insane idea per a new feature (OCR)

radioman

状元
No... but if I did some gesture and was immediately in the mode, then yes. Or, or maybe a separate application launch rather than Pleco proper.

Point taken on its "niche" fit challenge within the greater portfolio.
 

mikelove

皇帝
Staff member
radioman said:
Point taken on its "niche" fit challenge within the greater portfolio.

Yeah, and while in general I'm all for those sorts of features it's tough to justify them when they need to be this closely-integrated to be useful; it's very easy for us to add new features via Settings or the flashcard configuration screens (which is why they have so many darn features :) but it's much tougher to add them if they're taking up space right in the main dictionary UI.
 
If you develop the OCR for Android, will you be making it available as an IME?

This would be a killer feature. To be able to do Chinese OCR from within any Android app.

You might want to consider releasing a standalone Pleco IME app which could incorporate both OCR and handwriting features. You could also hook in Google's voice recognition which works well and can be integrated in just a couple of development hours. The IME app could also link through to the dictionaries (as shown in your OCR video).
 

mikelove

皇帝
Staff member
That's actually somewhat dicey both from a business and from a technical perspective. Even handwriting is looking like it may be problematic, in fact.

Business-wise, the handwriting and OCR features are both fairly low-margin items for us at the prices we charge / are planning to charge for them - features we developed ourselves like the document reader are naturally a lot more profitable, but even several of our dictionaries make us more money per copy than handwriting / OCR do. And that's largely dictated by the way the larger mobile content market operates - people are quite willing to pay $20 or $30 for a high-end piece of licensed content like a book / movie / season of a TV series / etc, because they're used to that, but an app that consists of nothing but code - even expensively-licensed code - has a tough time breaking the $10 mark.

So if we were to sell those modules as separate IMEs which one could then use with any Android app that one wanted (including another Chinese dictionary), we'd probably have to charge a lot more for them in order to make up for the fact that we weren't getting to sell people anything else at the same time. And I'm not even sure if Android will be able to sustain our current iPhone prices, given how difficult it is to get people to pay for Android software at all, so charging, say, $25 for a Pleco handwriting IME on Android seems unlikely to work.

Technology-wise, the big worry is that our IMEs are fullscreen and (in the case of OCR) extremely processor-speed-intensive. Each of those qualities has the potential to create *massive* numbers of compatibility issues, and many Android apps may be flat-out unable to support them; plenty of Android apps are likely to impose significant enough drains on processor speed / memory as to render the OCR module totally unusable, particularly if there are other apps running in the background at the same time.

Fullscreen handwriting is dicey on Android even against a black background due to the fact that we use OpenGL to make it run nice and smoothly - we're already worried we're going to have to come up with half a dozen different implementations of it to accommodate variations in digitizer accuracy / graphics chips, so throwing the possibility of application conflicts into the mix (can you even stack two OpenGL layers on top of each other? we can't really do that on iPhone) means we could end up spending way more time dealing with compatibility issues in our IMEs than the extra revenue from offering them would be worth.

OCR, along with the fullscreen and performance issues I already mentioned, is questionable as an IME just in general because it's really designed more for looking up words than inputting them; in our first implementation we actually didn't pop up a definition at the bottom of the screen, we just treated OCR as a fourth input method tab (along with HWR/Rad/Key) and let you tap on the screen to copy the currently-selected word to the search input box, but it didn't work nearly as well - the mode-switching time to open up the search box / frame up the camera / etc pretty much erased any advantage OCR offered over handwriting input. Now certainly an OCR mode that captured an entire line / paragraph / document could be useful in other apps, but with Android's open filesystem there's no need to do that as an IME - we'd just save the results of that capture as a text file and other apps could open that file.

So basically I'm starting to feel like cheaper, more reliable handwriting / OCR modules that work only in Pleco are a better bet than more expensive IMEs that work anywhere.
 
Top