Whether this is a good tool for you or not is really going to depend on what exactly you want to use if for. I'm using it simply to pluck sentences I like out of books and dump them into Anki. Or maybe I'll use it when I'm reading a book, and I want to run a particularly nasty sentence or kanji through an online translator (just speculating here. I haven't done that yet). Also, the text needs to be clean. It's not going to be able to read something you can't read, and it won't handle text that's been boxed in and text covered with strikethrough marks. You need clean and clear text. It can handle a single underline by ignoring it, but anything more is pushing the software.

Anything else would be faster to either scan on a flatbed or just type in yourself. Again, it depends on what you want to do with it.

It faithfully records the image you produce, then it analyzes it and tries to figure out what you did to it to screw it up, and then tries to undo it. That's how it appears to function to me, anyway. Honestly, I don't think they got too sophisticated with the hardware. I think they do all the heavy lifting in the software.

For example, when you scan vertical text, you actually rotate the book 90 degrees counterclockwise, so the text is aligned horizontally, where up is to the left, and down is to the right. Then you just scan it like a line of horizontal text. The software rotates each character 90 degrees clockwise back to vertical, and stacks the series of little images up. Then it OCRs the resulting overall image. There isn't a long wait, either. It processes the image pretty quickly.

That's at the extreme end of its performance. For most scans where the pen is wayward, it will account for the distortions a little less drastically. It has a little roller by the scan head that activates the scanner, so it has a way to know when the scan head is moving, and how fast its moving. But I think the software does the majority of the image interpretation and OCR, so I would guess that you'd need their proprietary OCR software to make it work.

Workflow issues are always going to be subject to individual variations. Once I got comfortable with the pen, getting good scans wasn't so much of an issue, although there are a lot of little annoying environmental variables that can either really make things go fast or bog things down. (Like a thick book with a very springy spine, that doesn't want to stay open for me.)

For me, it's probably about the same speed as typing, only with much less stress on my wrists and hands, and less messing around with the language bar. If you screw up, it's not a biggie. Just scan it again. They designed the scanner so if you hold it so the scan head is flat on the page, it will be a the perfect angle.

I don't have a big problem with making an extra pass or two, because the stress on my wrists is just about zero, compared to typing, which is fatiguing. I wouldn't necessarily say it's faster than the old way, but I would say that it's easier for me. I can see how it would infuriate some people.

And you point out a very nice thing it does do, and that's figure out what the correct kanji is for you, without you having to look up its proper reading so you can type it in. But a flatbed will do that for you, too.

Give me a week to work with it some more, and ask me how I feel then.

Did that help any?

rich_f wrote:

-Boxes around text will screw up the scan of the words in the box. No way around that.

Try covering the borders of the boxes with white paper.

I managed to create another 60 or so cards today with the pen. (About 150 sentences or so.) I found that the biggest cause of mis-scans is improper pen positioning, which is tricky when the book curves. I dug out some big heavy metal Japanese-style paperweights I bought in Tokyo, and they helped to keep everything in place.

At one point, the scanner stopped scanning English text completely. It would just show up as ~~~ ~~~~~ ~~~~~ and such, even if I set the native language to English. Rebooting solved the problem.

One time it suddenly decided that the pen was no longer connected. Exiting, unplugging the pen, plugging it back in and restarting the program with the pen reconnected solved the problem.

I also managed to solve the problem of scanning black text on gray boxes-- move the scan head *very* slowly. That alone will usually solve the mis-scan problems that scan head positioning won't solve.

Also, if you click on the preview pane that floats on your desktop, you can adjust its opacity so you can read the scanned text better, and see just how much you're torturing the OCR software with your bad scanning technique. Sometimes despite my best efforts, the text wanders like a drunken, wayward cow.

The best way to keep 。s from turning into o's is to move the pen slower, too.

I'm really digging the buttons on the side of the pen. Saves me the effort of hitting return and tab to do my entries in Anki, and while I found the beep annoying at first, I like it now.

Overall, I'm still pleased with it, and I really like how it helps my workflow.

vosmiura wrote:

How well does it work when scanning text with furigana on it?

It works just fine. It does *not* scan in the furigana, which is what I want. The scanner and software do a good job of ignoring it. It doesn't seem to affect accuracy.

Hey Vosmiura,
I am trying out a Demo version of the Readiris, but cant seem to get it to recognize the japaneses text.
Is your version of Readiris the Asian version? does it let you select japanese as an option?

Hmm..I guess not.
It seems to scan the image fine, but then all japanese text, except the large kanji seems to become garble.
What options are you using?

I just figured out how to make the pen scan a little better in some situations. In the software, there's a "Scan Test" option that will let you manually change things like brightness and despeckle. You can mess with the sliders while it tries to read the last line of scanned text. You can use that to fine-tune how the device reads a particular page.

In the case of UBJG, I finally got it to scan the dark text on dark gray background pages and recognize them at about 85% accuracy or so. Not bad.

I also found an article somewhere (forgot where) where the author explained why OCR software sometimes barfs when it tries to interpret Japanese and English characters in the same sentence. It has something to do with the way the software looks at the characters. Apparently JP characters are scanned differently than EN characters are.

Anyway, that explains why the pen is kind of spotty every now and then in scanning mixed Japanese and English text. It's not a huge problem, just a minor annoyance.

Ah yeah, I have the same problem with those kanji. It won't even bother to recognize 開 at all. So frustrating. It usually shows up as 聞 instead.

But the mixed forms of text appears to be a problem for most OCR software.

Question about the pen: is it able to scan unicode characters? Characters like ? ? ả ? ? ấ ẫ ỡ đ , etc, i.e. Vietnamese tone marks. 

I'm interested because I'm doing sentences from a huge variety (of very inexpensive) books for Vietnamese learning Japanese.  The problem is that typing the Vietnamese characters is, like Japanese, both a bit slow and painful on the wrists.  So the combination of the two makes it pretty slow going when entering sentences.  Any replies would be greatly appreciated.

Hmm.. says on the website that "irisPen Asian" supports 50+ different languages, but I don't see a list of the languages here.  I've sent an email to the manufacturer.  Apparently the non-Asian version supports 120+ languages, though I'm not sure why they're not listed... Maybe I'm missing something on the website. @rich_f : thanks for the reply.

IRISPen Express 6 (not the Asian version) sells for $117. It says it recognizes 128 languages. I am wondering if it can scan Japanese.

Okay, I just made a big flip-flop on this pen thing. I haven't used it in a couple of months, because I finally got sick of the cable constantly getting in my way, having to scan a sentence 2-3 times to get it right, and the random crashes. And now that I have the desk space to use my flatbed again, well...

I went ahead and sucked it up and dropped $190 and got the ReadIRIS Pro Asian 11 OCR software for my old flatbed (it's a 7-year-old Canon). It's actually a lot better than the pen for bulk scanning. MUCH faster, and more accurate. It does an excellent job at a page at a time... with a few exceptions. It makes some errors, but not a whole lot. Usually it's just the o instead of 。 errors. It's actually pretty neat how you can select the text you want to scan. (So that shoots down that 'advantage' of the pen.)

And yeah, now I kind of wish I hadn't gotten the pen scanner... although it does have its uses. Like on a long trip where you want to scan stuff, but you don't want to drag your flatbed with you, or when you're seriously lacking desk space. But that's a pretty limited application for it..

I'd say that if you were trying to decide between the pen and OCR software,  go with the OCR and skip the pen. It's not that the pen is bad, it's just a little buggy, a little awkward, and not at all faster than using a flatbed.

The pen would be handy if you're going on the road a lot, but really, if you already have a working scanner, skip the pen and just get the OCR.