Vectorization

Convert page images into searchable text. Talk about software, techniques, and new developments here.

Moderator: peterZ

spamsickle
Posts: 596
Joined: 06 Jun 2009, 23:57

Re: Vectorization

Post by spamsickle »

I run Acrobat Clearscan after PDF creation. I do it primarily to decrease the size of the PDF. Usually, the vectorized version is 1/10th the size of the raster version. The improved appearance and OCR are, for me, welcome side effects.
scanner
Posts: 9
Joined: 14 Jun 2010, 19:14

Re: Vectorization

Post by scanner »

spamsickle
Posts: 596
Joined: 06 Jun 2009, 23:57

Re: Vectorization

Post by spamsickle »

I definitely prefer its "vectorization" to Adobe's LiveTrace. I've never used LiveTrace, but I think I'll check out Vectormagic just to play around some time. Those results are surprisingly good. Naturally, being VectorMagic's own tests, they may have cherrypicked results which their software handles better than the competitors (or "tweaked" their own, but neglected to tweak the rivals), but if it does better in general (to be determined...), it definitely seems worthwhile. I don't like the fact that you only get to "try it out" on a couple of images before you're charged, but if that model works for them, I wish them the best. I guess since they have an online option, and say you can cancel at any time, you could sign up for three months at $8/mo, and if it turned out to be unsatisfactory, cancel after the first month.

I think we may be talking apples and oranges when it comes to ebooks, however. I use Acrobat's Clearscan, and I don't think it attempts to vectorize images, nor would I want it to. I don't know if it uses the "livetrace" code, but I suspect it isn't exactly the same, whatever it does. In any case, I'm happy with the results I get from ClearScan, and think it's probably also doing things (like OCR and compression via custom font) that VectorMagic wouldn't offer in the ebook arena.
Anonymous1

Re: Vectorization

Post by Anonymous1 »

Don't bother with VectorMagic. It took me ~ 10 minutes for a single scan (5.6 MP camera, and the program scaled it to 5.0 MP), and the result was worse than potrace. I doubt that me running the software in Wine affects much, since it installed and loaded perfectly.

Potrace works well, but when letters are deformed (like on printed copies of scanned books), it separates letters into smaller chunks. I want to play with it!
DSpider

Re: Vectorization

Post by DSpider »

I just wanted to say you can also use Inkscape to vectorize images. It uses potrace, it's multi-platform, works great on Linux (as does Gimp) and out of my 6-8 hours tweak marathon last week, I have yet to discover a bug.

First select the image, go to Path - Trace Bitmap...

Here's a descriptive example, although you can find many more on Youtube:




I gave Vector Magic a shot and it's not bad... It does have to be heavily edited, tho (in Inkscape, obviously). I tried it with a few covers and it outputted ~1.5 MB files. After learning about some of the tweaks in Vector Magic and getting rid of a great deal of nodes (but preserving the layout) I got it to 200 KB. Then saved it as Optimized SVG with Inkscape and it's now 150 KB. Not bad from a 1 MB JPG. Not bad at all. But it did take me several hours to get it *just* right... and honestly, with 1 TB drive being so cheap who gives a sh*t about a 1 MB cover that you're only going to look at maybe once.
scanner
Posts: 9
Joined: 14 Jun 2010, 19:14

Re: Vectorization

Post by scanner »

I think VectorMagic was/is a MIT's project?
with whatever that implies...
Post Reply