When programming, it is often convenient to emit debugging output, and it is very easy to do so in textual format using printf, syslog, console.log and their friends. And once emitted, there are ample tools to analyse our textual output, or even to graph it (gnuplot comes to mind).
However, I'd always been stumped by creating debugging.. images. Although most debugging proceeds well via text, some things just require a 2D image. Periodically I'd delve into libpng, or even whole UI libraries, but these always presented me with too many choices to satisfy my 'rapid debugging' urge. I want to quickly get a picture, not choose from many color spaces, calibration curves, compression levels and windowing models.
Recently, I ran into the awesome story of the ray tracer demo small enough to fit on a business card, and I tried to understand how he did it. And lo, this most karmic of codes used the venerable PPM format to.. printf an image! Of course I knew of PPM, but for some reason had never considered it as a debugging aid.
So how does it work? The header (as fully explained here) goes like this:
After this header, simply print pixels one row at a time:
In addition, if you are looking for patterns in your data, Gimp 'adjust levels' or 'adjust curves' is a great way to tease these out of your graph.
Good luck!
However, I'd always been stumped by creating debugging.. images. Although most debugging proceeds well via text, some things just require a 2D image. Periodically I'd delve into libpng, or even whole UI libraries, but these always presented me with too many choices to satisfy my 'rapid debugging' urge. I want to quickly get a picture, not choose from many color spaces, calibration curves, compression levels and windowing models.
Recently, I ran into the awesome story of the ray tracer demo small enough to fit on a business card, and I tried to understand how he did it. And lo, this most karmic of codes used the venerable PPM format to.. printf an image! Of course I knew of PPM, but for some reason had never considered it as a debugging aid.
So how does it work? The header (as fully explained here) goes like this:
printf("P6 %d %d %d\n", width, height, maxColor);maxColor indicates for red, green and blue which numerical value denotes maximum intensity. If this is less than 256 each pixel is stored as 3 bytes. If 256 or higher, each color component requires 2 bytes.
After this header, simply print pixels one row at a time:
for(int y=0; y < height; ++y)And that's all. I've found that most UNIXy image viewers accept PPM files out of the box, and if they don't, pnmtopng or ImageMagicks 'convert' can rapidly make a PNG for you.
for(int x=0; x< width; ++x)
printf("%c%c%c", red(x,y), green(x,y), blue(x,y));
In addition, if you are looking for patterns in your data, Gimp 'adjust levels' or 'adjust curves' is a great way to tease these out of your graph.
Good luck!
Depending on your data, writing Graphviz Dot debug output may also make sense.
ReplyDeleteI had to add newline, like so:
ReplyDeleteprintf("%c%c%c\n", red(x,y), green(x,y), blue(x,y));
[2021] Play Slots - Videoslots.cc
ReplyDeleteBest Online Slots in 2021! Play Slots with best payouts and highest RTP! ✓ Play with real people. Rating: 5 youtube mp3 · Review by videodl.cc