A few notes: PNG performance should not be much worse the xpm performance, except for the fact the images are larger amount of data (1024 bytes intead of 576 in a 24x24 image) The PNG loading is currently slower the xpm (just compare the time of running crossedit the -png vs -xpm mode). I plan to look into this some. I'm not sure how much this can be improved simply because there are more colors in the png image. There probably is room for some performance improvement. The biggest is to probably only draw the top image (ie, draw the monster, but not the grass it stands on). This would of course be an option, and IMO probably would look pretty terrible, but would get the performance up some. There certainly is some issue of minimum hardware support. I don't think anyone really expects to be able to go the store and by a modern commercial game and have it work very well on a 486. I can undertand the desire to be able to play crossfire on the system it first showed up on, but the game has evolved a lot on that time. The simple solution to this is to continue to run the version from that time frame. This is similar to the 486 example above - you can still run the games that were produced when the 486 was modern. I think it is always good for developers to try and keep performance in mind. But at the same time, if something could be done in 200 lines and run twice as fast as the same thing written in 50 lines, I would prefer the 50 line version in most cases - its less code to write, will likely have less bugs, and is easier to fix in the future. But I think the quick summary of this is: There are limited resources for working on crossfire, and so those limited resources need to be focused. If there were unlimited (or just a lot more resources), the focusing on performance for old hardware might be more reasonable. But realistically, I consider having crossfire run good on 10 year old harware still below a lot of stuff more important even if we did have a lot more resources.