Pixara

Most picture cross puzzles are generated rather than written, and a generated one usually has a hole in it: a place where two arrangements both fit the numbers and the only way on is to pick one. Half an hour of careful counting, and then a coin toss.

Pixara does not show a picture until a line solver has finished it from a blank grid.

That solver does only what a person does with a strip of cells and a list of block lengths: work out every arrangement the numbers still allow, keep the cells that are filled in all of them, cross out the cells that are empty in all of them, and go round the rows and columns until nothing changes. If it stalls, that picture is thrown away and another is painted. A solver that finishes also proves the answer unique, so the picture the numbers describe is the only picture there is.

WHAT IT DOES

• Every picture proved solvable by pure logic before it is ever drawn
• Three sizes: five by five, eight by eight, and a ten by ten that takes real time
• The clue for the aimed row and the aimed column is always spelled out in full and enlarged, because a paper clue strip does not fit round glass
• Aim, then tap: the first tap frames a cell, wider than the cell so it stays visible around a fingertip; the second fills it, holding it crosses it out
• Works the same with buttons on watches with no touch screen
• Filling a cell that is not in the picture costs thirty seconds on the clock
• A hint that names the row or the column proving the next cell, and then marks it
• A daily picture taken from the calendar date, the same one on every watch, rotating through the three sizes
• The grid is written down after every move, so leaving the app never loses it
• Best time per size, pictures finished against pictures started, and a run of dailies
• Five languages, on-watch settings, nothing sent anywhere

WHAT IT NEEDS

Nothing but the watch. No phone, no signal, no account.