9.10.2011

Another fail

So it seems my compression idea was a total fail. Yep, life's like that. Though I guessed that something like that will happen, I completely finished the compressor and was only able to compress files with very large amounts of null, due their highly often occuring 00 bit pair amount. I don't see any reason for wasting time with explanations or so, just don't think you can make files smaller by masking bit pairs away. It's not worth doing so. However, I create a couple of very useful macros for manipulating memory bit-wise using a pointer-alike structure including a byte address and a bit offset. With some cleverly designed operations, I was able to the same possibilities as with normal pointers, only the retrieval of multiple bytes was too variable-demanding for a macro. So yeah, I'll probably need to move my ass on and work on other things... It's really a shame for all the energy wasted in carefully planning a compression technology that works like that. I wrote quite some functions for it, but in the it doesn't pay off except for the frame around it. *sigh*, I don't think there's any way of changing this algorithms input data for having a better bit pair layout... Who knows? I even tried to "derivate" input data using a very simple differential triangle for a more "fitting" bit layout, but this only increases entropy for already entropic files. An interesting effect, yes, but it doesn't do anything to the actual data that'd worthwhile for compression. In fact, I was able to make some files even MORE unsuitable as they were before. I wonder whether this could be of any use. Some files brought up very high differences for each derivation, so this could be of use for some kind of exotic algorithm. Whatever. I'll think a day about how this bit pointer stuff could be of use and will then continue with other stuff It's just not worth wasting time that much.

No comments: