OCR!

[hackinger]

OCR! Price?

Hi,

is the price for the OCR add-on already announced? (I did not find it, but may noft have searchad hard enough ...)


Cheers

hackinger

 

[mikelove]

is the price for the OCR add-on already announced? (I did not find it, but may noft have searchad hard enough ...)

We've given estimates, but we're not announcing the exact pricing until Apple approves it.

 

[dustpuppy]

You should do another youtube video to promote the app ! But it should be a funny viral one, not a tech demo.
I'm probably going to be refreshing the appstore all day tomorrow again.

 

[doscrash]

Hehe, me too. I've been refreshing the app store daily since the video first came out!

 

[mikelove]

You should do another youtube video to promote the app ! But it should be a funny viral one, not a tech demo.
I'm probably going to be refreshing the appstore all day tomorrow again.

I'd almost like to see what users can do with that - maybe we could have some sort of "post the best Pleco OCR demo video" contest; get recordings of it reading Chinese characters in famous places (the signs in front of Tiananmen, say), or of interesting people (monks, say) trying it out.

 

[dustpuppy]

Has apple purchased the OCR module yet ? I can't believe how long they're making us wait.

 

[mikelove]

Has apple purchased the OCR module yet ? I can't believe how long they're making us wait.

Nope, and it's now past 7pm Cupertino time so it looks like it'll be at least another day.

 

[YoshiCookie]

How about now?

Are we going to have to wait another weekend!? Grrr.... Haha.

 

[kun4]

While we're waiting for Apple, a small question.
I've added a pdf with Chinese characters in different point sizes. What size characters can Pleco OCR read?

 

[mikelove]

Are we going to have to wait another weekend!? Grrr.... Haha.

Still only 2pm in Cupertino, it's quite possible they could approve it by the end of the day.

I've added a pdf with Chinese characters in different point sizes. What size characters can Pleco OCR read?

It's more a question of what your iPhone can focus on clearly than anything else, but lighting conditions / paper quality / typeface can all make a difference at small font sizes - seems to work pretty reliably with type of the size you normally see in printed novels and such, but there's no simple point size or other formula that'll predict whether or not it can work well with a particular piece of text.

 

[Entropy]

Still only 2pm in Cupertino, it's quite possible they could approve it by the end of the day.

But they still haven't released the white iPhone 4!

paper quality / typeface can all make a difference at small font sizes - seems to work pretty reliably with type of the size you normally see in printed novels and such

So how's it working with McCawley's book?

And, will it have support for those "artistic" characers which are even more messy than handwriting? For that matter, do you even try to read handwriting?

~ Kiran <entropy@io.com>

 

[mikelove]

But they still haven't released the white iPhone 4!

Funny that being white turned out to be the most groundbreaking / hardest-to-get-right iPhone 4 feature... it sounds like they're changing materials again for the iPhone 5, so at this point I'm inclined to agree with the rumors that they're going to delay it one more time in the spring and have a white iPhone 5 available when it's launched.

So how's it working with McCawley's book?

And, will it have support for those "artistic" characers which are even more messy than handwriting? For that matter, do you even try to read handwriting?

Still haven't got a copy, actually - not in a place with particularly reliable postal services at the moment.

And we don't try to read handwriting, no - OCR is a very different problem than active handwriting recognition, we have our hands full even reading clearly printed characters. Artistic characters in a standard font might work (the database has coverage for a few of the big ones) but I certainly wouldn't count on it for any particular document / sign.

 

[Entropy]

Artistic characters in a standard font might work (the database has coverage for a few of the big ones) but I certainly wouldn't count on it for any particular document / sign.

Ah, I was assuming you'd have a full database for those since they seem to be common in signage.

~ Kiran <entropy@io.com>

 

[YoshiCookie]

Artistic characters in a standard font might work (the database has coverage for a few of the big ones) but I certainly wouldn't count on it for any particular document / sign.

Ah, I was assuming you'd have a full database for those since they seem to be common in signage.

~ Kiran <entropy@io.com>

書法 Calligraphy is not databased. Just like anything hand-written, all calligraphy is unique. The more 楷書-ish the font, then the more the program will be able to recognize it. The more 草書-ish it is, then the program will be much less likely to recognize it.

@Mike: Now we have to wait until Monday? Lame... :-(

Different topic: Any chance of ever adding Lesser Seal Script, Greater Seal Script, Clerical Script, Grass Script, Running Script, Oracle Bone Script... Or any combination of those? A lot to ask, I know... But there must be font/character sets out there.

Some other developers that I won't mention have started incorporating some of those fonts.

 

[mikelove]

@Mike: Now we have to wait until Monday? Lame...

Well they did just send a note apologizing for the fact that the review was taking longer than expected, and a few minutes after that we saw a purchase logged in our system of the OCR module, so I guess someone there is at least aware that it's been a while... who knows, they could approve it over the weekend, though I'd hazard a guess it won't happen until Monday or Tuesday at least.

Different topic: Any chance of ever adding Lesser Seal Script, Greater Seal Script, Clerical Script, Grass Script, Running Script, Oracle Bone Script... Or any combination of those? A lot to ask, I know... But there must be font/character sets out there.

Some other developers that I won't mention have started incorporating some of those fonts.

Do they state where they obtained the fonts from? We haven't found any yet that we could actually use without a hefty license fee - developers smaller than us can sometimes play fast and loose with copyrights, there are a few "free" English-Chinese dictionaries floating around that are actually ripped off from copyrighted titles, but we're scrupulous about all of these things and we can't use a copyrighted font without permission from its creator.

 

[Entropy]

書法 Calligraphy is not databased. Just like anything hand-written, all calligraphy is unique. The more 楷書-ish the font, then the more the program will be able to recognize it.

I'm not referring to hand-written, I'm referring to the fonts that, for example, are stamped into my friend's doormat. They sure don't look unique to me, they look like a font trying to simulate brushstrokes.

~ Kiran <entropy@io.com>

 

[mikelove]

I'm not referring to hand-written, I'm referring to the fonts that, for example, are stamped into my friend's doormat. They sure don't look unique to me, they look like a font trying to simulate brushstrokes.

True, but there's a lot more potential variation between free-flowing calligraphic fonts than between fonts that follow more rigid character styles; a 宋体 font from one foundry is likely going to look very similar to a 宋体 font from another one, and moreover will likely be close enough even to 宋体 characters from a 600-year-old woodblock print that the same OCR template would likely match all three, but in order to capture all of the common brush stroke fonts you'd probably need a separate set of templates for each.

 

[Entropy]

in order to capture all of the common brush stroke fonts you'd probably need a separate set of templates for each.

That can't be more than a few GB... can it?

(Man, amusing to say such a thing about a phone.)

~ Kiran <entropy@io.com>

 

[YoshiCookie]

Do you think it'll make it out today?

 

[mikelove]

Do you think it'll make it out today?

No way to know - they've done everything short of approving it now, so the next update we get from them should be to tell us it's approved, but that could happen 5 minutes or 5 days from now.