Yo, it's me again with another Shiroi Tentoumushi release.
It's still me with DeepL and Google Transate..
But I really did some digging on jisho, retranslated paragraphs multiple times, and edited the script a lot to get best translation I could get.

Hope you guys enjoy the adventures of these two gyaru!

fair to say I did not see that ending coming. Good stuff.

Base +6, Cressent67 +8, Veroxious +6, Stallordblitz +6, Zemyih +6, jilin +6, xdc123456 +10, Novanity +7, Encuilion +9, Porokichin66 +9, Otiropas77 +7, unknownshadows +7, AncientPhoenix +7, and 2 more...

While it's nice to have an effort in translation, is there really no one you could have asked to at least check the JP source and compare it to the machine translation? Machine translation very often completely omits words and sentences.

Base +8, ron_hayato +7, Boiii6969 -6, Zakuzelo +14, xdc123456 +10, unknownshadows +7, somerandomdude33 +8, rjimenez666 -6

The editing needs some work too. Instead of "editing out the texts, you should learn to use the text-less versions more.

I usually create 2 new layers for it. In this case I would definitely do it for the transparent boxes using these steps:
Depending on what shape of the textbox I want to edit, I do it like this:
1a. If the box is rectangular or square without frame, just copy a portion of the text-less that covers the box. In this case, you can create my own black, transparent box that fits the size of the translated text better.
2a. Change the transparency of the box layer as close as you can get to the original.

1b. If it's a speech bubble or something with a frame that contains a different color (in this gallery, black box against white 'inner' frame, which is perfect), use the magic wand to select as much of the darkened box as possible. Go into the new layer and then fill it with black. Use the magic wand again to select the transparent part outside the box and then invert the selection, and refill with black.
2b. Copy the completely filled box and use this selection to copy a portion of the text-less and put it in a new layer (Gimp will create a new layer when copy-pasting automatically) below the box.
3b. Change the transparency of the black box as close as you can get to the original.

This might sound like a lot of work, but the more you do it, the quicker it goes. This is one way of making clean speech bubbles or boxes in the cases where you got text-less version of each image. It's almost a crime NOT to do that.

This way, you can make even those without transparency have transparency, like this:
https://e-hentai.org/s/aa87b8a271/1680855-2 (org)
https://e-hentai.org/s/e882716ceb/1751637-2 (translated)

As for my own translation, I'm also guilty of using google and machine translation, but I try to follow tools like Rikaichamp (now called 10ten) more than whatever Google and Bing actually translates it to.

These are the free tools I use for OCR, translating and editing images:
GIMP (For editing)
CAP2TEXT (for OCRing)
Any browser that supports the 10ten(Rikaichamp)/Rikai-chan plugins
Google/Bing for the dirty deeds, but together with 10ten to read what each phrase really means.

Sometimes you have to write the kanji in hira/kata to find out what actually been said. Artist occasionally forget to see the correct kanji.

Last edited on 16 August 2021, 13:03.

Base +24, frozenoctorok +5, sonnytag123 +14, Zakuzelo +14, unknownshadows +7, TheUnknownable1234 +6, thr0w4w4y4249 +6, eastxx +6, somerandomdude33 +8, SkullDraecos +6

@raziafire
Well.. not really.. While I do have friends that speak Japanese, I'd rather not trouble them with this.. also its super emberassing ^_ ^;

@q91
Wow, thanks so much for that super detailed reply! To think you took that much time to write out such a detailed instruction..! Honestly, that's super nice of you!
For tools, I use:

Capture2Text or ShareX for OCR,

SickZil to remove text from speech bubbles (that works really well in my opinion, except for the more calligrapy-esque bubbles when the girls orgasm.. those are the ones I just slapped the hand drawn bubbles on top.. definitely will look into you tutorial for those.. but for regular bubbles, it is sooo much faster)

ImageTrans (costs just 10$) for Typesetting. It actually does a pretty nice job of OCRing too, since I figured out I could implement tesseract (the same engine Capture2Text uses) into it. It can also directly query translation services straigth from OCRing, have a look at it, made by a Chinese graduate student or something, pretty cutting edge.

And definetly check out DeepL for MTL, in conjunction with Google Translate and its character drawing feature, you definitely can get the jist of things. Indiviual kanji / phrases I usually look up on jisho but i will 100% check out 10ten.

Base +6, xdc123456 +10, SonicBlu +6, thr0w4w4y4249 +6, SkullDraecos +6

I appreciate your work. good job

Base +6, thr0w4w4y4249 +6, cutegyaru -12, somerandomdude33 +9, LoliShin +6, rjimenez666 +6, SkullDraecos +6

Excellent work, great translation - Thank you for sharing it with the community

Waiting for new translations of other works from this great artist, which have not yet been translated

Base +10, LegoArielFanMaster -10, assisted_vape -6

Did anyone else laugh like 3 times during this?

Base +6, rjimenez666 +6, TNGBill +4, SkullDraecos +6

Weirdly wholesome, happy, and cute. I love it!

Base +7, rjimenez666 +6, SkullDraecos +6

>It's still me with DeepL and Google Transate..
tagging rtl (title, uploader comment)

Base +39