Page 1 of 1

Suggestions re Picture Zones

Posted: 06 Jan 2011, 20:53
by jimboh
I scanned a chess book with a thousand or more square diagrams consisting of chess board positions. Given the difficulty of seeing black pieces on dark squares, I was obliged to select Mixed for the output. Unfortunately, for this project, 0.9.9.2 (compiled by me for Mac OS X 10.6.5) turned some dark squares within the chess board, but not the entire chess board diagram, into a picture zone. For an example, see the screencast at: http://www.diybookscanner.org/forum/vie ... =792#p7704

As a result of this problem, I had no choice but manually to create square picture zones covering the entire chess board diagrams … hundreds and hundreds of times. In addition to being tedious, my mouse-related tendinitis flared up. I say this, not to complain, but to stress the significance of the problem. I would like to share some ideas that might be useful when we must manually create a lot of similar picture zones, for whatever reason.

It would be a nice enhancement if, in addition to custom shape drawing, several preconfigured shapes -- e.g. a square, rectangle, triangle, circle, and oval -- were offered on a palette and could be dragged, dropped, and resized to create a picture zone.

Similarly, the ability to copy and paste an existing picture zone (i.e. the dimensions, not the content) would be useful, especially if we could center its placement with a mouse click.

To deal specifically with mouse-related tendinitis, some keyboard shortcuts would be nice. Perhaps, for example, we could draw a square or rectangular picture zone by dropping a red dot with the mouse, then use the arrow keys to draw the lines in four directions. Indeed, the arrow keys might be used to draw rather complex boundaries, like an Etch-A-Sketch.

Thank you for this great program.

P.S. Pardon me. I'm new here. After posting these suggestions, I saw that the main developer does not want more feature requests. They would be unnecessary, in any event, if the problem with auto-selected picture zones, described at http://www.diybookscanner.org/forum/vie ... =792#p7704, were fixed. Not sure if that is too close to a feature request, or would require too much time. Since I am not a programmer, this is the only way I can be "useful."

Re: Suggestions re Picture Zones

Posted: 06 Jan 2011, 23:10
by daniel_reetz
I'm wondering if another utility or workflow might work for you -- your scans look super clean (or is the TIF you posted already processed output)? If the main operations you're performing are like crop and brighten/contrast, I'd bet that another utility would be less work.

Re: Suggestions re Picture Zones

Posted: 07 Jan 2011, 04:08
by jimboh
That tiff is the output from my scanner, before running it through ST. I need ST to flip every second page, crop edges, make margins equal, etc. Nothing else will perform those functions as efficiently as ST, and ST does so amazingly well. ST does an excellent job with most chess board diagrams in Black and White. However, when the dark squares are a solid color rather than the more common, closely lined background, I cannot process the diagrams in Black and White; I have to use Mixed to keep the diagrams in greyscale. The problem, in short, is that the automatic selection of the picture zones in Mixed does not work well.

Let me give you an example showing why I must use Mixed. Here is the lovely final output from ST using Mixed (with 3 of 6 picture zones created manually because the automatic zoning did not work):
214-mixed.tif
(4.66 MiB) Not downloaded yet
Here is the unusable final output in Black and White:
214 - b&w.tif
Obviously, Mixed mode is necessary.

I am unaware of any other program or utility that does what ST does as easily as ST does it. It is a great program. I will continue to use it even if I must manually create and place picture zones. I probably would not have said anything if my mouse-related tendinitis had not flared up during an especially tedious, 500+ page project full of problematic diagrams. Fortunately, most chess books use diagrams with dark squares that are easily processed in Black and White, with no special problems. This one project, however, was a real doozy.

Who knows, maybe the work being done on edge recognition or whatnot for dewarping will be transferrable to identifying picture zone edges with greater accuracy in Mixed mode? Just speculating.