Problem statement: Diagrams for
chess illustrations and problems exist in numerous digital graphic formats - in pdfs, web pages, jpgs, gifs, etc. Nearly all modern chess programs recognize
FEN (Forsyth-Edwards Notation) notation which can greatly facilitate the setting up of these problems in a chess program for analysis, which would otherwise take much manual setting up of pieces in the UI. Many, if not most posted diagrams do not have available with them the FEN notation which would easily allow the setup of those diagrams in a chess program unfortunately.
Solution: Develop a screenshot capture utility to import a graphic of a defined chessboard area on a computer screen and using optical character recognition-type technology convert that diagram to FEN notation which would then be sent to the user clipboard for easy pasting into a computer chess program.
The user would have to be able to specify whose move was next (Black or White) and if castling is still available since this cannot always be determined just by reading a diagram without any context.
It seems that most of it would be trivial except for the OCR. Perhaps it would be possible for it to have an internal database of all the popular chess piece "fonts" and compare squares in the grid with that and discern if they are black/white, lighter/darker and also allow for user correction of the piece placement in case of error after it had performed its analysis.
I realize this might take more than a few hours to code, but I do think that it has potential to become widely popular among chess players, who would be willing to donate a few dollars for it.
This idea came to me while using "Screenshot Captor". Thanks for that, BTW
Currently, the closest utility is
DiagTransfer, which allows a user to do it more easily, but still manually, by overlaying the diagram with a transparency where pieces can be manually set up without switching screens. It is able to export FEN.
There was an attempt at such a utility as evidenced by
this forum post, but it is no longer available and was probably never really finished. I have tried it in the past, though it did not work very well at all for me. A couple obvious problems with it if memory serves were that it was not a screenshot capture utility, instead opening graphics files and converting them to what seemed to be a very low bit grayscale blob which I could not even then read. In other words, it was clunky and did not work.
Thanks,
Jay Coover
