Frustrating experience - mkultra

I keep finding new ways to make this app crash, and that isn't good. And I just lost 50 pages (the limit it can recognize) with this last crash. That means all the interactive learning (pattern recognition of a batch of pages) has been lost for that session. There's no recovery mode, and that sucks being the app clearly needs it.
If the scan isn't next to perfect this app will simply crash. You have to batch process your scans at 600dpi+ and perform the best contrast possible in photoshop prior to use in readiris.
It also seems to have more problems using serif than san-serif fonts. It seems less prone to error with helvetica than with times for example.
I wanted this app to work out because acrobat's own built-in OCR isn't interactive the way this app is. It gives me a chance to review and correct characters the program can't decode.
Still, my main beef with this app is the fact that there aren't proper tools for selecting items on the page. The automatic analyzation draws boxes to recognize text or images. Often in the process, it cuts off descending letters (like p, g, y, etc) so you have to manually extend the boxes below the descender letters otherwise the letter y, for example, may get cut into a v. Sometimes it even cuts off letters on the far right of the boxes. So it may well decode, for example, that the ending letter is an 'r' instead of an 'n'.

I found myself wasting time correcting the little mistakes.
To add to the frustration, the interface is horrid. Whenever I needed to make adjustment to the boxes I found myself clicking all the way up to the top tool palette that's part of the upper part of the window. The tool palette should be ideally a float window and above the document for easy, fast access. Better yet, add freggin shortcuts. I found no shortcuts in the manual to toggle between drawing text, or drawing graphics boxes. To add another layer of clumsiness to this app, you can only manipulate the boxes with these very tiny dots at the box boundary. Put some god damn handles. You can't even select multiple boxes by dragging. Instead you have to shift click boxes to erase them. If the boxes turned out too little by the automated recognition part then you can't select them. Instead you're left with a command to 'erase all small selection' which more often than not erases the wrong stuff.

The place holder for the pages isn't re-sizable. So if you have pages that have long filenames, let's say a very common name like Untitled 00.tif., what you'd see instead are a bunch of truncated names labeled Untitled.., Untitled.. you won't know which page number they are because the window isn't expandable to the right.
If you're going to make your app a whole window without detachable parts for each command, or function, then at least make their respective area re-sizable.

Another thing I found is that unlike other apps I use, whenever they use my 'canon lide' scanner for importing/acquiring files, the scanner would remember the settings (that is if you don't quit the app).
However with readiris, the behavior is exactly as quitting the app. I had to re-adjust every god damn thing on the scanner every time I would scan a page. If you tried to get around that by making multiple copies on the canon scanning software, then readiris would only read the last scan made by the canon software, not the previous scans. In other words it doesn't cue up the scans like in other programs. Readiris could very well process incoming scans in the background as you keep scanning, but instead you have to scan, process, close the scanning software, repeat the cycle ad naseum.

There may well be more problems with the software. But I had enough with it.
IRIS needs to either get a new UI designer, or keep the programmer away from UI design process.

Reply to This

Monday, May 19 2008 @ 03:33 PM PDT