https://homeostasis.scs.carleton.ca/wiki/index.php?action=history&feed=atom&title=Game_Engines_2021W_Lecture_20 2026-04-08T03:16:30Z Revision history for this page on the wiki MediaWiki 1.42.1 https://homeostasis.scs.carleton.ca/wiki/index.php?title=Game_Engines_2021W_Lecture_20&diff=23101&oldid=prev 2021-03-25T14:07:06Z

<p>Created page with &quot;==Notes== &lt;pre&gt; Compression in Game Engines --------------------------- How much do you know about compression? - lossy vs lossless? * huffman encoding * perceptual coding...&quot;</p> <p><b>New page</b></p><div>==Notes==<br /> <br /> &lt;pre&gt;<br /> Compression in Game Engines<br /> ---------------------------<br /> <br /> How much do you know about compression?<br /> - lossy vs lossless?<br /> <br /> * huffman encoding<br /> * perceptual coding<br /> * lempel-ziv<br /> <br /> Why?<br /> - games care about performance, space and time<br /> - compression trades off time for space<br /> - but sometimes you effectively get both<br /> <br /> lossless compression<br /> - compressed version can be used to reproduce original bit-for-bit<br /> - zip files<br /> <br /> lossy compression<br /> - some information is thrown away during compression<br /> <br /> Information theory<br /> - information: number of bits needed to represent something<br /> - but, LEAST number of bits<br /> - so part of information theory is always compression<br /> <br /> Maximum information is from random processes<br /> - random bits can only be reproduced by copying them, no algorithm<br /> can do so without being equal or bigger than the bits<br /> <br /> Run length encoding<br /> - represent repeated runs explicitly<br /> - i.e., say 20 0&#039;s instead of 00000000000000000000<br /> - only works if you expect to have many runs<br /> - if you have few, you can actually make the output bigger than the input<br /> <br /> All compression algorithms try to transform data into a representation equal to the amount of information in it<br /> - i.e., the theoretical amount of information = size of data<br /> <br /> An approximation of the amout of information in something...how big is it when you compress it!<br /> <br /> With lossy compression, we throw away information<br /> - output doesn&#039;t have all the information of the input<br /> <br /> lossy compression seems like a bad idea<br /> - don&#039;t do it with code!<br /> <br /> idea: throw away information that won&#039;t be missed<br /> - patterns that are imperceptible<br /> - so, you have to have a model of what is perceptible!<br /> <br /> For audio, lossy compression is common because our ears can&#039;t detect many<br /> patterns in sound<br /> <br /> MP3 was developed by someone at the Fraunhoffer institute, listening to a song by Suzanne Vega<br /> - did it until he couldn&#039;t hear any artifacts after compression, i.e.,<br /> the frequencies that were thrown away were imperceptible<br /> <br /> More recent codecs, i.e., AAC, use more sophisticated models of human perception<br /> <br /> Same thing for images and video<br /> - what patterns do people not notice?<br /> <br /> More sophisticated perceptual models allow for higher compression ratios but at the cost of more computation, especially during compression<br /> - typically compression is much more expensive than decompression<br /> - but the tradeoff varies<br /> <br /> <br /> Latest trend in compression is AI-assisted compression<br /> - e.g., feed data into a neural net, let it reproduce it<br /> - neural net identifies patterns that can later be reproduced<br /> <br /> most compression is based on some form of dictionary<br /> - identify the &quot;words&quot; that make up the document<br /> - then compress by storing the dictionary and then which words are used, with<br /> the words identified as indices in the dictionary<br /> <br /> Can lead to great compression if you use a pattern recognizer, but artifacts get very weird<br /> - instead of weird blocks, you get weird faces<br /> - you see this in image enhancement, which is really just reverse compression<br /> <br /> <br /> As a game developer, why do you care about all this?<br /> - need to choose the right compression for game assets<br /> - PNG vs JPEG (lossless vs lossy)<br /> - wav vs ogg/mpg/aac<br /> <br /> You&#039;re making a space vs time tradeoff<br /> - bigger files for faster processing<br /> - but the cost of bigger files can be more of a cost than the processing<br /> <br /> PS5, XBox Series X<br /> - part of performance gain is hardware decompression<br /> - so you actually get better performance with compressed assets<br /> - particularly important for textures<br /> <br /> Why do games get so big?<br /> - a large part is using uncompressed assets, duplicated assets for performance<br /> - duplication is for slow seek times of spinning hard drives, optical media<br /> <br /> But new consoles have SSDs and hardware decompression<br /> - so can get smaller games that perform better<br /> <br /> compression has big future as perception really is compression<br /> &lt;/pre&gt;</div>

Soma