Cardboard box training rig - first book done!

[rob]

Also, don't forget, I think Scan Tailor is optimized for "real" dpi in some of its algorithms, so be sure you actually measure your dpi. Don't just guess!

[600dpi]

Rob & Heelgrasper,

Thanks for your help and advice. I think the DPI values came from the information screen in Irfanview, but clearly their wrong! It shows 180 x 180 and I had to tweak the second up to 204 before ST would enable the OK button. Last night I played about a bit and with some better values got a much better PDF that seems to behave in the Android tablet and is quite quick and responsive loading and navigating. But I'd like to know what the real DPI is...

So how to Calculate the DPI more effectively..... I thought about it and then knocked together a "DPI Calculator", I guess there must be some about? Here's my effort:



You load a typical image. In this example it's just a blank page from a paperback (autofocus is on so obviously it's out of focus). Also, I'm using the images after they've been rotated by the Homer bash script (in the \renamed folder).
You measure the books length with a ruler (it must be a plastic one).
You check the pixels for the length of the image.
Enter the values in the DPICalculator.
Adjust the spin boxes so that the brown bar (which represents the book) is a good match.
Click "OK" and voila!

In this example it turns out to be 400.

Some inner workings...
It just uses the proportion of the 'brown book bar' and the grey areas above and below to the left of the image to work out the pixel length of the book, then divide by it's measured length...

nMeasured = 8.45 // you must use a plastic ruler
nImageHeight = 4000 // from file explorer details view
nTotal = 407.00 // total height in screen pixels of the 'bars' to the left of the image
nProportion = 0.85
Book Pixels = 3380.84
DPI = 400.10

Not sure if it's right though?

[600dpi]

Run into a problem:

I'd spotted that the version of ScanTailor that comes with the Homer suite is the beta version. So I upgraded to the current version.
But now it stopped working! I get the following error message:

!The following file could not be loaded: <file.jpg>

This displays in the central image pane (and where <file.jpg> is a real jpg with full path etc).

I tried re-creating the projects from scratch using the images output from Homer option 1 (rotate and renamed), but get the same result.
I tried un-installing and re-installing ST, but to no effect and still get the error message.

Does anyone have any ideas about this?

[dpc]

Another person with a mysterious problem loading images - weird. Since you can write apps, why not attach a debugger to ST and set a breakpoint where the image loading occurs and then see why it's failing? Should take all of 5 mins to determine the cause.

You could automatically determine the DPI by framing your shot for a particular book, then put a piece of graph paper with known grid size in its place and take a shot. The app would just look at the resulting image and count the number of pixels per grid(s) to determine DPI.

[600dpi]

Another person with a mysterious problem loading images - weird. Since you can write apps, why not attach a debugger to ST and set a breakpoint where the image loading occurs and then see why it's failing? Should take all of 5 mins to determine the cause.

Yeh, I'd thought about that.... I did find references on the SourceForge to ST in the early release versions (and presumably the beta version) not un-installing correctly, so I'm wondering if some DLL file or other from the beta version has been left behind (there were also a bunch of entries in the registry)...