> It is still pretty nasty in the 1.3 series. I got around the issue > somewhat by throwing out several letters that could be indistinguishable > after deformation. Things like "h" and "n," "I" and "1," and "t" and > "f."
Good idea! I tried throwing out all the letters and using only numbers but even then there was a lot of ambiguity between 1 and 7. I might try going back to letters and avoiding the letters that you mention. > But yeah, about 10% of the time, they get pretty bad. Yeah it gets so hard to read that you almost need software to do the recognition for you =] > > Chris Colman wrote: > > I'm using the wicket extensions CAPTCHA feature which works perfectly > > except: > > > > I know the purpose of CAPTCHA is to make the text unreadable to a > > certain extent but it's *TOO* good at making the text unreadable - such > > that humans often get it wrong too. > > > > I have a client who has complained that while's it's good that a > > computer can't decipher the letters it's bad that he has trouble doing > > it. Dang those customers! Just think how much more successful we could > > all be if it wasn't for those pesky customers... ;) > > > > I am currently using wicket 1.2.6. I was wondering if any post 1.2.6 > > releases have "improved" (depending on your perspective) the capture > > appearance to make it slightly more readable by humans. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
