New screenshot web API

[dancmc]

OCR Web API

I decided to port my OCR app to a web version so it's more useful. Taken a while because I had to learn python, basic html/js, and hosting on an unmanaged VPS from scratch haha, but was a very educational project. There's a lot more room for improvement with the OCR pre-processing, but image processing theory is pretty complicated.

I've written a readme on the project page @ https://github.com/dcmc87/MFFRosterParserWeb, and the POST url is www.dancmc.io/mff/ocr. That address is also hosting a functional demo website where you can upload screenshots to test.

I'm thinking you could ideally build a front end UI on your website and just POST requests to my link, it'd probably look much nicer than anything I could make and allow direct integration with the results.

Issues

1. One issue with copy and pasting JSON generated by me into your site currently is that it overwrites all character info. For example, if I post character detail data in, it sets all gears to 0, and vice versa. One way around this is to allow importing of JSON fragments, or to have an option to not replace values that are 0 in the JSON.

2. I realised there are some duplicate gear names between characters, what I've done on my demo page is to display the thumbnail sent back from my server, and ask the user to choose between the possible characters.

3. If you build your own frontend, you could aggregate all the results after the user has resolved any duplicate issues, and apply it directly. Or even highlight missing values for users to fill in (JSON contains 0.0 or "" if OCR cannot recognise some words/numbers)

I'd be happy to make any changes to my result data structure, just let me know =)

[VLstack]

Wonderfull. I have no time to check out today, but tomorrow week-end I will for sure. Sounds very interesting.

That is an awesome news.

[VLstack]

I lied :) I took a little time to check out and so far it is great. And I will definitely use it on the site.

However I have a few "enhancement requests" to be able to use the API. Do you want us to discuss about enhancements and sync our data here or on your repository ? I think your repository is more dedicated for it, but I will follow your recommendations.

Can't wait to sync my tool and your API, it will be very great.

[dancmc]

Thanks, yes we can discuss it on my repo. And definitely just let me know what enhancements you need and I'll try to add them, it's my first time building something like this so I'm sure there are a lot of things to improve on =)

By the way I'm probably going to have to retrain the font and recalculate all the bounding rectangles when they completely revamp the UI next update lol

On 15 Apr 2017 01:42, "Laurent Vilday" notifications@github.com wrote:

I lied :) I took a little time to check out and so far it is great. And I will definitely use it on the site.

However I have a few "enhancement requests" to be able to use the API. Do you want us to discuss about enhancements and sync our data here or on your repository ? I think your repository is more dedicated for it, but I will follow your recommendations.

Can't wait to sync my tool and your API, it will be very great.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-294199655, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs6Ucdz7-UmGldZ6St6zTb38XQSFaks5rv7AWgaJpZM4M62Ii .

[VLstack]

Great. I have prepared my suggestions to help me call your API from the site. I am creating "issues" on your repo.

[VLstack]

UI so far. Of course work in progress :)

[dancmc]

Looks great! I've finished reconfiguring my api response by the way, Have a look at the readme for the single file mode and ignore the multi file which returns the old response. The endpoint is still www.dancmc.io/mff/ocr

I'm currently trying to figure out if my server settings are causing some delay in file uploads. A typical response takes about 2 seconds on my test page using my server, but less than 500ms on testing on localhost. Granted there's some upload time and server latency, but for a <100kB file do you have any idea if the 1.5s lag is simply due to this? Using Apache default settings

[VLstack]

I have no idea about server architecture unfortunately.

[VLstack]

Almost completed. I hope I can finish it by tomorrow evening and let you play with the beta version :D Thanks again, that is a very huge thing to have for the tool.

[dancmc]

I've finished migrating to my new server, but the DNS change has only propagated to about half the world at the moment. You can post requests directly to the IP at http://173.212.216.199/mff/ocr in the meantime

[dancmc]

I'm also working on getting SSL set up again, you might run into warnings for now trying to post to http from your site

[VLstack]

Sorry for the delay, it is almost ready for first tests. Details is ready, gear is on the way.

[dancmc]

Take your time, no rush at all =)

On 20 Apr 2017 1:43 AM, "Laurent Vilday" notifications@github.com wrote:

Sorry for the delay, it is almost ready for first tests. Details is ready, gear is on the way.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-295362089, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs5mJOWhuTKz1Qwx-QbaTGFU-fRz9ks5rxke3gaJpZM4M62Ii .

[VLstack]

Details : ready Gear : ready Remove all errors : ready Apply all : in progress

Way more complicated to check everything than I thought at first :p

[dancmc]

Haha features often turn out more complicated than we think. I spent ages trying to make the base OCR software read it accurately and failing, then I finally sat down and learnt how to do the custom font training and solved it all in a day

On 21 Apr 2017 02:00, "Laurent Vilday" notifications@github.com wrote:

Details : ready Gear : ready Remove all errors : ready Apply all : in progress

Way more complicated to check everything than I thought at first :p

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-295835804, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs8pvzWqY3zUN3eJMYkEDFppdYP64ks5rx51TgaJpZM4M62Ii .

[VLstack]

Beta test is available online https://www.mokhet.com/MFF-beta/

You should export your data before testing, safer to save it :D

So far, only tested on Safari Mac. I am going to do some tests with Chrome Mac and Firefox Mac later tonight, and tomorrow at work I will be able to do tests on windows for IE11, Edge, Chrome and Firefox ESR. I hope it is working good enough for the first tests.

[VLstack]

Feel free to tell me if anything can be improved/updated. I hope it is simple enough, but you might have a fresher eye on the UI.

[VLstack]

Oops, fixed an issue with gear position 1 (which in fact is position 0 in array and did not fit correctly my checks).

Beside this bug, all my tests are ok for Firefox, chrome and safari Mac. I hope I will have same successes tomorrow on windows 7 and windows 10

[VLstack]

Arghhh and now I find a bug in chrome. grrr I definitely hate this browser.

[VLstack]

Fixed :D Yoohoo, it was not a bug related to OCR, sadly it was already there before :(

[dancmc]

Whoo looks good. I also fixed a bug on my end.

OK some thoughts from my limited testing :

1. Could you put a regular file input multiple button element there as well in addition to the drag and drop? I often have my browser on Mac running in full screen and it's quite difficult to do drag and drop

2. It doesn't seem to be doing any file compression client side and just uploading the full size pics, which is slowing uploads and taking up disk space =p I wrote a some JS to do compression of image file arrays, let me know if you want the code

3. In the gear upload result, maybe should let people change the attribute type instead of just the value? Also for blank results, maybe put a couple of blank boxes for input or remove the word invalid? Because say my gear level is +12, it's correct that the last few attributes would be blank, but putting invalid property may make people think there was an error processing

Other than that looks fantastic for now =)

On 21 Apr 2017 07:21, "Laurent Vilday" notifications@github.com wrote:

Fixed :D Yoohoo, it was not a bug related to OCR, sadly it was already there before :(

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-295963151, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs9jpCB9EDb3DEI9j-2WGYxs9SlD_ks5rx-hkgaJpZM4M62Ii .

[dancmc]

Oh yeah the other thing with point 1 is that you can't drag and drop on iPad/phone, so a button is needed

On 21 Apr 2017 11:41, "Daniel Chan" dcmc87@gmail.com wrote:

Whoo looks good. I also fixed a bug on my end.

OK some thoughts from my limited testing :

1. Could you put a regular file input multiple button element there as well in addition to the drag and drop? I often have my browser on Mac running in full screen and it's quite difficult to do drag and drop

2. It doesn't seem to be doing any file compression client side and just uploading the full size pics, which is slowing uploads and taking up disk space =p I wrote a some JS to do compression of image file arrays, let me know if you want the code

3. In the gear upload result, maybe should let people change the attribute type instead of just the value? Also for blank results, maybe put a couple of blank boxes for input or remove the word invalid? Because say my gear level is +12, it's correct that the last few attributes would be blank, but putting invalid property may make people think there was an error processing

Other than that looks fantastic for now =)

On 21 Apr 2017 07:21, "Laurent Vilday" notifications@github.com wrote:

Fixed :D Yoohoo, it was not a bug related to OCR, sadly it was already there before :(

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-295963151, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs9jpCB9EDb3DEI9j-2WGYxs9SlD_ks5rx-hkgaJpZM4M62Ii .

[VLstack]

1) Of course, what was I thinking about ? Easy fix, will add the multiple browse button under the previews.

2) If you already have a js code for this, I will happily use it instead of digging the internet to find something usable.

3) Once again all valid points, I did not think about characters with gear below 20. I'm so use to track nothing on those characters until they reach 60/20/20/20/20 :)

All are valid points, I will address them tonight when I'll be back from work, but for now I am so very late, time to leave (run) house :D

[dancmc]

Thanks! I'll have the js code up in a separate GitHub project and link you in, currently working fine but I'm trying to figure out how to use web workers so the browser can compress multiple images concurrently

On 21 April 2017 at 14:50, Laurent Vilday notifications@github.com wrote:

1.

Of course, what was I thinking about ? Easy fix, will add the multiple browse button under the previews. 2.

If you already have a js code for this, I will happily use it instead of digging the internet to find something usable. 3.

Once again all valid points, I did not think about characters with gear below 20. I'm so use to track nothing on those characters until they reach 60/20/20/20/20 :)

All are valid points, I will address them tonight when I'll be back from work, but for now I am so very late, time to leave (run) house :D

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296096671, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMsy_eABYSsfNXT-Eh-GJLIX91NNuSks5ryFGogaJpZM4M62Ii .

[VLstack]

1. issue #46 : fixed
2. issue #47 : implemented
3. issues #48 and #49 : implemented

[dancmc]

Sorry about the delay, finally got stuff configured the way I wanted. Compression code is at https://github.com/dcmc87/JSImageCompress. Demo is at https://dancmc.io/projects/JSImageCompress.

I'd just use the imagesToResizedBlobs method and submit each result.blob

[VLstack]

np :)

Do you mind if I create my own reworked branch ? My linter is not pleased with the code style and global methods :)

[dancmc]

Haha sure go ahead and do whatever you want with it. As long as you send a blob or base64string my server is happy haha

On 22 Apr 2017 19:04, "Laurent Vilday" notifications@github.com wrote:

np :)

Do you mind if I create my own reworked branch ? My linter is not pleased with the code style and global methods :)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296365763, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs1oyt6rLdYDzsbnF1m_XQezCUVSGks5ryd6qgaJpZM4M62Ii .

[VLstack]

All done and live on https://www.mokhet.com/MFF-beta/

For the compression options, I have used { "maxSize" : 1, "speed" : 10 } with default units to "mb"

I hope your server will like it better :p

Tell me if you want me to reduce the maxSize or add another specific option.

[dancmc]

Max size should be 0.1 (100kb images have all worked for me so far) and speed can probably be around 5 (1 is fastest) though honestly 2-3 would be fine too

On 23 Apr 2017 00:50, "Laurent Vilday" notifications@github.com wrote:

All done and live on https://www.mokhet.com/MFF-beta/

For the compression options, I have used { "maxSize" : 1, "speed" : 10 } with default units to "mb"

I hope your server will like it better :p

Tell me if you want me to reduce the maxSize or add another specific option.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296386275, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMszDxVB68Xn8oLgz7KlMW6WmBCXjaks5ryi_dgaJpZM4M62Ii .

[VLstack]

Done, updated to { "maxSize" : 0.1, "speed" : 3 }

If we happens to have too many fails we will see :D

[VLstack]

So far, all my tests are working very great. This is going to be so simple now, your work is really amazing.

[dancmc]

It's working really well =) But I'm still receiving the full sized image on the server side, I'm wondering if something went wrong with the compression code or I'm not loading the new js file or something?

On 23 Apr 2017 01:32, "Laurent Vilday" notifications@github.com wrote:

So far, all my tests are working very great. This is going to be so simple now, your work is really amazing.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296388977, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMszf_8i7SZ8YhlLLZig2eb7jls8kWks5ryjmygaJpZM4M62Ii .

[dancmc]

Ok it's not a problem with cached code, I'm scanning through the compression code in inspector to see if I can figure out the issue

On 23 Apr 2017 01:38, "Daniel Chan" dcmc87@gmail.com wrote:

It's working really well =) But I'm still receiving the full sized image on the server side, I'm wondering if something went wrong with the compression code or I'm not loading the new js file or something?

On 23 Apr 2017 01:32, "Laurent Vilday" notifications@github.com wrote:

So far, all my tests are working very great. This is going to be so simple now, your work is really amazing.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296388977, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMszf_8i7SZ8YhlLLZig2eb7jls8kWks5ryjmygaJpZM4M62Ii .

[VLstack]

Hummm. this probably a mistake on my side. I'll check in half an hour when I'll be back.

-- Vilday Laurent

Le 22 avr. 2017 à 19:44, dcmc87 notifications@github.com a écrit :

Ok it's not a problem with cached code, I'm scanning through the compression code in inspector to see if I can figure out the issue

On 23 Apr 2017 01:38, "Daniel Chan" dcmc87@gmail.com wrote:

It's working really well =) But I'm still receiving the full sized image on the server side, I'm wondering if something went wrong with the compression code or I'm not loading the new js file or something?

On 23 Apr 2017 01:32, "Laurent Vilday" notifications@github.com wrote:

So far, all my tests are working very great. This is going to be so simple now, your work is really amazing.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296388977, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMszf_8i7SZ8YhlLLZig2eb7jls8kWks5ryjmygaJpZM4M62Ii .

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub, or mute the thread.

[dancmc]

I found the bug =)

In compress.js, in resultArray.push, you put "file" : file, but that's pushing the original uncompressed file. I think you meant to push something like dataUrlToBlob(dataurl) instead haha

On 23 Apr 2017 02:02, "Laurent Vilday" notifications@github.com wrote:

Hummm. this probably a mistake on my side. I'll check in half an hour when I'll be back.

-- Vilday Laurent

Le 22 avr. 2017 à 19:44, dcmc87 notifications@github.com a écrit :

Ok it's not a problem with cached code, I'm scanning through the compression code in inspector to see if I can figure out the issue

On 23 Apr 2017 01:38, "Daniel Chan" dcmc87@gmail.com wrote:

It's working really well =) But I'm still receiving the full sized image on the server side, I'm wondering if something went wrong with the compression code or I'm not loading the new js file or something?

On 23 Apr 2017 01:32, "Laurent Vilday" notifications@github.com wrote:

So far, all my tests are working very great. This is going to be so simple now, your work is really amazing.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43# issuecomment-296388977, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMszf_ 8i7SZ8YhlLLZig2eb7jls8kWks5ryjmygaJpZM4M62Ii .

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296390758, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs7c2xm9WnEOREl_gHiWLxikbfH5Qks5rykCigaJpZM4M62Ii .

[VLstack]

Yes, I just figured out this 15 seconds ago, working on it. Will be ready in a few :) sorry.

[VLstack]

Fixed, should be better. I hope :D

[dancmc]

Yup everything's working perfectly now =) =)

Btw, is there any purpose to the window closing in 5 secs after pressing apply?

On 23 Apr 2017 02:29, "Laurent Vilday" notifications@github.com wrote:

Fixed, should be better. I hope :D

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296392390, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs5UtT3mhWXDwaewAcJsSHM4xrRPzks5rykcCgaJpZM4M62Ii .

[VLstack]

Great.

Only purpose for the 5 secs delay is to let the user acknowledge everything is complete. At first I was automatically closing the popup and it felt like it was bugged.

[VLstack]

OMG again, a problem with chrome. I so definitely hate this browser.

The image is correctly parsed from safari and firefox but got errors with chrome (all on Mac).

Do you think it is related to compression ?

• Original screenshot :

• Safari result : good

  

• Firefox result : good

  

• Chrome result : some mistakes

  

[dancmc]

That's really strange. I had no issues so far with chrome on windows, safari on Mac and chrome on android. Will test on chrome Mac when I get home and see what's going on.

It sounds like I may need to tweak something on the server side. Possible that chrome Mac is using a less optimised jpeg compression so the image isn't quite good enough, but will have to check

On 23 Apr 2017 03:12, "Laurent Vilday" notifications@github.com wrote:

OMG again, a problem with chrome. I so definitely hate this browser.

The image is correctly parsed from safari and firefox but got errorr with chrome (all on Mac).

Do you think it is related to compression ?

-

Original screenshot : [image: img_3221] https://cloud.githubusercontent.com/assets/24301953/25307326/75e1d6fe-279f-11e7-9f4d-9900aeed8a21.PNG

Safari result : good

[image: safari] https://cloud.githubusercontent.com/assets/24301953/25307336/b023d7f4-279f-11e7-9b1a-1a3058b55703.png

• Firefox result : good

[image: firefox] https://cloud.githubusercontent.com/assets/24301953/25307375/4dfc659a-27a0-11e7-849a-ccd2c8513d4c.png

• Chrome result : some mistakes

[image: chrome] https://cloud.githubusercontent.com/assets/24301953/25307378/533b523c-27a0-11e7-8657-e8dfb1a08b73.png

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296395043, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs-0n1gL0lFCrMOxGwXB3vmM-rpk5ks5rylEfgaJpZM4M62Ii .

[dancmc]

Had a peek at the server logs on my phone, looks like the Mac chrome image really is just noisier and introducing random characters. I'll see if I can replicate it on my macbook at home and tweak the OCR algorithm.

If I can't then we'll just have to increase the maxsize of the compression a bit

On 23 Apr 2017 03:24, "Daniel Chan" dcmc87@gmail.com wrote:

That's really strange. I had no issues so far with chrome on windows, safari on Mac and chrome on android. Will test on chrome Mac when I get home and see what's going on.

It sounds like I may need to tweak something on the server side. Possible that chrome Mac is using a less optimised jpeg compression so the image isn't quite good enough, but will have to check

On 23 Apr 2017 03:12, "Laurent Vilday" notifications@github.com wrote:

OMG again, a problem with chrome. I so definitely hate this browser.

The image is correctly parsed from safari and firefox but got errorr with chrome (all on Mac).

Do you think it is related to compression ?

-

Original screenshot : [image: img_3221] https://cloud.githubusercontent.com/assets/24301953/25307326/75e1d6fe-279f-11e7-9f4d-9900aeed8a21.PNG

Safari result : good

[image: safari] https://cloud.githubusercontent.com/assets/24301953/25307336/b023d7f4-279f-11e7-9b1a-1a3058b55703.png

• Firefox result : good

[image: firefox] https://cloud.githubusercontent.com/assets/24301953/25307375/4dfc659a-27a0-11e7-849a-ccd2c8513d4c.png

• Chrome result : some mistakes

[image: chrome] https://cloud.githubusercontent.com/assets/24301953/25307378/533b523c-27a0-11e7-8657-e8dfb1a08b73.png

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296395043, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs-0n1gL0lFCrMOxGwXB3vmM-rpk5ks5rylEfgaJpZM4M62Ii .

[dancmc]

Yup just as I thought it was just different browser's jpeg compression affecting quality. Although strangely enough chrome on my macbook worked with your original image.

But I tweaked my OCR threshold setting and the details page should be fine now. Gear page might need a bit more testing to make sure the thresholds are correctly tuned, but no point trying too much till 3.0 hits

On 23 Apr 2017 03:35, "Daniel Chan" dcmc87@gmail.com wrote:

Had a peek at the server logs on my phone, looks like the Mac chrome image really is just noisier and introducing random characters. I'll see if I can replicate it on my macbook at home and tweak the OCR algorithm.

If I can't then we'll just have to increase the maxsize of the compression a bit

On 23 Apr 2017 03:24, "Daniel Chan" dcmc87@gmail.com wrote:

That's really strange. I had no issues so far with chrome on windows, safari on Mac and chrome on android. Will test on chrome Mac when I get home and see what's going on.

It sounds like I may need to tweak something on the server side. Possible that chrome Mac is using a less optimised jpeg compression so the image isn't quite good enough, but will have to check

On 23 Apr 2017 03:12, "Laurent Vilday" notifications@github.com wrote:

OMG again, a problem with chrome. I so definitely hate this browser.

The image is correctly parsed from safari and firefox but got errorr with chrome (all on Mac).

Do you think it is related to compression ?

-

Original screenshot : [image: img_3221] https://cloud.githubusercontent.com/assets/24301953/25307326/75e1d6fe-279f-11e7-9f4d-9900aeed8a21.PNG

Safari result : good

[image: safari] https://cloud.githubusercontent.com/assets/24301953/25307336/b023d7f4-279f-11e7-9b1a-1a3058b55703.png

• Firefox result : good

[image: firefox] https://cloud.githubusercontent.com/assets/24301953/25307375/4dfc659a-27a0-11e7-849a-ccd2c8513d4c.png

• Chrome result : some mistakes

[image: chrome] https://cloud.githubusercontent.com/assets/24301953/25307378/533b523c-27a0-11e7-8657-e8dfb1a08b73.png

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296395043, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs-0n1gL0lFCrMOxGwXB3vmM-rpk5ks5rylEfgaJpZM4M62Ii .

[VLstack]

Probably a version issue in the chrome jpeg algorithm, my chrome was not on auto update and the version was 57.0.2987.133 (64-bit)

Now I updated to 58.0.3029.81 (64-bit) and everything is fine with the exact same image.

[dancmc]

Oh I already updated the server code so it's probably hard to tell. But either way it's fixed =)

On 23 Apr 2017 05:35, "Laurent Vilday" notifications@github.com wrote:

Probably a version issue in the chrome jpeg algorithm, my chrome was not on auto update and the version was 57.0.2987.133 (64-bit)

Now I updated to 58.0.3029.81 (64-bit) and everything is fine with the exact same image.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296403107, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs0HoIV7446_L1pBie9Kvy6JhVv5jks5rynKlgaJpZM4M62Ii .

[VLstack]

Great :)

[VLstack]

So, what is our next move ?

1) public release 2) beta public release 3) wait for MFF 3.0 before option 1 or 2 4) alliance release 5) another option

What do you think ?

[dancmc]

I'm leaning towards option 3, if only because I'm hopeful it will be released this week and then I can rework the OCR. Otherwise if not we can just go with option 1

On 25 Apr 2017 01:59, "Laurent Vilday" notifications@github.com wrote:

So, what is our next move ?

1. public release
2. beta public release
3. wait for MFF 3.0 before option 1 or 2
4. alliance release
5. another option

What do you think ?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Mokhet/MFF-roster-tracker/issues/43#issuecomment-296771285, or mute the thread https://github.com/notifications/unsubscribe-auth/AKYMs98gdMa3P8cEPycWxfc4mC2kgv70ks5rzOLugaJpZM4M62Ii .

[VLstack]

Good, give me free time for a few code tweaking :)

[VLstack]

Just finished reading the patch notes, I saw nothing about the 2 screens we are using, I hope the "new" UI did not messed up with ´em. Only a few hours to wait, first time I can't wait for the update, not for the game update itself but for our own update :D