Tired of these old LLLL OOOOOOOOOO GGGGGGGGGG OOOOOOOOOO SSSSSSSSSS LLLL OOOO OOOO GGGG GGGG OOOO OOOO SSSS SSSS LLLL OOOO OOOO GGGG GGGG OOOO OOOO SSSS LLLL OOOO OOOO GGGG OOOO OOOO SSSSSSSSSS LLLL OOOO OOOO GGGG GGGGGG OOOO OOOO SSSS LLLL OOOO OOOO GGGG GGGG OOOO OOOO SSSS SSSS LLLL OOOO OOOO GGGG GGGG OOOO OOOO SSSS SSSS LLLLLLLLLL OOOOOOOOOO GGGGGGGGGG OOOOOOOOOO SSSSSSSSSS ???? If so, then ... _wæ*°"9æ, gØ° â# _dH´ _#@ Xg#Ø gØP æ´ ø #Øùw_wwæØ@" dF dP dØ ¯¯¯ J@ J# ¸Ø´ ¸w-mw_ -g#^^-g#9^ ,a*ww_ _æw,¸wwp Æ´ _¥" §# ,#F ,#F ,/ #b ¬g#b°"¬² Æ" Ø __wÆ" ØF ØF ,P _wæP" _#P _wæwwww@ ÍL ^"¯ ØK ØK ØL·°" ÆF ¡#_ __ض#µww____òm ]#, _w d#_ d@_ ¶#_ _w JP ¬@*#P" "9**@P° °M*Æ" #° @° °M#P¯ * ___wwww___ __ww¸ _æ@°"¯¯¬"°°@*mww_ _*" J@" ___ a" w , ¬°¶@*-^°¯ _wd' dP¯¯7´ _wØ~ æ´ g° w ¯#P °^+æb*¤"W" _dK _w" dP d@ 9_ dF _,mS@ _Ø J# dØ ¬WØÝ*¤" ,# g# __wwa -g#** _ww_ J#´ _a_ _www_ _w¸ ùw_ g#´ ,#´ #¹ wæP" æc ,#F g°¯ ¬# #Ew¤"#F _æ" `#z ¤9#þ*°¶F ,#F ,#þ d# gØ° dP ØP d" °9 ØP" d@ ,@ _æM° _Ø@" ,#@ ¸#Ø ## dØ _dØ ØP z# dØ JH #F *°" ÆP ,## Ø#F ## # _w##¹ dØ |# g JH´ ,#´_ #L _ ,@ J##F Ø## ##w_w° ¬#Ñ*°¯#@~ jH*" ¶#ææ°¯ aHF #@°¯ ¬MWææ°¯ ,# V1.0 ""¯ is just for you. So what the heck is LetterMatcher ? ----------------------------------- LetterMatcher lets you create character representations of bitmap images. This way, you can design your logos in any paint program and have them converted to ascii text output. There are some programs that do this by using Amigas own ansi-sequences and building the image row by row but LetterMatcher creates ordinary ascii output, and even 7-bit if you want it. Some other novelty uses for LetterMatcher might be: - Make those wacky ``if you take a step back you can just make out the face of Queen Elizabeth II from this mess of characters'' pictures - Make large graphical banners with your text-only printer - Design IBM-character menus (for your BBS) in a paint program (if you really want to abuse LetterMatcher) How to use LetterMatcher ------------------------ For a quick list of available options you can type 'LetterMatcher' in your CLI. LetterMatcher is a CLI only utility, no other interface is currently provided. Currently the source bitmap must be in raw format. Use any available iff to raw converter to get raw bitmap data from your iff pictures. The source image must also be one bitplane deep (2 colors) as must be the source raw bitmap you pass to LetterMatcher. The command line syntax is: LetterMatcher [-stmlq] OutName Picture Width Height [FontRange [FontName]] Where: Outname is the name of the output text file created from your Picture. Picture is the source raw bitmap data file. Width is the width of your picture in pixels (max 2048). The width of your picture MUST be divisible by 8, otherwise LetterMatcher will complain. If you specify a wrong value for the width (other than the original width of your picture) you get screwy output (slantig horizontal lines). Height is the number of lines to convert from your Picture. If you specify the full height of your source Picture or greater, the whole Picture is converted. Note that this also should be divisible by 8. Height is always rounded downwards to the nearest number divisible by 8. FontRange specifies the characters to use for matching in your font. The default FontRange is 32-126,160-254 which gives you all the printable characters in topaz.font. The example FontRange 3-6,21,32-126,128-254 makes it possible for your output to contain the characters 3-6,21,32-126,128-254 and not the characters 0-2,7-20,22-31,127,255. FontName specifies the font to use for matching. The font must be 8 pixels wide and 8 pixels high and it must be nonproportional. This makes it possible to use your olde topaz and the many ibm-style fonts that come with comms packages. Default is the topaz.font. -s Strip trailing spaces from your output. -t Display the time elapsed while matching your picture in EClock ticks. -m Add a CR/LF pair after each output line. If this switch is not in the command line only LF is added. CR/LF is usefull for printing purposes. -l Loose matching of characters. This gives you a slightly different output and is about twice as fast as the normal matching. Sometimes the output with this switch can even look better. -q Suppress any CLI output but error messages. If this switch is in the command line no matching information is outputted to your CLI. Note that timing information is still printed, if the switch -t is on. OutName, Picture, Width and Height MUST be provided in the command line. The switches -stmlq are optional. The switches and options MUST be in the order specified in the command line syntax above. FontRange and FontName are also optional and if you want to specify a FontName you MUST also specify a FontRange. Theory of operation ------------------- Each 8x8 rectangle of the source bitmap is matched with all of the defined characters in the font. The character that has the highest number of matching pixels is displayed in the resulting output. Currently the program runs somewhat slowly, and I've tried to speed it up as much as I could. The innermost loop is unrolled for maximum speed under 68000 making the loop 566 bytes in size. Sorry for the 256 byte cache owners. You can roll these loops tight again. After all, the number of matches to make for a single picture is quite large (example picture 640x400): source rectangle 8x8 pixels = 64 pixels target rectangle 8x8 pixels = 64 pixels number of font characters = 256 picture width in characters = 640/8 = 80 picture height in characters = 400/8 = 50 total source rectangles = 80x50 = 4000 to match with 256 target rectangles of 64 pixels each so in total 64*4000*256 = 65536000 pixel matchings The work load is made smaller by matching totals of 32+32 pixels in each iteration thus making it down to 1024000 matches. I'm sure there is a better and quicker way to do this with a smaller amount of pixels to match, but this is how I've done it. Of course, some optimizations are made. If a perfect match for an 8x8 area is found no more characters are matched. This boosts up the matching speed of pictures that have a lot of empty spaces in them tremendously. The more loose matching method works by or'ing the matching pixels of the upper 4x8 and lower 4x8 areas of the source and the target and then counting the matching pixels. Giving a pretty good approximation of the number of matching pixels in each 8x8 rectangle. Of course, using a smaller font range will speed up the matching. Fiddling with the result ------------------------ The obvious thing to try is to generate a different output with the -l switch on. Another thing to try is to change the x- and y-offsets of your source bitmap. Since the matching is done in 8x8 rectangles you have 64 possible starting offsets for your bitmap. Each of these produces a different output. A clarifying illustration of this follows: 1 2 3 4 5 6 7 8 +-+-+-+-+-+-+-+-+ 1: : : : : : : : : : +-+-+-+-+-+-+-+-+ : 2: : : : : : : : : : +-+-+-+-+-+-+-+-+ : 3: : : : : : : : : V +-+-+-+-+-+-+-+-+ 4: : : : :X:X:X:X: y-offset +-+-+-+-+-+-+-+-+ 5: : : : :X: : : : +-+-+-+-+-+-+-+-+ 6: : : : :X: : : : +-+-+-+-+-+-+-+-+ 7: : : : :X: : : : +-+-+-+-+-+-+-+-+ 8: : : : :X: : : : +-+-+-+-+-+-+-+-+ ------> x-offset The Disclaimer -------------- If this program eats your dog or blows up your hard drive, I'm not in any way responsible. (no, I'm not responsible for any other kind of damage you may acquire whilst using this program) The Author ---------- This particularly magnificent piece of coding was done by Henri Veisterä. Version 1.0 was finished some time in August 1993. You can leave me e-mail at hveister@niksula.hut.fi if you have some comments or suggestions to make. There are no bugs in this program =). Distribution ------------ LetterMatcher is PD. Do whit it what ever you want. If you want to distribute modified versions of LetterMatcher indicate somewhere in the distribution that it is a modified version. For instance, loading of iff-pictures should be trivial to add if you already have the code ready, or use an external library.