Scanlation Thread

You lazy anons actually working on those chapters, or just fapping to eroge?

Attached: the thread.png (512x512, 16K)

Other urls found in this thread:

pocket.shonenmagazine.com/episode/10834108156631700425
pastebin.com/e8abid0s
pastebin.com/iBr5zPXk
pastebin.com/ry1WvpC1
python.org/downloads/release/python-372/
twitter.com/SFWRedditImages

what if I'm translating eroge?

Any tips about redrawing for causals?

Actually yeah I just finished rance iv the other day instead of actually translating anything.

If you're TS, push all the redrawing to your dedicated slave/redrawer. If you're also the redrawer use a big white stroke with the font when redrawing over text dialouge. You will be hated by every one of your readers including me if you do this though.

shitty ungrateful EOP readers have no power

just fapping

Realise anything you do will be shit, then do what you want.

Do any other stores have sales as good as BW? Those 40-50% coin sales are pretty nice if you have the patience to wait for them.

Typeset before redrawing.

Would anyone want more online resources for scanlation? I used to have one of my websites in the general tutorial posts that went around for a while, but I've been working on making more JavaScript pages that can literally just do things for you. Just finished the foundation for practicing levels that has an auto-levels function, but I can also make one for auto-rotate and auto-crop, etc.

Old trick I used to do before I got gud, redraw on an upscaled image and then downscale it to the final size to help blend redraws. Otherwise, just typeset over however much you reasonably need and go from there. Make a new layer and quickly draw where you think lines connect to get a better idea.

this, it helps a lot to just paste the text over the hardest parts

Forgot image. This is all entirely in JavaScript and runs smoothly, just need to spruce it up and add some more functionality.

>That's not white
Didn't use the auto function since the page I was leveling is a mag scan and has crazy bleed. Consider doing two leveling steps to minimize the amount of dodging and burning you need to do.

Attached: It's not white.jpg (519x947, 148K)

Bless you.

I would want that.

I haven't seen any sites that have as many sales as BW.
You can even abuse that coin system by making new accounts.
Initial purchase has always 50% return + campaign, so if there's a campaign with another 50% and you purchase a manga worth 600 yen you get 600 coins afterward. Expiration is an issue, but patience is key as you said.

As many is a given, I was more asking about sales as deep. Even without abusing you could buy a crapton of Kadokawa stuff for as little as 100円 earlier this year. (40% off, 40% coins)
I don’t know if I’m just not watching other stores closely enough, but I’ve never seen a blanket 40-50% on everything sale anywhere else. (Granted, newly released stuff is excluded from this on BW, but that’s about it)

I guess it's not a surprise that they offer such sales for Kadokawa since they own BW after all.
I know that amazon and EBJ always followed on the free stuff but never really paid attention if sales are on several stores as well.

How to we force Japan to scan their raws again? 10 year old raws from Share are excellent quality compared to what we get now.

start a weeb jihad against online publishers

That would just kill all raws entirely

how so? If you could no longer buy (shitty) raws online people would probably start scanning again

There's like 5 active japanese raw uploaders left and they sure as hell aren't going to scan when they can pump a few books daily
Scanning is too much effort and there's just not enough people left

It's like the prohibition in the united states, making moonshine in your basement sure as hell isn't comfortable but what the heck are you gonna do if it's the only option?

What is going to happen when they die?

You'll have to buy raws and scan them yourself.

Digital raws are generally so shitty the world would be better off without them. Fuck kadokawa in particular.

no way...

>in the future, DRM security will be so tight that only coding geeks could truly own digital books while plebs like me could only rent them through their shitty app readers or pay for overpriced apps made by those geeks

Attached: 1537546393177.jpg (1280x720, 463K)

I wholeheartedly agree on Kadokawa. All of their digital magazines (or everything in general) is shit. Which is surprising because they actually get it right when it comes to LNs.
Although, there's one acceptable magazine from them out there, Maoh.

Attached: 0035.jpg (1426x2048, 1.2M)

their manga are all shit though, besides the gay aquarium one.

They're just going to introduce server-side watermarking so every served image is going to contain your unique account info.

they have such abysmal security that you can apparently download almost everything for free from their sites. By the time they get to DRM shit like there'll be ways to filter that out

And only coding geeks will have the access to those ways. I wonder what's the maximum lifespan of the current dedrm plugin from apprentice. I feel paranoid if I can't truly own stuff that I bought.

and coding geeks will provide it to everyone else, like they currently do

will they?
doubt.

That's for Amazon and Kobo though. Even non manga purchasing people use it and they have a broad amount of users.
You can be sure that there will always be a way for that.

That's also the very reason we never got BW and proper EBJ because they have a very limited market.

Yes they will, for the e-peen and because fuck corporations and fuck drm

How do people get their raws from smartphone apps? And what's the trick to ripping free webmanga? (like shonenjump plus).

For the former you decompile the source and look at it.
The latter, open dev tools in Firefox/Chrome and check out what the website is doing.
Once you know, you just write a tool.
Manga has been going hand in hand with coding for a while now if you want to do it on your own.

>yet another coding stuff
It's unbelievable that there are hundreds of scanlation groups and every single one of them know this stuff while I'm powerless.
how do you rip this for example?
pocket.shonenmagazine.com/episode/10834108156631700425
can't find any image in dev tools.

Attached: 1529749382384.jpg (1280x720, 393K)

>It's unbelievable that there are hundreds of scanlation groups and every single one of them know this stuff while I'm powerless.
i very much doubt that. Most scanlation groups either buy raws, have someone to scan it from, beg someone to rip it or use public raws

You aren't looking properly. Sites using that reader are extremely easy to rip.

Attached: asd.png (1485x966, 540K)

Shounen jump readers are almost trivial to crack. You can download the scrambled images directly and finding the part of the js that unscrambles them is easy too. (Unlike sites like sunday webry that have js files so fuckhuge that my chromium died half of the time when I set breakpoints)

I don't think every major store has such flaws, it's rather easy to prevent leaks if you have sane web developers.

have you ever had a midget woman scream at you in urdu while you're drunk, very tired and just try to open your front door so you can collapse into your bed? That's what looking at code is like for most people

How do you get those from inspect element? and I have to unscramble those images as well, huh? I wish there's a guide to do this kind of stuff.

You don't inspect the element, switch to the Network tab.

I must be a tard because I never managed to find that part. Not that it matters, ripping unscrambled images from canvas is trivial.

I don't know what that is, but I can get the scambled image through view source. I have to do it 1 by 1 though.

Canvas is always the worst solution and never really delivers the actual quality, though. And they can easily watermark it.

In dev tools you have several tabs at the top. Element, Console, Source, Network, etc. And finding the scrambling code is also extremely easy via Chrome.
You can beautify the output in it as well.

Attached: src.png (1131x551, 64K)

Just search for the drawImage function, most sites use that one. The descrambling code is just a couple of lines for shounen jump and the same ripping technique works for all sites with that reader. (tonarinoyj etc)
>ripping unscrambled images from canvas is trivial
Probably, but I hate and don’t know much about js so I prefer to do everything in python. And I care more about automated ripping of online manga that I read and could never keep up with without ripping anyway. (How do normalfags do this? Set up two dozen alarms in their calendar for the release dates of two dozen different manga?)

Then? what now?

Attached: this.jpg (1646x715, 190K)

If you go into the JS files in the sources tab, you can search for getContext('2d'), which is the JS call for images. From there, there should be a single file that calls to this.solvedImage, which you can set up a breakpoint by clicking the line number. Reload the page, then step through the functions using F9 until you get to a function key called "solve". This function will be called through a loop that places the squares in the correct spot, but it gives you the formula if you watch it enough.

Attached: The formula.jpg (1920x1011, 717K)

Just use Firefox, you can download the unscrambled images directly from it. Just press Ctrl+Shift+C > Inspector and check canvas. Then right mouse button > copy > image data url and paste it in your address bar.

How can it differ in quality unless their script fuckups something? Watermarking may be a real issue though.

Funny you mention it, because I'm tapping into that method to rip the canvas right after unscrambling. But yeah, now after checking that site's script the unscrambler is in plain view, unlike ebookjapan's or bookwalker's.

I get them horizontal and the next instructions are just confusing.

Anyone else get one of these emails? 吐きそうになるところだったよ

Attached: absolutely disgusting.png (1628x1745, 736K)

Other user told you already, now you basically have everything you need. All that's left is to automate it.

Just some examples of the math inside.

>cell_width = (960 / (4 * 8)) * 8 = 240
>cell_height = (1378 / (4 * 8)) * 8 = 344

Now you have two variables already.
e = 0
e will always increase by 1 until it reaches 15 [ e < (4 * 4) ].

The rest is just copy and paste.

I already encountered canvas rips that had worse quality than descrambles so I don't trust it. It's also a huge pain as the other user said.
Ripping via script is just a single small command in a terminal and you have the chapter.

Aren't those free chapters? Why do you care about watermarks?

I give up. It's hopeless.I can't even rip a random shitty webmanga. I'll just pray that this calibre dedrm stuff is going to work as long as possible.

>automate
>math
>formula
>copypaste
It might be surprising for coding geeks like a lot of you, but I don't understand a single thing from this post. A lot of otaku are computer science students, huh? I should've listened to my dad and went to coding school shit instead of accounting & finance. at least the former could be useful for the hobby that I'm actually invested in.

Attached: hmm.jpg (1239x676, 53K)

I honestly don't know how to explain it easier to someone who can't code at all, or rather never had anything to do with it.
Also, your picture, you should have a symbol { } somewhere on the bottom-left. This will beatify the output, it's easier to read.

Let's break it down a bit then.
We have our for loop that initializes the e variable at 0, this is going to be our defacto counter for the entire unscramble. this.DIVIDE_NUMBER happens to be 4, I know that because I looked at the value in the inspector. You could also guess this just by how many rows and columns the image is split into. So we have the loop going until e is less than 16, the total amount of cells in the split image. After that, we have a mess of variables that make more sense in context.

To get that, we need to go to the parent function that called solve, and look at where it's using the returned variables. We can see that it's in the drawImage call, which has 9 parameters total. Looking at the JS documentation, it's easy to see what does what.

At the end, n will turn into our clipping x boundary, t into our clipping y bound, our cell width, our cell height, a our top-left x coordinate, o our top-left y coordinate. The variable names get mixed around, but that's what they are. From there, it's just following the math knowing that this.cell_width is 240 and this.cell_height is 344.

Attached: context.jpg (1920x1007, 735K)

Watermarking as in inserting shit into image that can be used to identify the person who ripped it. Nobody is doing that right now as far as I know, but who what will happen in the future.

Unfortunately I have no idea how to rip manga any other way, so I will stick with my scripts. It's lossless and one click per tank is not really that painful.

If it's any good that'd be useful as hell. But I doubt it's any good, like fucking Trados.

Not sure why you’d even want to manually figure out cell_width and stuff. Pretty easy for this site admittedly, more annoying on others and simply copying the code just werks with a lot less hassle.

Because that ripper, once finished, will work on other websites with the same reader and they're not always the same width and height.

If you were standing right behind me and giving me a direct coaching from the very beginning until the end, maybe I would be able to do it..
Thanks for trying to explain, but this is far harder than passing N2 exam.
I can't imagine getting into Python JavaScript stuff.

Attached: 1541052131682.jpg (482x549, 57K)

Get yourself a friend who knows how to code. Plenty of code monkeys on the internet these days. Or maybe that user should just post the script here, it's not like they're going to notice and fix it

You obviously copy the code that calculates width and height with divide_num and multiple too.

>Beijing
the eternal chink at it again

Attached: unscramble.jpg (1939x1378, 1.1M)

I don't mean to derail but has anybody here translated from English to Japanese? How much did you charge per word?

*professionally translated, whoops

I actually have one ironic weeb programmer friend, but I'm too scared and embarrassed to reveal my true power level to him. Not even my parents know that I'm actually a super weeb who's also pretty good at jap.

Attached: 1534255265883.jpg (540x756, 85K)

It posted without the text I was going to write.

If we follow the initial math in the first post I showed my inspector in, we can see that the function iterates up to down, left to right on the scrambled image, and puts it left to right, top to bottom on the solved image. This shift is a common cryptography technique called a Caesar Square.

To reverse it, all we need to do is put the image cells into a matrix, rotate the matrix (Initial position i, j goes into final position j, i), and ignore the excess since 344 high blocks don't go into 1378 nicely.

Python is extremely easy to get into.
You can probably do this much within days.
Here's just a quick and dirty descrambler for the images in your example.
I especially used the same variable names and everything as the source so you can see just how similar it is.

pastebin.com/e8abid0s

This will basically descramble the input image.
>python3 script.py scrambled_image.jpg

Attached: 10834108156639880687-2c71434131ef86846700a2f895ea2a42_out.png (960x1378, 556K)

Maybe the typesetting bit if there was some magic tool to auto-place straight from a word doc it without making it look like shit, but we all know that's impossible as you need a human to judge what good placement really is.

Enabling and proudly advertising the fact you're enabling machine translating faggots is pretty kimoi though.

And how do you download the images?

Why would you be embarrassed about not being monolingual
I don't get you americans

Don't know about murrica or that user, but here you still get scornful looks sometimes if people know you're into anime, and depending on your job you may want to keep your power level reasonably hidden.

I didn't include that. You just have to write a bit more that will do that as well.
The re and requests module will be sufficient to do that. Just look it up.

Open the network tab, find a scrambled image in it, right click "Show in Sources Panel". Load more, profit.

Attached: In the sources folder.jpg (1920x1007, 1012K)

I went to a translation school for uni and I would always laugh when people were like "holy shit machines are gonna take our job one day."
The attorney's office I worked for bought a $1000 machine that scanned and translated any document "automatically" and I told them I'd take a $1000 cut to my salary if it gave them usable target texts. They returned it the next day and gave me a bonus when I asked for it a few weeks later.
My school was 99% weebs and I've worn my Madoka and Junji Ito shirts to every single job I've had and never had any single issue about it, in fact I've met more people for being a weeb.

>americans
>implying
You should've noticed from my English that I'm an (elite) ESL.
I just don't like the stigma and my parents are extremely... playful and like to tease me for literally everything and I'm just too embarrassed and prideful.

Your English is good enough to fool me, user, but I too am an ESL.
What would they tease you for, anyways? "Haha our son is near fluent in a language with which we don't share a writing system how hilarious!"

I actually made and error but it surprisingly works because the srambling is so easy.

This line:
>out.paste(im.crop((a, o, a + width, o + height)), (n, t))
has to be this:
>out.paste(im.crop((a, o, a + cell_width, o + cell_height)), (n, t))

I have to click those images and save them 1 by 1 manually?

I'm asking about finding out image urls automatically. I can't find where the source script is making requests.

Can anyone recommend anything that's easy to rip from? I've been using Rakuten Kobo so far, but I'm unable to get it working on Linux, and it's a bit much wasting 1/5th of my shitty laptop storage on a Windows 7 virtual machine.

Because it doesn’t, they’re in plain sight in the html.

>Linux

Attached: 1538643492156.jpg (320x317, 18K)

In this case the JS of the website doesn't make any requests because all image URLs are present in the source from the very beginning.

*JSON, not html, sorry.

Oh, funny, they’re actually in the html too, I didn’t know the different shounen jump readers actually differ this much.
The Json is there too anyway, even though it doesn’t seem to be in use on this particular site.

That's some quality programming, thanks guys. Writting a proper canvas ripper for a reader this helpless may be actually harder than with other sites.

Here, couldn't be bothered to add threading or anything to increase the speed since I don't feel like it. Also keep in mind that it will download the images into the current directory because, yet again, I'm lazy.
Tried it with pocket.shonenmagazine and shonenjumpplus.

pastebin.com/iBr5zPXk

>python3 script.py VIEWER_URL

Attached: 10834108156647664758-ebf2679bea0e7179a21a5a0b5c97b633.png (764x1200, 464K)

Wait. Don't use it for shonenjumpplus after all, it messes up. I'll see if I can see why real quick if not, that's that.

Really thankful the series I'm doing is from Maoh.

Any booklive canvas ripper? this unknown oneshot is stuck there in my library.

Attached: ゴレムと魔女.jpg (1283x916, 374K)

Found it.

pastebin.com/ry1WvpC1

Attached: 10834108156647664758-ebf2679bea0e7179a21a5a0b5c97b633.png (764x1200, 485K)

From what I can see that majestic thing isn't using canvas at all, but plain images cut in 3 parts.

Some are cut in 3 parts, some are scrambled. I guess it's impossible, then? And they use their own app as well for desktop. sucks.

Last time I looked at BookLive it had horrible quality so I never bothered again.

How do I even use these codes?

I will see what I can do. It should be using canvas to unscramble somewhere.

python.org/downloads/release/python-372/
Install it.
Copy and paste the script into a text file and name it whatever.py.
Open a command line tool and input:
>pip install Pillow
>pip install requests
Afterward you can simply use the script if I didn't miss anything.
>python whatever.py URL

umm

Attached: Python.jpg (352x274, 21K)

You don't start python itself, but a normal CMD.

Even if you find it (probably hardcoded in the .net code) the keys are encrypted now IIRC

What does it mean?

Okay I did it, but in the last step
>python: cant open file 'whatever.py': [Errno 2] No such file or directory

Proofreading Karakuri Circus volume 29 as we speak, OP.

26's chapters will start coming up somewhere around there soon.

Well, the script needs to be in the same folder you're currently in. From your previous screenshots that would be "C:\Users\MSI".
You can open the CMD in any folder you like by simply holding SHIFT and pressing the right mouse button.

black magic...

Attached: surprised pikachu dot jpg.jpg (1703x322, 245K)

Have fun. Also, if that was a new chapter you missed the uncesored version by 5 minutes.
Things on these readers are uncensored for 8 hours after release (midnight Japan).

I found it. This at least seems more feasible than the garbage that is their web viewer js code.

If you know how to inspect process memory then you can find the decrypted epub file in there and dump it

Attach a debugger or use a tool like process hacker, search for 504B03 byte sequence in the process memory, one of the results is going to be your decrypted file you just need to save that block of memory

The problem is that you don't know where the data ends but as long as you dump like 200MB starting from the place you found the byte sequence you gonna be fine

I just clicked random stuff for testing. that manga is shit anyway, not even nudity can save it. It still sells like hotcakes though since teenagers are horny creatures who want to remove those black bars no matter what.