ScanChess

Deadshot IO shows me that even basic mechanics of FPSs can be entertaining. You do not need a lot of complicated systems for a shooter to be enjoyable. Just decent movement, shooting, different kinds of weapons, and competitions are enough. https://deadshotgame.io/

Built ScanChess after too many OTB games were lost to phone photos and paper scoresheets — wanted a fast OCR path to FEN/PGN for players and coaches.

Really clean execution on ScanChess! Having a focused, low-friction workflow makes onboarding much smoother for first-time users. Curious how you are prioritizing power-user shortcuts versus keeping the core interface lightweight as you add new features?

The scoresheet-to-PGN direction is the harder half and the more useful one: digitising an OTB tournament game after the fact is exactly the chore that stops people from reviewing their own games. Two questions from someone who has photographed a lot of scoresheets: (1) when a player writes one move across two columns or crosses out a move and rewrites it, does the OCR follow the correction or does it produce two plies? (2) do you keep the original photo alongside the generated PGN so a coach can verify a questionable move against the paper? Being open source makes the second one easy to trust either way.

The board-photo to FEN flow is the part I would use most as an OTB player. Curious how it handles partially occluded pieces or a board shot at an angle — does it ask you to retake, or infer the position?

The scoresheet-to-PGN step seems like the place where legality checking could help a lot. If a handwritten move reads ambiguously (say Nf3 versus Nf5), does the app use the game's move sequence to pick the only legal option, or flag low-confidence moves for review? A simple highlight on uncertain plies before export would make proofreading much faster.

How does ScanChess handle messy handwriting in scoresheets — does it flag illegal or ambiguous moves for review before producing the PGN?

On a handwritten scoresheet one misread move makes every move after it illegal. Does the PGN step use legality to choose between close readings like Nf3 and Nf5, and does it point to the first move where the game breaks? A coach could then fix that one move instead of rechecking the whole sheet.

Scoresheets to PGN is the bit I'd use most. Typing up a whole tournament by hand is the worst part. How does it cope with messy handwriting or a crossed-out move?

Turning handwritten scoresheets into PGN is a nice niche since messy handwriting is where OCR usually fails. How does it handle illegal moves caused by a misread character, does it flag them for review before exporting? It might help to show a confidence highlight on uncertain moves.

Handwritten scoresheets are the part I'd expect to be hardest. What happens with a smudged move or one that's illegal in the position? Does it flag it for review or pick the closest legal move?

For board-photo imports, how are side to move, castling rights and en passant handled? Those details are part of FEN but cannot always be recovered from a photo. An explicit confirmation step would help avoid exporting a position that looks correct but behaves differently in analysis.

Separating board-photo FEN from handwritten scoresheet PGN is the right split. A coach photographing a dozen student sheets after a tournament needs the scoresheet path more than another board viewer, and a no-account browser viewer means the club can open the PGN without another signup. Are the free starter credits enough for one tournament's sheets, or do they run out after a few pages?

The scoresheet-to-PGN direction seems like the most useful part for coaches. How does it handle messy handwriting or crossed-out moves — does it flag low-confidence moves for review?

Turning a phone photo of a board into editable FEN is a practical shortcut for over-the-board players who want to analyze a position afterwards. Digitizing handwritten scoresheets into reviewable PGN also sounds useful for coaches, especially if the review step lets you fix misread moves.