Latest version: https://scratch.mit.edu/projects/919975360 Added some optimizations to improve scanning speed of @griffpatch's scanner Tap on picture to scan Griffpatch's reference scanner project: https://scratch.mit.edu/projects/827805681 Camera version on https://scratch.mit.edu/projects/915814045
@Layzej/@geotale/@griffpatch did amazing advances in scanner logic it is difficult to get any improvements. Marginal perf increase about 40% from using last y lookup and combining scanpass 1,2 & 3 using a 405 (9 reds x 9 greens x 5 blues) colors1st list and then use check8 with ordered edge fallback on scanpass 2 and 3 @Geotale's binary tree search is so fast, you only can perform one extra color check before any extra logic slows the default BT search. The render method uses a branchless RLE data structure Lasty : 156k vs 205k color checks or 12s vs 14s :check8: 135k vs 205k color checks or 11s vs 14s :colors1st 101k vs 205k color checks or 7s vs 14s 20-Apr-2023: Added check7 to final ScanPass based on @chrome_cat suggestion for a good performance boost 5 May 2023: Scan 1 combined old scan 1,2 & 3 using 405 color list and added check8 for scan pass 2 & 3 Scan 2 require the full 64 color list/checks needs the 64 check 2300 times I experimented with 2 pass scan but checking 2601 rather than 405 color list, but slows the the initial scan too much on low speed devices.
