Lobko-Lobanovsky on January 10, 2017, 05:36:57 pm
... that won't change anything when loaded in memory ...

That's correct, and the problem seems to have no solution in programmatic user space.

However both DX and OpenGL (https://developer.nvidia.com/astc-texture-compression-for-game-assets) support a number of image formats (including DDS/DXT) that allow for on-the-fly decompression on the GPU hardware only when the respective texture is needed for a particular render pass.