Screenshot app - development and testing

[Aivoturso]

I had the IE open all the time and tried again after FF updated. No luck. However, I do not see this as a major problem. There are two easy workarounds:
1. Restart the browser.
2. Don't use IE

BTW, hotlink update was what I set out to test. Opening the app in different browsers just came as afterthought

 

[tobiasn]

2. Don't use IE

Heartwarming

 

[tobiasn]

So guys, I'm thinking we can do it like this:

I'll make 3 folders:
turns - for turnlogs
strategy - for dotmaps, war strategy, whatever
demographics - for.. demographics

each folder has subfolders that corresponds to the current turn we're on. when it comes to strategy this won't always make sense 100%, but as long as we know that the folders will be sorted alphabetically backwards, it's all good.

Key points are that it will be easier to navigate the Fast Worker site to find the right screenshot and embed code.

I'll implement the changes in Dropbox and in code later tonight or early tomorrow if this makes sense to you all.

 

[Bowsling]

Is it possible at all to be able to view spreadsheets and text documents within the Fast Worker? Or will we have to download the files?

 

[tobiasn]

Everything's possible

However, google docs makes it way easier to collaborate on stuff like that. If we don't want to use GDocs though, we could just make a folder for files like that too. But then you'd need Excel to view spreadsheets (for example). And it would be collaborative just for the people with Dropbox and the shared folder.

I might be able to iframe a spreadsheet and a text doc, and that way make it a one-stop shop for all info. That could work. It would be kinda ugly, but who cares? We could make the doc readable (security through obscurity... but i don't think anyone would ever find it without the link), but you'd need to be logged in to a google acct to be able to edit.

 

[Aivoturso]

If we use a standard format, such as CSV (ASCII only), for the demographics data it should not be a too much of a hassle to hack a system generating neatly formatted tables and desired graphs at will, right? I must admit that I'm a bit biased since I have no experience (or account) with Google docs. But still, I believe having everything in the Dropbox would make the reporting faster for the turn loggers and hence improve our chances maintaining complete and timely turnlog for the whole game.

 

[2metraninja]

Google docs is great thing, essentially an Excel put in a browser. I am not advanced Excel user, but it have all the tools I use at daily basis and most important it is always accessible and online and I think all authorized people can modify the documents there.

 

[Aivoturso]

Yes, from what I know, Google Docs is an easy to use multiuser Office platform. But my point is that as long as we are aiming to use DropBox for posting and processing the screencaps it makes perfect sense to use the same platform for demoreporting. From usability point of view it is always a suspicious idea to use several tools for achieving a single task (in this case, logging a single turn).

 

[tobiasn]

I agree, that's why I pondered on putting the GDoc in an iframe, so that you can enter your data when you go get your embed codes. It would be two systems, but it would appear to be just one.

CSV/ASCII might be a bit daunting to some, I mean, the whole point of the app is to make the user experience easier.

It's almost as I wish there was some easy way to parse the data from a screen grab. Wouldn't that be kinda awesome?

Oh btw, work was kinda intense today. I just finised (it's 00:25 atm..). So I guess I'll do the app update tomorrow. It's easy as pie anyway

 

[2metraninja]

It's almost as I wish there was some easy way to parse the data from a screen grab. Wouldn't that be kinda awesome?

That could be done easy with a photoshop action and an OCR software. Still away from fully-automatic

 

[tobiasn]

I'm thinking more some kind of PHP mod, or Linux deamon.. Doing it manually in Photoshop defeats the purpose entirely, yeah.

I'm gonna google the hell out of this one

 

[Caledorn]

Resizing the screenshots can be done easily with PHP engines like ImageMagic or similar. Parsing data from the screenshots would be extremely novel - but I have never heard of an OCR PHP mod. OCR software is generally quite expensive, but I would be surprised if there's not something out there that at least does a semigood job!

In regards to the documents thing, I am voicing in my strong opinion in favour of Google Docs which is simply put: Awesome for collaboration.

Sent from my HTC One X using Tapatalk 2

 

[tobiasn]

I like "novel". I'm testing some open source OCR package as we speak. But it's getting late, so I'll post my results tomorrow

 

[Aivoturso]

As long as the screen logger can dump the data in one place and we can get the results easily from The Fast Worker, I don't really care what happens under the hood. What I was concerned is that the turn logger would need to go to two places to enter the data. Anyway, image recognition would be awesome. I can do some googling around later today too.

BTW, if we're going to have an intra-team game before the actual game starts, it would be good if we could test the app in realistic conditions.

 

[tobiasn]

I tested the ORC just now.. Hehe. Well, you can see result here:

Spoiler :

I 1% iiii >
11 11 V1:w [44] 11 11
1' cl-m<=\ En &#8216; 11> 1 1115111111 Calendar (51) &#8216; Cruz Suenun Clnhlg: (21; ~~ 1- , 11,-, 1;,-
. c S 1 n
3 a 2 I i
11% iiii -
. W . I _ ,__ _
~ l _; __ 1 J 11,11
1' _ '
2 - ,1? 1 -1 Xi? _
_ -1 -11., _,»
.' &#8216; ' "-
cw (G111)
- M11111 G111
Mfg <;1111(1>m1)
- M11111&#8220; T1111
AM
'0
&#8216; *;&#8216;r.."&#8216;.4\
M &#8216; &#8216;
JJ,
\_\" , Crop v1=11(1111>
' &#8217; 1:1?
&#8216; - 0* f * 9
1 ._§ _ . 1 _&#8216; . > . c
1 - _-§_ g
111
- M11111&#8220; 111111111
s1111=1=
1.1.1 1.1.
. 2 *7.
_ 'Squlr:km
&#8216; "- "1. ;
r' ' _&#8217; _
&#8216; &#8216;J11 315-?
£4&#8216; 51".. "T &#8220;&#8216;;&#8216;?_
1111111111
Appruv&#64258; 11&#8220;
1_1= EXp!ln(y
- Y1&#8220;
11111117 Irnpan:
- M11111&#8220; Gold
&#8216;Ii
,.
Q .
4i~
4 ____
41
141
111111
112111
1111111
51%
11
14
&#8212;..<»-=-_+-=\--,.~&#8217;7- -
51 41
15 11
221111 115142
111111 122151
1111111 111142
11% 11%
11 12
15 '2
[I&#8216;91'1'1:1g1'1p\'111:&#8216;: '~ '11» 11:11 E1» F&#8216;1~1,~z1 fwerige F&#8216;1~1,~z1 111.111 F&#8216; =11:
1 11 ' ' H&#8216;: 1 &#8216; 45 2
11 1
11 1
152111 1
11111 1
111111 1
54% 1
11 1
'21 2 1
:1 1
1:1
17",
, 1I"'~"
". 1 ~ 1*&#8216; » . Ga. 1;" » I &#8220;R - &#8216;
' n
l'I*l
c
L32
1.
1 1
J&#8216;
4;&#8216;
_ w
'l"I 1 C_'IiL /1 CL _ L'l&#8216;_#&#8216;_'I_L'l_CL E1"? ' 3 W
_ 411
" 4 I &#8220;&#8216; . 1 -W11, -"J? " av
1-\
V \
Di
&#8216;&#8212; .
Q .
I&#8217; " &#8216;
111111.11... 1111.1. 11
1111111 Ernplrc
1>1».1> 11.511511 1111,11"
.1?&#8217;
7 '1' 7' 11
O1
Q7
5'1!

This was just a whole screen with lots of data floating around the demographics.
This OCR engine is trainable, and this test run was just straight vanilla, something that tells me this is definately doable, but might need some work.

Installing and testing tesseract was very easy btw.

 

[2metraninja]

I knew the results from a demoscreen will be like this ( I use quite often OCR) and thats why I suggested slicing the screenie in to strict zones with photoshop action. I am away from any PHP but I think PHP can make it even easier. Cutting the big demographics screenshot in to small discrete pics named and numbered so the OCR knows what where to put the data he gets from each and every piece is something I can imagine can be done with PHP.

 

[tobiasn]

Yeah that could be done with the ImageMagick API. Problem is actually that the images are probably bound to be slightly different - different screen size, different local setting on font size, and whatnot. So these actions we're talking about probably needs some low level image analysis before they can run.

I'm interested in this from a professional viewpoint, so I don't have anything against running some tests. But I don't have any real hope that it will actually work

I'll come up with some contingency plan involving Google Docs iframed, probably.

I'll also make the changes to the app now, and remove the dummy data, so that we can test TFW in the pitboss, as requested.

 

[2metraninja]

Yeah, I though of that - screen size can be set to something sustainable by most/all of the turn reporters/demographics data collectors.

About fonts, I dont have answer ready Maybe leave the slices slightly bigger to avoid possible cutting off of letters and see how it works?

 

[YossarianLives]

Hi Tobiasn. I am very clueless to everything that has been discussed in this thread, but appreciate the time and effort you've put into this app. I was wondering if you could explain again exactly what it does? Do turnplayers save images into Dropbox, and then the app automatically posts them into a dedicated thread (that has not yet been created) here in our sub-forum? What still needs to be done so that we can start implementing this?

I created an account on Dropbox using your link. My email is vhinderl at gmail. What happens now? 2metra posted some screenshots into Dropbox that I'd like to take a look at, but I don't know how to do that. Please help! Thanks!

 

[tobiasn]

Hey guys,
Sorry I've been absent, I had to take a impromptu business trip last week, and started my vacation this week.

YossarianLives, yeah the app works like this:
- You put screenshots in a designated folder (folder structure, actually) within the TheFastWorker folder in your Dropbox.
- There's a small mashup between the Dropbox linux deamon and some PHP code that shows the images on web-page. Right now it only works for the turns-folder. I'm working on the demographics folder too.
- There's some rewriting going on, so that the "latest pic" always has the same URL (useful for auto-updating threads)

Anyway, I'll move the app to a new server, and fix the demographics thing, and hopefully that will work better for people.

Also, I need to read like ten thousand forum posts to catch up with all the action