Full OCR Test Last one captured input as you were writing it, this one captures it fully optically. Turn on turbomode Press space after you are done to scan and recognize Going into fullscreen may break it (haven't really tested this fully yet, expect a full breakdown on correct detection rates and other stuff soon) When you press space you have to select an area to scan you do this by clicking and holding on the left top corner(you should leave at minimum 1 pixel gap between the letter and the scan area, but it wont hurt if there's a bigger gap, it doesn't even really affect performance) then let go of clicking on the right bottom corner, after this it will try to detect it, if it comes back as "c1" with a highest percent chance of "0" then that means it wasn't able to detect anything(check if you selected the right area) ------Saffroth's Observation / Random Info Corner------ The fully optical version seems to have a better rate of correct detection than the earlier semi-optical version despite using the same sets and recognition system. also yes, it uses the exact same sets and recognition system, which is crazy. My running theory of "why" it works is that because the resized resolution is so low (16x16) it might miss the differences between fully optical and semi-optical(more like getting the exact pixel data but wtvr) The 16x16 resolution has actually been a major advantage, it's faster(despite there being 33 sets across 4 letters, it is still instant), easier to use(i can modify small stuff without having to replace the set completely), smaller(256 max pixels, though no letter should use up 256 pixels... unless it's just a cube that is perfect and divisible by 16...), and better for recognition(it struggles sometimes with some complex letters d vs b(especially with curvy d's and straight b's(how i write d's and b's on trackpad :skull:)) O yea and the weird scoring/weight/percent thing, with the +1 if it matches a pixel and -1 if it doesn't, is actually reaaaaly good, now does it look bad? maybe. but it is much much better than just +1 for a match, reason being, l and i (rn we don't have l and i, but in the future) if you were trying to write i, there is a much higher chance of it detect it as "l" than on my system
