Q: What is the difference between the free version (ChessOCR) and the paid version (ChessOCRProKey)?
A: The only difference is that in the paid version there is no limit to how many chess figures that are actually transferred to the pgn files or fen string. Please note that while ChessOCR only allows 25 figures to be transferred it does OCR exactly the same as the paid version. To remove the restrictions from ChessOcr you buy ChessOcrProKey and install it on the same machine as ChessOCr. You do not need to run it, just install it. Then when you invoke ChessOcr next time you will see in the activity line that he app name has changed from ChessOcr to ChessOcrPro, which tells you that all such restrictions have been removed. However, both the free and the paid versions display the recognized figures in the same fashion. Assume you own many chess books, and you want to see how well the OCR works on your books before you buy the paid version. Then just keep using the free version even if you have gone beyond 25 recognitions because you can tell if the OCR works well for a given figurejust by looking at the displayed result.
Q: I touched the screen to start the recognition process. However, sometimes the program keeps trying to recognize a figure without every returning with a result, what is happening?
A: When you touch the screen the program starts recognition. It tries to focus, then recognize, and it repeats this process until either recognition is successful, or you stop the recognition process by touching the screen again. If the program fails to recognize the figure, you should first try to move closer so that the figure fills up as much of the screen as possible, and make the book as flat as possible. You should also try to improve the lighting by moving under a brighter light or by turning on the flash to automatic in the settings menu. These three things (move closer, book flatter, better lighting) solve almost all recognition problems.
Q: I do what you just said above but still recognition fails for some figures, what is happening?
A: If the boundary of the chess figure is not completely closed, that is there are gaps in the boundary (even if they are small) then the program does not realize that it is looking at a chess figure. Therefore recognition fails for figures with small boundary gaps , and I do not currently have a good solution for this problem. However, this happens rarely, usually only for older reprinted books that have poor quality figures. An example of such a book is my copy of “Pawn Power in Chess” which has many such slightly damaged figures. The printing in older books is simply not as good as modern books, and there will be more OCR problems for such books because of the figure quality problems.
Q: For some figures an empty square has a piece in it, which is wrong; why does this happen?
A: For older books there are sometimes small imperfections or smudges in the diagram on empty squares, which are often misinterpreted as a piece. From an OCR point of view this is not so easy to deal with, so it is best if you just edit the final position by hand to remove the error.
Q: I got an error in the OCR, what should I do?
A: You could try to edit the position to remove the error a-posteriori (after recognition). But first move closer so that the figure fills up most of the screen, flatten the book and improve the lighting. These three things almost always improve the results and make editing to remove OCR errors unnecessary. And remember to set double recognition to true in the settings if you are doing OCR on older chess books. For older books of poor quality the OCR may require from five to ten tries before it has results that it considers to be reliable enough, so for such cases the OCR takes longer. If you do not get an OCR result after ten tries then you can stop the OCR since it will not work on this figure.