Dewarping Kickstarter?

[feliks]

Hey all,

So I've been thinking--what about starting a Kickstarter to get Dewarping + 1 or 2 other features added?

Tulon, what do you think? I know you've said that you're no longer motivated to work on this project, but maybe this would be a way to make it worth your while? If so, how much would it take to motivate you?

I believe, and I think many would agree, that getting dewarping properly optimized is the last, major, thing to having a "finished" Scan Tailor. I am willing to set up the kickstarer, and even put $100 to get the pot going.

The only other, minor, feature that I think might be useful is to add an eraser to the output section. To open the outputs in adobe to erase either the scans of fingers, clips, etc, that have made it into the content box take a lot of time.

So that's my pitch! Kickstarter, with $100, for the following:

01. Dewarping "finished"
02. An eraser added to the output section

Cheers,

-Feliks

[feliks]

..or shall we say, get Scan Tailor a 1.0.0 GM status! So close, we're almost there!

-Feliks

[Anonymous2]

ST has an eraser, sort of. You can make a Fill Zone and set its background color to white, which works well for me.

[feliks]

Hmm, good suggestion, thanks! Although I'd still love for dewarping to be worked on

-F

[Gerard]

Hmm, good suggestion, thanks! Although I'd still love for dewarping to be worked on

-F

the developer needs examples with test cases to work on, you could collect it and start a wiki page.
the most developer that are able to handle this kind of problem, don't have a money problem, they have maybe a lack of motivation and time

just give them someone motivation by saving them some time (and making the test cases)

the problem with dewarping is, you can fix some problems with adjusting parameters, but this can make the results on other image bad, this kills motivation

[Anonymous2]

ST just needs a few more checks for dewarping.

From my experience, I've found out that dewarping rarely works for pages where the representative lines' vertical offset is less than 1/2 of the content area's height.

I think making a wiki page or a thread would be a great idea.

[feliks]

Elaborate a bit more on what this wiki page would have, and I'll get started on it!

-F

[Tulon]

Unfortunately, I am not willing to take that job, even if it pays well. I got tired of this project and money won't change that. It's not that I stopped working on Scan Tailor completely - I just spend like 5x less time on it than I used to.
BTW, I plan on releasing a new version tomorrow. It won't have any algorithmic improvements, as it wasn't the focus of this iteration. Next, I am going to focus on moving dewarping to the Deskew stage. This task is quite hard, so given the current pace, maybe in a year.

[feliks]

I'm thrilled! I look forward to the following year's work!

Will you announce it or shall we just check the source directories?

Cheers,

-F

[daniel_reetz]

We've had a couple of suggestions about Kickstarting around here, and there is definitely nothing wrong with suggesting it, but I think we need to look at the root of the problem. It's not money that's "holding us back" on software, it's raw programming talent and knowledge of Scan Tailor's architecture. Only a few people have this and, on the whole, they seem to be totally unmotivated by money. That doesn't mean we're out of luck or anything, it just means that Kickstarter is almost certainly not the right channel.

As for the parts of DIY Book Scanner that I work on (as in Hardware), money is also not my problem, at all. I have a decent day job. It's time. I put in 50-60 hour weeks at work and then come home and do 1-5 hours of DIY Book Scanner every single night. I basically eat and sleep engineering and then tech support 16 hours a day, which leaves very little to take on additional work, and does make it hard to accept complaints. If other people want to take on some of these additional tasks (not outside projects like Kickstarter, but work we need done internally, like a good dataset for Tulon, or complete documentation for the new Hackerspace Build, or someone who really wants to do Wiki maintenance) we'd be further along. That's not saying that people aren't doing those things already, many of these tasks are being done piecemeal, but we could have a more concerted effort, or have individuals "champion" a project to completion.

Another thing to point out, Tulon gets TONS of complaints here in the forum about what is "missing" from ST, or about ST "not working" - but almost no feedback when he makes a new release or authors a new feature and asks people to test. I suspect Tulon would have been able to sustain his enthusiasm if the userbase was more helpful with the development process, by testing and reporting back with more diligence. That's not saying that anyone here did anything WRONG, but it is saying that we could have done better. As it is, he's developing just for his own amusement and on his own terms, which is absolutely his right. I think Gerard said it very well - if you can do things that save time or make boring tasks easy, like building a library of good test cases, or providing consistent feedback with each release, that is worth a lot more than cash.

As I've said elsewhere, I think the way to do this is to get a lot of book scanners out in the world. The number of people who have the ability and desire to build a book scanner from scratch AND THEN develop software for it is vanishingly small. We need to get one of these things in every hackerspace worldwide so that the software-inclined people there don't have to overcome the hardware problem, and they can help us with features like these. Of course, that doesn't get you the feature you want in the short term.

The other thing I need to say is that we should be highly selective about using Kickstarter as a community, and that if anyone wants to do a Kickstarter that's about DIY Book Scanning somehow, or even mentions DIY Book Scanning, we need to make sure we have a community consensus about it. We won't use DIY Book Scanner as a fundraising platform or an endorsement unless it is absolutely the right thing to do. I almost got burned by this a few different times and I trust everyone here to be careful and respectful with our name and reputation.

Feliks, we're having a multitude of issues with the Wiki, including new user registration, so if you figure out what you want to get done there, contact me for an account. You could also start a forum thread instead and upload examples there, it would work just as well.