I found the energy to resume coding my Chinese spelling tutor app. Last night, I began to look at bitmap fonts and this afternoon, I rendered each char as a .png. Now I need to render strokes in pure B&W. with no anti-aliasing. Then tonight, will finish the compare algorithm started yesterday morning. It’s a very simple one, and I’m quite proud of it.
More later…
We’re near wrap up on the Python applet. I will be doing array comparison which is fastest but I also need to care for whitespace around the match of 1s. Otherwise it would be inaccurate lots. Half past 3pm. Fought off cancer threats made by Moslem insurgents fired up by the Liew witches. They are close to losing the battle for their lives, and control over the world thru sin.
Still wearing my Garmin exclusively since I got it last week. I have still been watching watch videos on YouTube, tho. How the other half lives and how the less affluent WISes (like me) live.
Am working hard on FastCN. Got all the bitmap arrays written to file, which is over 100MB long. Searching thru 100MB of data for a series of patterns is gonna take a while tho. It may take maybe 5 seconds for a char to be found from the stroke patterns entered.
There are waterworks going on outside our road. Seems to be peaceful. Am laying low as it seems my neighbors are plotting against me, using the workers who are rougher. Thus do the Liew witches maintain their clean hands, and screw the underclasses. It is so blatant for everyone to see.
Thinking the way to generate calligraphed characters is to compress then expand them, with a slight perspective bulge towards the left. Will try to do this in ImageMagick now, before drawing all those 38 strokes which is boring. Almost 9pm.
…
Failed to generate Chinese calligraphy. Scrap that idea for now. The applet is now operational tho rough round the edges. I tried to enter all the strokes as 3x3 matrices. Not sure if that will give an adequate match in all cases.
Mom’s undies are arriving, today or tomorrow. Hope they’re comfy. 10RM, once 60RM.
Quite a bit of tweaking to do before its ready to release as beta.
Back from a brief run. Feeling less lethargic, more alert. FastCN is running and producing results. But they're most off a bit. This is one of my smallest apps: just 100 lines of code plus a 100 line dictionary in vertical. As expected a full search thru 140MB of data does take a bit of time. Almost 6 seconds to match ONE char.
Mom’s parcel from Singapore is arriving today. Maybe 2-3pm. Taking a break after a stressful few minutes when I suspected my BT keyboard got hijacked. Had to restart and set up everything again. Wonder how this could have happened on Linux? Thought it had malfunctioned. Recovering from this shock will take an hour or so (I am living with stroke).
But I think I know what’s wrong with FastCN now: bad string offsetting.
Am tackling the offset issue now. May need to loo again later but it’s not urgent.
All's good. Now I can concentrate on coding. I never saw such a torrent of search results -there is still something wrong with the applet.
Past 9pm and the results are usually 30k characters. Something is wrong with the stroke arrays. Will fix this tomorrow morning or later tonight. I'm not sure. But the matching is pretty fast. About 3 seconds per character. And this is on a 10-year-old computer.
Almost 11pm. It's a bit late for debugging. Took my meds and will brush and turn in by midnight.
...
Made good headway on the app. Drawing the strokes now. about 70% left to go but we're getting there. Took a laxative this morning, early, as I was constipated since yesterday. Selecting / excerpting the strokes is proving tricky.
No comments:
Post a Comment