Contained in this page is a bunch of useful code. This page is still a WIP as I pick through stuff to upload to the site.
MPEG WriterLicense: public domain
Features: Single File, 256 lines of C code, No memory allocation 
GIF WriterLicense: public domain
Features: Single File, Extremely small (398 LoC), No memory allocation 
Simple Windows 2D Drawing ShimLicense: public domain
Kodak Image Set 2x Super ResolutionLicense: public domain
|
JPEG WriterLicense: public domain
Features: Single File, Extremely small (336 LoC), No memory allocation
MPEG-1 Audio Layer 1 (MP1) DecoderLicense: public domain
Features: Single File, Extremely small (397 LoC), No memory allocation
Image ResizingLicense: public domain
PCM/ADPCM WAV File WritingLicense: public domain
| ||||||||||||||||
TGA file Writer
|
|
WAV file writer
|
|
Bilinear texture sampling
|
|
HBAO Shader
|
|
DCT/IDCT 8x8
|
|
Portable fsync()
|
|
Quality Experiments with Audio at various Bit Depths
These experiments are part of a twitter discussion on the quality of audio at various bitrates and its suitability as a super fast near lossless audio codec (a replacement for Flac) when combined with Zip or Lzma. The only artifacts from quantization is the volume of tape hiss and the dynamic range of the audio itself. Therefore if you can make the tape hiss inaudible, its pretty much equivalent to lossless.
Try our your ears at this blind test between 8-bit and 16-bit audio here. 8-bit hiss is audible with quiet audio, but at 12-bits it becomes nearly inaudible at normal volume levels.
Check out this download for some experiments. Contained in the Zip is the same audio at 1 to 16 bits per sample. The quantization method is naive and can certainly be better, but results are pretty good at bit depths >= 5.
Try our your ears at this blind test between 8-bit and 16-bit audio here. 8-bit hiss is audible with quiet audio, but at 12-bits it becomes nearly inaudible at normal volume levels.
Check out this download for some experiments. Contained in the Zip is the same audio at 1 to 16 bits per sample. The quantization method is naive and can certainly be better, but results are pretty good at bit depths >= 5.
| archive.zip |
Javascript NES emulator
VA
static const char *va(const char *fmt, ...) {
static thread_local char tmp[0x10000];
static thread_local int at = 0;
char *ret = tmp+at;
va_list args;
va_start(args, fmt);
at += 1 + vsnprintf(ret, sizeof(tmp)-at-1, fmt, args);
va_end(args);
if(at > sizeof(tmp) - 0x400) {
at = 0;
}
return ret;
}
static thread_local char tmp[0x10000];
static thread_local int at = 0;
char *ret = tmp+at;
va_list args;
va_start(args, fmt);
at += 1 + vsnprintf(ret, sizeof(tmp)-at-1, fmt, args);
va_end(args);
if(at > sizeof(tmp) - 0x400) {
at = 0;
}
return ret;
}
using C++ Lambdas as C callbacks