Scanlating the Sunfish Scans Way Part 4: /a/nonymous, ydwtt, and TDX

This is Part 4 of our scanlation tutorial series. You can find all parts here.

My, it’s certainly been a while! 😛

Last time, we talked about Vexed Scans and their awesome scanlation of Mitsudomoe.

Today, we’ll be taking a look at not one, not two, but three groups at once!

NOTE: As always, if you have difficulty reading this article, please zoom-in on the page!

Group #1: /a/nonymous

Ahh, could there be a better embodiment of the spirit of scanlation than /a/nonymous? A bunch of RAW providers, cleaners, translators, typesetters, and quality checkers who don’t know each other all working together to bring series that they love to fellow fans – and with no desire for internet fame, I might add!

It’s always a marvel to see scanlation threads actually make progress, let alone actually finish a chapter. The timing has to be juuuust right so people that can contribute are around and the thread doesn’t get archived too quickly due to other threads’ activities.

With the Mitsudomoe anime already being a decade old by this point (the manga being older by four years), I was pleasantly surprised to learn that there were still some fans of it hanging around on /a/. I was even more surprised when they actually banded together to work on a Kaieda chapter all on their own.

Not to discourage anyone from participating in future Mitsudomoe Live TL threads, of course. My only intention here is to share my thoughts on all current releases in order to learn from them and hopefully improve future ones regardless of group affiliation.

I have decided to tackle today’s groups in order of release count in ascending order, which is why /a/nonymous is up first despite being the latest group to work on Mitsudomoe so far.

Note: When I started writing this post, /a/nonymous only had a single release. However, they have since released two more chapters which have also been added here.

Once again, I will try not to repeat any praises and criticisms I’ve already mentioned for other groups in previous posts, so please don’t assume that any of the following groups are terrible in case I don’t have that many praises to say – in fact, I think they’re all awesome!

Now, with that out of the way, let’s take a look at:

What They Did Right:

Efficient Typesetting

1 clever letter

You can tell it’s Futaba’s because of the boobs.

While I’ve stated in the past that 100% typesetting should be every group’s goal, it’s not a hard rule by any means. Really, with compact translations, some clever typesetting, and a good eye for spotting vacant spaces, you too can make your life a lot easier if you’re willing to sacrifice some visual purity.

This probably goes double for anons because their thread is always on life support – as other threads on the board get more replies, their scanlation thread slowly gets pushed to the bottom of the board, and if it gets too low, it gets archived.

Now, it’s not necessarily the end of the world whenever that happens because they can always make a new thread or continue through ghost-posting (which is harder due to not being able to post images), but they’re probably going to lose some collaborators when that happens, which is never a good thing because their numbers are usually already very small.

Typesetting being the second-longest phase in any scanlation process (the first being translation itself, depending on how much of a perfectionist the translator is and the simplicity of the source material), I can see why the typesetter elected not to redraw this panel. Of course, they can also just come back with the initial typesetting on the next thread, but scheduled TL threads are very rare from what I see.

If it were me, I would’ve redrawn the paper texture, then clumsily write Futaba’s letter with the mouse. That last part can take a few tries because it still has to be legible despite the poor handwriting while still fitting neatly on the paper, which could take anything from 5-20 minutes because I have a personal obligation to get everything just right (not that I have to, I just really want to).

If you’re under strict time pressure (which is rare but can still happen), don’t hesitate to take the easy way out and treat the text like you would any other SFX that’s too much trouble for its worth. I’m generally not a fan of this approach, but it’s understandable when done in the right place.

Preserved the Western Canon

2 preserved canon A

2 preserved canon B

/a/nonymous’ version points to Ch. 214 while the RAWs point to Ch. 215.

This was particularly interesting because it shows that someone actually bothered to double-check the English fan translations instead of just relying blindly on the RAWs.

Let me explain: TDX’s later releases were actually using magazine scans instead of tankoubon scans. My only guess for them doing so is because the tankoubon scans weren’t available at the time or they weren’t aware of them, because those are the only reasons that I will accept due to how low-quality the magazine scans are. Their cleaner not being good at restoration certainly didn’t help things either (in fact, they actually made the pages look worse, but we’ll get to that part later).

For Mitsudomoe, the magazine scans are different from the tankoubon ones in the following ways:

1. The page art is clipped differently due to magazines having smaller pages. If you compare both versions, you will see that the magazines have a small portion of new art at the top of the page, but are missing a big portion of the art on the bottom. This makes the tankoubon version superior in terms of chapter art.

2. The magazine scans include chapter-specific colored pages (usually special title pages) while tankoubon scans only include the grayscale versions.

3. While both versions have special colored chapters, the magazine scans have more vibrant colors while the tankoubon scans have more muted ones. This makes the magazine scans feel more alive in my opinion.

4. Quality-wise, Volumes 1-12 of the tankoubons are superior to the magazine scans. The magazine scans have better scan quality (though a bit faded in some areas) than Volumes 14-19 of the tankoubons. Sadly, my RAW provider didn’t give me magazine scans for Volume 13 but I assume that it’s better because the tankoubon scans are sadly blurry.

5. Some chapter numbers are different due to the magazines and tankoubons releasing chapters at different times, usually the special colored chapters.

It’s Item #5 above that makes working with Mitsudomoe particularly easy to mess up because the English fan translations use both the magazine and tankoubon scans thanks to TDX’s releases (that’s not an insult, by the way. It’s just a fact).

I was pleasantly surprised to find that the anons working on said chapter actually got the “See this chapter” note right because it meant that someone actually cared enough to check if Chapter 215 (the chapter the RAWs were pointing to) was actually the same chapter in English.

It was surprising because it meant that said person actually knew there was a discrepancy between the RAWs and existing releases, and that the English fan translations were actually using mixed RAWs.

Of course, such situations are very rare because most groups generally stick with just one version of the RAWs (usually tankoubon scans), but /a/nonymous once again proves that if you get the right people on the right projects, the results are nothing short of astounding.

If you’re adding new releases to existing ones that don’t belong to you, make sure to double-check your work so they don’t conflict with those that came before you.

Had Great Eye Candy

3 eye candy

This one’s actually from their third release which absolutely blew me away. The typesetter really went all out for this one! They even typeset all the SFX, which was an absolute treat! (Anonymous typesetters that do so are so rare nowadays.)

Works like these are what really give me hope for the scanlation industry. You don’t even need to inspect each panel because just looking at the page as a whole already screams quality.

Such is the power of having great eye candy – all that variety pulls the reader into the page, which is the first step towards any immersive experience.

But I’m not done singing praises for this chapter just yet, because they also had:

The Secrets to Great Eye Candy

4 secret to eye candy

Enough charm to make any doujin typesetter blush…

Study the above image and try to guess what makes for great eye candy in typesetting. Go on, I’ll wait.

All done? Alright, here it is. The secrets to having great eye candy are:

1. Variety

2. Liveliness

3. Fits the Intended Delivery

Let’s study each one in detail.

The first key to having great eye candy is variety. Because our minds get bored very easily, just looking at the same things over and over is going to put the reader on autopilot which is the exact opposite of what we want.

We want the reader to be immersed and engaged in our works, which is what most people refer to as “being in the zone.” When a reader is on autopilot, they’re just mindlessly going through each panel and are absorbing very little (if any at all) of what is actually going on.

While having zero variety in typesetting will rarely put the reader on autopilot (because the author’s art will be providing the other half of visual variety), we should still try to add them whenever the situation calls for it as it will always enhance the experience.

Of course, choosing different fonts is only the first step towards having great variety. Thankfully, the other steps coincide with the next items in our list.

The second key to having great eye candy is liveliness. This can take many forms, from tilting and rotating your text, to giving them fancy transformations like arcs and squeezes, to even varying how each individual character looks in relation to the whole line. Of course, we’re barely scratching the surface here because there’s virtually an unlimited number of ways one can inject life into their typesetting.

While all lines are easier to read in their plain forms, these often feel very static and plain to look at, which is just another way of saying that they’re lifeless. Lifeless lines aren’t necessarily bad, however, because they will be the most common lines you’ll be typesetting throughout your scanlation career. “Why is this so if they don’t contribute towards having great eye candy?” you may ask. Well, it’s because while every line will look more interesting with lively typesetting, they will also be very distracting and exhausting to look at.

This is why there must be a careful balance between lifeless and lively typesetting, with lifeless lines roughly being 60-70%, depending on how much the emotional tone of the series you’re working on varies. By being seen less often, lively lines will pop out more for the reader and will thus be more effective.

The last key to having great eye candy is fitting the intended delivery of the line/SFX. You can’t just vary your fonts and use fancy effects without thinking about it because it will almost always result in a complete and aimless mess. You must carefully choose what font will fit the line best, usually based on what the RAWs did. Then, you check how the line looks without any additional effects. If it looks good without any, then that would usually be enough. I won’t discourage trying out some things, however, because that also usually leads to good results.

Lastly, the effects you use need to fit the emotion, feeling, and intended delivery of the line/SFX. For dialogue, this usually means knowing how the character is currently feeling (e.g. are they tired, excited, angry, in pain, etc). For SFX, it usually has to fit how loud/impactful the SFX is (even if they don’t necessarily make a sound), the material making the sound (is it hard, soft, liquid, hot, cold, etc.), and the intent of the SFX (is it meant to be scary, serious, dramatic, epic, casual, comical, etc).

Of course, you’ll sometimes need to consider more details than the ones I described here. What matters is that you completely understand how the line is meant to be perceived by the reader and deliver accordingly. Use the RAWs as a basis, but don’t restrict yourself to them if you feel you can do better.

Alright, that’s all the new good points I found. Now it’s time for:

What They Did Wrong:

Went Too Black

1 went too black A

1 went too black Ba

You know what they say, “Once you go too black, you never…” You know what? I think I’ll stop there.

This one is really easy to mess up if your monitor isn’t set up properly or you’re not paying enough attention to the original page’s quality.

Generally speaking, all RAWs that belong to the same scanning batch should have about the same quality, so any configuration that makes one page look good should usually work for the rest too.

However, special attention must be given to grayscale and colored pages because it’s very easy to make a page so dark that already-small details in the RAWs become very hard to spot in your final version.

A good rule of thumb is to look at the RAWs and find details on the darkest portion of the page (Hitoha’s hair, in this case). Because our goal is to darken the page, we need to make sure that our adjustments will not make said details harder to spot than they already are. The same principle applies when you’re making a page lighter (which is rare, but not impossible).

If it were me, I would’ve made the page darker while still keeping the lines in Hitoha’s hair easy to spot with a duplicate layer of the page set on Multiply with low opacity, probably 25-40%. It’s not going to be as dark as /a/nonymous’ version, but it should still be better than the RAWs.

Do note that adjustments that work for grayscale and colored pages probably won’t work for regular ones. I generally use different adjustments for grayscale, colored, and regular pages, and while it usually takes a bit of trial and error for each RAW batch, it’s always in your best interest to do so because it makes appreciating the author’s art a lot easier for your readers.

More Missed Dialogue

2 more missed dialogue

Kaieda: “Gyaaaahhh!!! I’m being purified!!!”(Credit: Oshima Azusa)

I’ve said this before and I’ll say it again: If you plan on skipping SFX, make sure you don’t accidentally skip dialogue too!

It’s very easy to mess this up if you’re just skimming the page, so pay special attention to SFX-like text in every panel.

(Nitpick) Letters Touching The Art

3 nitpick letters touching

“Bad touch! Bad touch!!

Generally speaking, you never want lines to coincide with each other, be it in text or the actual art.

In the art industry, we call these things tangents – lines that fuse or lead into one another that create immersion-breaking results because they just don’t look right.

While these things do occur in real life (especially in photographs) and authors usually leave them in by accident, our main concern is with our additions to the source material – namely in our typesetting and redrawing.

In the interest of brevity, below are some images to help you get an idea of what to avoid:

NOTE: Feel free to check the sources as they lead to some very helpful articles!



A line from one object runs directly into the line of another.


The containing lines of two objects run alongside each other.



The containing line of one object “bumps up” against the containing line of another object.


Two lines meet intentionally, but a third (accidental) line runs directly into the place where they meet.

Here are some more from real life:








“My next creation, the Lemon Centipede, will somehow be even more medically accurate! Mwahahaha!!!”

For more information on how to spot and avoid these pesky things in your work (scanlation or otherwise), Google “art tangents!”

(Nitpick) Rotated Translations

4 nitpick vertical A

4 nitpick vertical B

Let me begin by saying that I’ve got nothing against rotated text per se, especially when scanlators have no choice because the target area is just too small or is at too steep an angle. I think I’ve done this a couple of times myself now that I think about it.

I would, however, like to discourage their use as much as possible because they are very difficult to read (sometimes requiring readers to rotate the whole page or their devices if the line is long enough) and are generally bad to look at – especially for character dialogue.

As much as possible, aim for vertical, non-rotated text because even if it covers up more art, it’s going to be less immersion-breaking for your readers in the long run.

Not Enough Breathing Room For Text

5 not enough breathing room

Double minus points for tangents, natch.

You might remember me saying in a previous post that text inside speech bubbles generally require some breathing room, preferably of equal amounts on all four sides.

As you can see here, while there is enough breathing room on the top and bottom of the text, the left and right sides are literally pushing on the inside of the speech balloon, which just doesn’t look right.

The solution here would be to transform the text so it becomes thinner (more on that in a later post), thus giving it more breathing room on both sides.

Of course, this problem goes both ways because you can also have:

Too Much Breathing Room For Text

6 too much breathing room

As typesetters, we must strive to make as much use of the space we are given with. We may not want to take up too much space inside speech bubbles, but we also don’t want to leave too much space either.

Unless your intention is to make the line sound like it’s being spoken really softly (and I only advise to do so if the RAWs also have such an effect), you’re better off going with a well-balanced typesetting most of the time.

NOTE: I actually compared the left-most panel with the RAWs and I didn’t get a “soft-spoken line” vibe from it, so I personally would’ve gone for a lovey-dovey/charming effect here. If anything, Norio-sensei drew the speech balloons too big or kept the font size too small (it’s the same for every other line on the page). I assume that manga artists don’t like to fine-tune their font size for each speech balloon the way I do, which is understandable because it is too much work – but it does make everything look better in the long run!

Sloppy Editing

7 sloppy editing A

7 sloppy editing B

When I first saw this panel, I thought the cleaner accidentally left in some brush strokes. I only realized that they opted to directly use the white border from the RAWs when I compared them. Either way, it looks still looks awful.

Though I’ll give the group the benefit of the doubt here and assume that this was a missed-out error (or they decided they didn’t want to fix it due to time constraints), because the easier solution would’ve been to just clean the whole panel like normal and give the text some Outer Glow so it stands out against the background.

You never want to leave these things in on purpose because it tells your readers that you don’t have as much dedication to the craft as you should.

(Nitpick) Inconsistent Editing

8a nitpick consistency

While I applaud /a/nonymous for typesetting Kaieda’s sweater in the top-right panel, I have to point out that it’s a bit jarring for readers to see English text in one panel and then Japanese in the next. This is because typesetting something implicitly sets a precedent that you’re going to typeset it for every other appearance for consistency – and when that consistency isn’t met, it becomes distracting.

If they never intended to typeset Kaieda’s sweater in every panel, a better approach would’ve been to simply add a TL note or a small translation next to it similar to what they did with Futaba’s letter earlier.

This is why I opted to use a TL Note for Chiba’s cap and Tabuchi’s shorts instead of typesetting them:


My second reason for not typesetting them is because I’m lazy, but that wouldn’t sound as cool now, would it?

Props to them for going through with it though. Their hearts were definitely in the right place.

They might have initially intended to typeset her sweater in every panel before realizing how much work it would be, or simply thought that the smaller panels were not worth the effort (they did redraw Kaieda’s sweater two more times in a later page, probably because it was the focus of the panel).

Sometimes, the easier way out is the more elegant solution.

(Nitpick) Did Not Redraw

9 nitpick did not redraw

Redrawing is perhaps one of the more difficult topics because it requires digital art skills and (sometimes) digital art tools like pen tablets (mostly to get some lines just right).

Then again, if you’re going to drink poiso- I mean, go into scanlation, then you might as well drink the whole bottl– I mean, go all the way. Dammit!

We’ll get into redrawing for Mitsudomoe in a later post, but this is something that I fear not a lot of people can do very well (the ones just staring out, at least).

While Mitsudomoe’s art is thankfully crude enough in most places that you can easily redraw it with a mouse, for most people, I fear that a pen tablet will be necessary – or at least some basic Photoshop Pen skills. I’ll be focusing on the mouse/pen approach because it’s the cheapest and most accessible option.

In any case, it’s always a bit jarring to see bits and pieces of the art missing in the final version, so if you want to give your audience a seamless reading experience, redrawing is an absolute must.

Text and Art Merging

10 covered text

This one is actually from their second release. While the text isn’t actually being covered here, that’s certainly the impression that most readers will get because the characters are merging with the thought bubble lines on the right.

If it were me, I would’ve typeset this vertically and probably would’ve gone with a font that distinguishes between “O” and zero more clearly (the ones where zeros have a diagonal line). Here, the reader might get the wrong first impression that Matsuoka is just saying “Oooo” like a ghost until they read Hitoha’s line that comes afterwards.

If I had no choice but to have my text overlap with the art such as here (it’s happened before), I would’ve used some Outer Glow to separate my text from the art. Such a clear separation will make your text easier to read, which always contributes towards a seamless reading experience.

The main reason you rarely want this issue in your works is because right after the art, dialogue is the second most important part of any manga. Therefore, anything that will overpower the text should be avoided unless that’s what the original work was going for (such as overlapped text which we’ll get to in a bit).

Forgot to Switch to Grayscale

11 forgot to grayscale

You don’t like yellow sh*traws? – anon

As much as I applaud Typesetter anon’s brilliant efforts, it’s a darn shame that they forgot to set the image mode to grayscale before releasing the chapter as it could literally be done in just a few mouse clicks.

While I’ll admit that there’s a certain charm to reading an old book’s yellowed pages, most readers expect to read regular manga pages in grayscale (especially if previous releases were in grayscale too). Thankfully, it doesn’t look too jarring or distracting to look at, but it would still look a lot better in grayscale in my opinion.

Sadly, this issue could’ve been easily caught in the QC phase had they waited for the next Live TL thread, but there’s no sense crying about yellowed RAWs.

And hey, no harm done – it doesn’t really look that bad, and anyone with an image editor can make the pages grayscale themselves if they really wanted to.

(Nitpick) Had Blurry Text

12 blurry text

I’ve never really seen this font before, and regardless of whether the great character variety comes from mixing font faces and text transformations or the text being completely handwritten (in which case, Typesetter anon is waaay more dedicated to the cause than I am), the blurry nature of the text makes them a bit jarring to look at when compared to how sharp everything else is on the page.

If it’s a font, there must be a “Sharp” anti-aliasing method available for it, and if it was made with brush strokes, then a brush with sharper strokes would’ve been more preferable. Resizing handwritten text can also lead to them getting blurry due to how resizing regular pixels works. Thankfully, there are some Photoshop techniques out there that can sharpen these blurry texts, so they’re not completely unsalvageable.

While not a deal-breaker by any means, blurry sections on our pages should be minimized as much as possible – especially if we added them in ourselves, as they can break the reader’s immersion by making them think their vision is getting impaired whenever they see them. “Are my eyes really that tired..? No, wait… This line just wasn’t added right.”

Alright, that about does it for /a/nonymous. Let’s move on to…

Group #2: You Don’t Want These Translations (YDWTT)

ydwtt (You Don’t Want These Translations) only made three releases before they decided to hand over the reins to TDX. They originally planned to continue where Vexed Scans left off, but seeing as how TDX picked up Mitsudomoe at around the same time and have translated several chapters already (they seem to have just been waiting for a typesetter), ydwtt decided to drop said series and work on something else.

I actually only remembered that Mitsudomoe was their first project while reviewing their credits pages for this post. Turns out that they also started out as a one-man band and seem to have grown from there (the similarities in their origins with Sunfish Scans was quite interesting, especially since I would’ve also immediately quit once another group decided to pick up Mitsudomoe).

Such are the marks of a well-meaning scanlation group: They aim to work on series that they feel deserve more love, and if a group willing to work on said series shows up, they move on to the next unloved series. From my understanding, their mission statement was to work on series that have seemingly been abandoned, which is one of the most noble goals that any scanlation group could have in my opinion.

Rather than make a name for themselves with a series that’s never been seen before, they choose to aid fans that have been abandoned for one reason or another. Such dedication to the craft makes me respect them even more. It’s a darn shame that they seem to have become inactive since 2014, but their legacy will live on in their works.

Personally, I would’ve loved to see ydwtt and TDX work on Mitsudomoe together. While their translation styles are a bit different (as we will see in a future post), they certainly would’ve covered a lot more ground that way.

What They Did Right:

Horizontal Translations

1a horizontal tl

1b raw

Every now and then, authors sometimes use more space than usual to make room for multiple speech bubbles. As you can see in the second image, the second and third lines actually belonged to two vertical lines in the RAWs.

What I like about ydwtt’s version here is that they didn’t confine themselves to the formatting of the RAWs. You may remember that I advocate emulating the original text style in a previous post – well, that doesn’t necessarily apply to formatting because Japanese and English are fundamentally different orientation-wise.

While Japanese text works best in vertical spaces, English text works best in horizontal ones. Here, ydwtt wisely chose to typeset the second and third lines horizontally because it not only looks better that way, it actually gives them more room to be liberal with their translations.

That’s right: spacing doesn’t just affect typesetting – it affects translations too! More on this in a future post. (Man, so many previews!)

Typesetters must never forget to maximize usage of empty space, as forgetting to do so tends to limit your typesetting and translation by emulating the RAWs too closely.

Creative Typesetting

2 creative typesetting A

2 creative typesetting B

I know, I know… WordArt-esque text is super cheesy in manga, but it just fits here so darn well!

Comparing ydwtt’s version with the RAWs, you have to admit that their typesetting is more fitting for the coolest Sentai show in the universe (There, Hitoha. I said it. Now please get out from under my desk).

The only nitpick I have is that it’s a bit too dark to read. I’d personally make it larger and a bit lighter to improve its readability. All in all though, a most unexpected font style!

Every now and then, I encourage bursts of creativity and unconventional techniques with typesetting. Emphasis on “every now and then,” as using it too much will not only get old and annoying very fast, but will also clutter up your pages.

While it’s usually a good idea to mimic the RAWs, the occasional well-placed clever detail will certainly be a surprising treat for your readers.

Overlapped Translations

3 overlapped TL A

3 overlapped TL B

You gotta appreciate such fine attention to detail. I know it’s basic, but it’s a sophisticated use of a basic.

Technically, this falls under mimicking the RAWs, but because not all groups do this sort of stuff, I elected to give it its own section here.

The feeling the author was trying to invoke here was obviously an interrupted line (or the first line being immediately followed by the second).

To express that feeling in visual form, the second line overlaps the first. Naturally, any text in the first line must be overlapped too. Failing to do so will lose some of the spirit from the RAWs, and I only advise you to do so when you have one of the following reasons:

1. Overlapping the text makes it difficult to understand due to too many hidden characters.

2. It doesn’t look good due to too many hidden characters.

Though if I could change one thing on ydwtt’s version, I would actually make the “W” and “N” more visible and lightly trim them instead, just so readers won’t have to think for even a second, as unintentional thinking sessions usually lead to their immersions being broken.

Emphasized Words

4 emphasized words

In this case, please DO!

If there’s one major difference between Japanese and English, it has to be emphasized words. I don’t think I’ve ever seen any Japanese media where characters emphasized words the way English speakers do because it’s such a Western concept.

We’ll talk more about this in the translation tutorials, but watch/read any well-written English media with tons of dialogue (e.g. books, films, cartoons, comics) and you’ll find emphasized words up the wazoo.

This is because, as native speakers, their writers know that emphases help bring line deliveries to life because that’s how real people would say them. Bonus points for really emotional or personality-driven lines.

Alright, now let’s move on to…

What They Did Wrong:

Partial Translations

1 partial untrans A

The Japanese text says “Energy” for those who were wondering.

1 partial untrans B

Box 1: Brain. Box 2: Puppet.


Unlike leaving untranslated SFX, leaving out key text like this is definitely going to make your readers stop and try to figure out what the heck is going on. This is because, unlike SFX, key text is on the same level of importance as actual dialogue and should always be translated – the reason being that they always contribute to whatever is going on in the panel.

Encountering untranslated key text is like buying a cake only to discover that someone already took several bites out of it. You just can’t enjoy the whole thing anymore because some chunks are literally missing.

I actually have no idea why ydwtt didn’t translate these leftover texts as they were perfectly capable of translating even Norio-sensei’s (terribly) handwritten lines (which takes a novice like me hours to decipher). As such, I assume that these were simply missed-out.

Whatever the reason, said issue could have easily been avoided with rigorous quality-checking – so always take the time to review your works! It’s as important as every other step!

(Nitpick) Character-specific Font Styles

2 nitpick char-specific fonts

The road to heck is paved with good intentions.

While I highly doubt that a group nowadays would repeat this issue, I want to discuss it here just in case.

You do not, I repeat: NOT need to give each character their own distinct font style. My only guess for ydwtt doing so is because they didn’t want readers to get confused on who’s speaking in panel 5 above, but that’s going way overboard in my opinion.

Ideally, a character’s speech pattern and word choice should be enough to identify who’s speaking even if you can’t see the speaker themselves. While issues of not being able to identify who is talking certainly exist even in the RAWs, for the most part scanlators can’t really solve a problem that’s so ingrained in the source material. That, and if the author thought it wasn’t such a big deal, then you probably shouldn’t worry about it either.

I also have yet to see a RAW manga where the author decided to give each character their own font style because that would just be too much work to add in their already busy schedules.

If you feel that you must help the reader identify who’s speaking, try giving each character their own voice and distinct vocabulary (more on these in a later post). By that point, your work will be easier to digest than the RAWs because you actually went beyond where most original authors are willing to go.

The reason I’m calling this an issue, however, is because having specific fonts for each character is going to be more distracting in the long run. It’s eventually going to be noise that gets in the way of the reader absorbing what’s being said because they’ll be too busy subconsciously mapping font styles to each character. No one will be able to keep track of such information over the course of several chapters and volumes, especially if your releases don’t have a fixed schedule (*cough* Sunfish Scans! *cough*).

Be smart yet clear with your translations. You don’t need to hold your reader’s hand at every panel. Trust me, they’ll appreciate you respecting their intelligence.

Missed Translations

3 again missed TL

For the curious, it says “ED Theme” (Ending Theme).

Probably another missed-out detail or something they didn’t feel was worth translating. Or maybe they thought it would be too difficult to typeset properly due to the small space (more on how to get around this in a future post).

I know I’m gonna sound like a broken record by the end of this series but I’ll say it anyway: Always quality-check your work two to three times, preferably with a few hours/days in between so you have fresh eyes to spot mistakes.

Removed the SFX

4 removed SFX A

4 removed SFX B

Quite sacrilegious!

I’m going to give ydwtt the benefit of the doubt here and assume that they just forgot to typeset the SFX that they cleaned, because it just doesn’t make any sense otherwise. After all, if you’re not going to translate the SFX, the least you could do is leave them in.

Once again: quality-check, quality-check, quality-check!

Alright, I think that should do it for ydwtt, now let’s talk about…

Group #3: TDX

At an astounding 76 releases, TDX was the most prolific group to ever work on Mitsudomoe before we came along.

While I will stand my ground that they could have done a lot better with cleaning the magazine scans for their later releases, I don’t mean to dismiss their monumental contribution to the Mitsudomoe fanbase. Their releases were what made me decide that I needed to know what happens next in the manga, after all – so without them, I wouldn’t be here working on Mitsudomoe with the rest of the team for you guys. We really owe them a lot.

Sidenote: While researching their releases for this post, I now realize how lucky we were that TDX decided to work on Volume 12 with the magazine scans, because every RAW site seems to be missing two whole pages for Chapter 221. Maybe that huge jump in releases wasn’t such a bad thing after all…

Alright, let’s get this review started because we have a lot to discuss… I suggest making yourself a nice cup of hot milk and grabbing some snacks for the trip. Fasten your seatbelts, everybody! We’re going in!

What They Did Right:

Knew When NOT To Clean Between the Lines

1a when not to clean between

While I have mentioned before that you usually need to clean the art between your text lines, you should only do so when the art is directly making your text difficult to read by acting as distracting noise. As you can see, there is no such issue here because the background is distinct enough from the text even when it gets in between them.

Ironically, cleaning between the lines here would actually make the typesetting look sloppy and lazy. That’s because additional cleaning will make the text stand out too much against the background and would make the whole thing look unprofessional.

With text over simple screentone backgrounds like this, you never want to give your text too much breathing room by cleaning the art between them. For a seamless reading experience, there should be nothing out of the ordinary unless the author intentionally made it so.

I will, however, point out that I would have personally cleaned some portion of the background, specifically in the below points:

1b clean points

My reason for doing so is because said areas do make it a bit difficult to read the text. This is because, by being mostly surrounded by text, they have become practically separate from background and now act like distracting noise between the text. Nothing too distracting, but would still be nice to remove.

Sometimes, you have to leave the art just the way it is, and knowing when to do so is a matter of trying things out and getting a feel for it as you go.

Don’t be afraid to clean between the lines if you feel the art is making the text difficult to read. You can always undo your changes if you don’t like how things turns out.

Used Expressive Fonts

2 expressive fonts

Try not to get too distracted by the shiny Futabutt.

You may have noticed that I moved on from only using the Anime Ace font in our releases and have actually been varying our fonts over the course of Volume 9. Why the sudden change of heart, you may ask?

Well, it’s because I had the long-overdue realization that, while I’ve been mostly successful at tweaking Anime Ace for almost every situation, it just doesn’t get the entire feeling across for some scenes.

Some things just require a completely different font style to perfectly mimic the RAWs, and thankfully, decent typesetting fonts are freely available on the internet. You just have to find them and start your own collection, not unlike collecting different pens and screentones that you like for making manga!

I remember praising other groups in the past for doing this, but I’ve never actually advocated for it myself because I saw it as too much work for too little payoff. I now see the error of such narrow-minded thinking. Of course different fonts will help convey a variety of different feelings! I feel so silly for only realizing it now – but hey, better late than never, right?

Different fonts not only help provide variety and delicious eye candy for the reader, they can help you be more versatile with your typesetting so you can make each panel pop with life and attention to detail.

Kept Special Text Characters

3 kept special chars A

3 kept special chars B

Although I’ve never actually been mean enough to compare any group’s releases with the RAWs on my own spare time, I must give credit where credit is due here because this is a very easy thing to mess up.

Because most groups start working from the RAWs directly, typesetters can actually forget to add in these special characters (the star and heart, in this case) while editing the page. Even I, with my obsessive attention to detail, don’t actually compare our final page with the RAWs because I’m too busy critiquing our translations and the whole flow of the chapter. This is actually quite dangerous because you can end up straying quite a bit from the original source material.

Some solutions that I see for this are:

a) Add in the special characters in your translation scripts. The typesetter should take these as a cue to look at the RAWs and reuse the special characters directly.

b) The cleaner leaves special characters in (the typesetter will extract them themselves), or puts them in a reusable layer for the typesetter.

c) Compare the final page with the RAWs.

Of these steps, I like approach A the most because it’s the most convenient of the three. Approach B is a close second because cleaners and typesetters are usually on par with each other in terms of image editing skills and thus shouldn’t be a problem for either of them. Approach C, while being the most effort-intensive, isn’t that bad either if you can spare an extra step in your quality checks, because comparing your final pages with the RAWs will help you make sure that you’re not straying too far from the source material.

While not a serious issue when it happens, it still means that your readers will be missing out on some personality and dialogue quirks that characters have, so try to avoid this issue as much as possible.

Cleaned as Best as They Could

3.5 cleaned

I’ve mentioned before that scans of Norio-sensei’s work aren’t exactly the best in the world. Such page quality makes having a good eye for spotting all kinds of dirt indispensable, because you will encounter bad RAWs if you scanlate long enough.

In the above image, you can see that there is some dirt near Futaba’s lips. I applaud TDX’s cleaner very much here because I’ve seen a lot of releases in my day where the scanlators just left the unclean art as is (myself included).

Though if I could change one thing here (boy, I always have something to say, don’t I?), I would’ve cleaned the dirt near her tongue too (see if you can spot it).

While not particularly immersion-breaking in the above image, it can still be quite distracting for some readers, especially if they’re the kind who loves to take in every little detail on the page because they love the author’s art so much. I suggest paying special attention to characters’ faces, other body parts, and clothing because these are what most readers will be paying the most attention to.

Although cleaners are expected to clean all immediately-noticeable dirt on the page, they should also strive to remove “barely unnoticeable” details that clearly do not belong in the art itself, as doing so is not only a great service to your readers and the original author, but also will hone your cleaning skills in the long run.

Distinguished SFX in Speech Bubbles

4 clear sfx in bubbles A

4 clear sfx in bubbles B

Note: That’s Sugisaki’s trademark SFX, “Piro-riro-ring!” It’s actually a (very conspicuous) camera jingle. The translators probably thought it was too weird and decided to just localize it as “Flash.”

Unlike English, Japanese is a very onomatopoeia-heavy language. Because they have an SFX for just about anything (even things that don’t make sounds), Japanese readers are naturally able to distinguish between text that’s meant to be read as SFX and those that aren’t.

Western readers, however, usually take a few seconds to determine whether what they just read was an SFX or not. This is because they have been trained to treat everything inside speech balloons as dialogue unless explicitly told otherwise.

While most authors will actually distinguish speech balloons meant only for SFX with those that are used for dialogue (here, Norio-sensei has made the tail of her SFX bubble into a square instead of the usual curved spike), not all authors do this so you can’t always rely on them to help you.

Some (like Norio-sensei) will also write their SFX by hand, which is a very common tactic because dialogue text is usually typed instead of handwritten. Unless you plan to write all your SFX by hand as well to match such authors (in which case, I salute you!), you’re better off with a more standardized approach.

Surrounding in-balloon SFX with asterisks is one such method: it’s a popular way of saying that whatever is inside the speech bubble is not actually dialogue. It even works for outside text! *takes a sip of milk*

Do take care not to mix using asterisks for SFX with using them for emphasis, as this will confuse most readers.

While not always necessary (sometimes, objects themselves will spout SFX bubbles which makes it clear they’re SFX), if you feel that your SFX translations could be mistaken for dialogue at first glance, it’s usually a good idea to surround them with asterisks to help your readers identify them correctly.

It is crucial to help your readers identify the type of lines they’re reading the moment they see them. Otherwise, it’s going to (you guessed it) break their immersion. A character suddenly going “Beep! Beep! Beep!” is going to take anyone out of the experience (unless they’re a robot, but I digress).

Translated the Covers

5 translated cover

Finally! Sweet, sweet, flavor text!

After what feels like ages, we finally have a group that truly gets it.

It may look like insignificant text to the common reader, but to true fans of any series, such bonus text actually serves to enhance the experience. If they weren’t translated, it feels like we were left out of something that the Japanese readers got, which is like enjoying an amazing dessert that had a kick-in-the-pants aftertaste – pretty good, but could’ve been so much better.

I won’t ramble on too much about this, however, because this post is already wordy enough as it is.

Again, my sincerest gratitude to TDX for showing true dedication and setting such a fine example!

Translated Bonus Pages

6 translated bonus pages A

6 translated bonus pages B

“Relatively-pointless” as it may be, it’s still part of Mitsudomoe heritage!

6 translated bonus pages C

Wow, the postscript too!?

Similar to how I feel about manga covers, bonus pages are also an essential part of any manga. They usually have details that can’t be presented normally in the manga such as miscellaneous character information and other world-building info, lets us get a glimpse of the author’s thought processes as they were making the series, and postscript pages especially let us hear the author’s special messages to their fans.

If you’re only interested in translating the chapters, I suggest giving these pages a shot as well, because your readers are sure to thank you for it (I know I will)!

Translated the Table of Contents

7 above and beyond ToC

Daring to go where no Sunfish has gone before!

And now for another praise that I don’t personally advocate.

If you release chapters one at a time (and especially if your releases don’t have a fixed schedule), I would advise against translating the Table of Contents because that would require you to translate all chapter titles up front without any context whatsoever (unless you’re willing to read them all in advance, which isn’t always an option). You’ll also be committed to them so they can’t be easily changed later on (you’ll need to change the Table of Contents in the earlier release as well).

From one scanlator to another, translating this page is usually not worth it unless you batch your releases so you know your chapter titles are no longer going to change. Trust me, readers don’t read these things – and even if they did, they’ll probably forget the titles by the time they reach the chapters.

Translated Background Details

8 translated background details A

8 translated background details B

Ahhhh, TDX… Always celebrated for their excellence…

Beautiful. Just beautiful… It always does my old heart good to see fellow scanlators showing such love for the craft. And on such a zany series, to boot!

However, as much as I commend TDX for doing this, they seem to have made the page a bit too dark which kinda overshadows their awesome work here because it makes the text difficult to read (the top part especially).

This is why I will sometimes fine-tune individual pages when a single configuration doesn’t work for all pages in a batch.

Angled Text

9 angled text

I mentioned before that I am a huge fan of angled text because they inject so much subtle personality and emotion to each line.

Think of them as visual inflections. Just like vocal inflections, they add a lot of life to the delivery.

Followed the Angle of Speech Balloons

10a followed angle of speech balloons


If you’re a typesetter who’s only concerned with getting things over with, your speed will actually make this issue very easy to make.

I mentioned before that we should always try to mimic the RAWs as closely as possible. It’s especially important in typesetting because you can usually trust the author’s art and typesetting skills to deliver the perfect experience even in English. After all, the best possible typesetting is something that doesn’t stray too much from the original work while still being easy to enjoy in another language (not always possible, but always good to aim for).

While not every scanlation group does this, I actually appreciate TDX’s efforts here because it’s such a subtle yet important detail. If they didn’t tilt their text to follow the speech balloon, it wouldn’t have fit in as naturally nor would’ve looked as good, and trying to make it look good without the tilt would’ve been very awkward.

If you look closely at the RAWs, you’ll notice that the original text does this as well. This is why following the RAWs not only tends to give the best experience, but also makes typesetting a breeze because your text won’t have to “go against the current” of the art as much.

Used Adaptive Vertical Text

11 adaptive vertical A

11 adaptive vertical B

Typesetting is a very tricky business. On one hand, you need to fit your typesetting on the panel, and on the other, you need to make it look good and preserve the original feeling of the RAWs. Rarely, this will be impossible due to the length of your translations or space limitations, but more often than not, it should be doable with some skill and tact.

In the above image, it would be impossible to typeset the text vertically with zero tilting because the words just won’t fit in the available space. With some tilting, the typesetter is able to maximize the free space a bit more, but the text will still feel awkwardly jammed into the art because it would cover some parts of it.

The final solution was to tilt and transform the text with an arc to completely maximize the available space. This gives the added benefit of not covering any art (not always possible, but always nice to have) and being easy to absorb, as typesetting that has too many hyphens is always going to be difficult and awkward to read.

Sure, the text is a bit on the small side, but it’s the perfect compromise between readability and aesthetics. Any larger and it probably wouldn’t look as good (although I’d still try just to be sure, because larger text will always be easier to read).

Spiced Up SFX Translations

12 spiced up SFX tl A

Personally, I wouldn’t tweak the speech balloons to make room for my translations, but to each their own!

12 spiced up SFX tl B

SFX: Fondle fondle fondle~

Consider this another preview of some serious topics to come.

Generally speaking, you should never tamper with the author’s original vision. It’s their work, after all. We’re just here to help bring it to the rest of the world.

However! That doesn’t mean that we don’t have any artistic licenses of our own to use every now and then. Anything subjective can be considered an art form, after all – and scanlation is nothing if not subjective. But with great power comes an even greater responsibility…

If you actually take the time to compare our releases with the RAWs, you will notice that I tend to be very liberal with our translations. While I generally won’t try to change the meaning of lines and SFX too much, if I feel that they can be improved to be more entertaining or make a lot more sense in English, I will do so without hesitation. That doesn’t mean I change things all willy-nilly, however.

On the contrary, I spend a lot of time thinking over such matters. This is because I don’t treat such changes lightly. I carefully evaluate all sides of the situation: what I feel is wrong/lacking with the original text, how I feel they could be improved, and whether there could be any ramifications in future chapters. It’s never easy, but I can say with confidence that the end result is always better for it.

Note that I don’t do this because I know no one will ever compare our releases with the RAWs. I do it with the best of intentions – to make an English reader’s experience as entertaining as possible. Now, does that mean that I think I’m better than Norio-sensei herself? Far from it! I could never do or think of the things she does, because if I could, I would find her work rather boring and predictable!

I actually make these changes because of my huge respect for her and her work. I feel that her series should be given all the help they can get to reach as wide an audience as possible, and if I feel that certain parts are kinda lacking, I don’t think twice about improving it in any way I can. This is all in the name of making readers enjoy her work as much as possible.

Does this mean that my changes are a bastardization of the original? I haven’t heard any complaints yet, so I think that’s a testament to a job well done. After all, if any detail in your final release is received poorly, it’s your fault as the translator whether you changed anything or not. Saying “It’s what was in the original!” is akin to defending stiff and rough translations because they’re closer to the original Japanese.

Scanlation is all about entertainment, and I strongly feel that, as scanlators, we have a duty to bring the best possible experience to fellow fans of every series that we work on.

Do I feel that every group should make their own changes to the original? Only if they’re confident enough in their love and knowledge of the source material. Otherwise, it’s going to be a disaster. I discourage trying this if you’re not confident with your skills and entertainment sense.

And while we’re on the subject of spicing things up, I should also mention that this isn’t just a scanlation thing – it’s present in all adaptions of any media. The Mitsudomoe anime is a fine example of this, because if you actually compare the animated episodes with the chapters they’re based on, you’ll find that they not only spiced up some lines from the manga, they also added new lines to help explain things better or make things more entertaining.

And no, the whole “Scanlation isn’t an adaptation” argument doesn’t hold water because it’s already an adaptation from one language to another. Of course some things are going to be different! They have to be – otherwise, they’re not going to make sense to, or worse, alienate their new audience. This is the very spirit towards localization (itself a very polarizing topic) which works exceptionally well when done right and with a lot of heart.

More on this in a future post, however. Now it’s time to move on to…

What They Did Wrong:

Overlapping Characters

0 Overlapping font

While I’m an advocate of an all-caps approach to typesetting for most lines, I must admit that not all fonts were meant to be used in such a way.

This issue is a prime example of why most comic strip artists in the West opted to use all capital letters for most things in the first place – to conserve space.

You see, when your lines are in all-caps, their height is uniform and you only need to worry about their top parts from touching previous lines. With mixed casing however, you need to worry about a line’s top and bottom parts because they can now hit lines that come both before and after them.

While you can use Photoshop’s text properties to adjust the vertical spacing between lines, that’s going to be very tedious if you use fonts that require fine-tuning in too many places, which is why I advise picking an easier font to work with for most lines. However, the real issue with mixed-casing is that you’re going to have less room for your translations overall due to needing to make room for both the top and bottom parts of your text.

It gets even worse when you miss mistakes like the one above in your QC phase, because lines touching each other is always going to be jarring to see.

If you’re going to use mixed-casing, I suggest that you only do so when you don’t have any better font that matches the feel of the RAWs, and to pay extra attention to lines that use said font in your QC phases.

Compressed the Images

1 compressed

Left: TDX. Right: RAWs. (Click on image to enlarge.)

I’m all for image compression to make downloading our releases faster for our readers, but never at the expense of quality.

I have a few points against what TDX did here:

1. Their pages are actually smaller than the RAWs. Why they chose to do this is beyond me, because you always want to give your readers every possible opportunity to take in an author’s work. Resizing is fine if the scanned pages are in an absurd resolution (e.g. 3200 x 2400), but if the pages will fit just fine on a single monitor, I see little point in resizing them. Your readers will miss out on some great details, which is why I never shrink pages for our releases.

2. This one actually gets worse in their later releases, but here you can see that TDX’s version is a lot smoother and has lost a lot of the original texture. The most glaring example here is that there are less stars in panels 3 and 8 in TDX’s version on the third comparison. This is a form of destructive cleaning because it makes the art look worse by removing details that were present in the original.

As a rule of thumb, if your releases end up looking worse than the RAWs, you’re doing something wrong. Special care must be given to textured and shaded areas (the ones with screentones) to make sure they don’t end up looking blurry/smudgy when compared to the RAWs.

At most, I would probably make the art more crisp by making blurry lines sharper, and the whole page generally darker (but not too dark!). I’d also make sure to remove as much dirt as possible on the final page, whether the scanner added them in or the actual pages already had them to begin with doesn’t matter – if it ends up looking like dirt on the final page, it needs to be removed.

(Nitpick) Too Small Adaptation

2 nitpick too small adaptation A

2 nitpick too small adaptation B

This one is related to mimicking the RAWs. Of course, resizing TDX’s translation to exactly match the RAWs’ font size isn’t going to work because the words are too long. I would personally resize it up to 2/3rds of the RAWs’ font size while compressing the text a bit so it has enough breathing room on both sides.

While not the most jarring of issues, as a reader you may feel that TDX’s typesetting here does feel a bit too small for the scene. Of course, we want to minimize such feelings as much as possible in all our typesetting.

Knowing how to mimic the RAWs without making your work look terrible in comparison is a learned skill. I say “learned” because it’s something that you’ll have to figure out as you go. Much like any skill that’s too subjective to teach effectively, getting a sense for what “feels” right to look at is only going to come with experience. Don’t worry too much about it though, because just like how every artist discovers their own art style, you’ll also find your visual sense eventually.

Crowded Text with Lots of Open Spaces

3 too much breathing room, too much text

Man… Look at all that wasted space!

We’ve already talked about the problem with having too much breathing room in our speech bubbles, but it’s especially bad when you have a lot of bunched up text inside them as well.

You want to maximize all the space inside a speech bubble in the following ways:

1. As much as possible, you want to keep your text centered in your speech bubbles. The same goes for irregularly-shaped speech bubbles, because there will always be an ideal spot for your translations regardless of their shape.

2. You want to have a well-balanced breathing room for your text, preferably on all four sides. Not having enough is going to make things feel crowded, and having too much is going to make the line sound like it’s being spoken very softly.

3. When aiming for a well-balanced breathing room, take care not to make it sound like the character is shouting. This can happen when the speech bubble is too large for the line or the line is too short for the speech bubble. In such cases, it is perfectly fine to have too much breathing room because the alternative is going to give your readers the wrong idea about the line. As a tip, you generally want to keep the font size about the same with all other non-shouting lines on the same page.

Keeping these rules in mind should help you make full use of the space inside your speech bubbles. The better you get at maximizing the space available to you, the more flexible you can be with your typesetting and translation work as well.

(Nitpick) Squeezed Text in Nooks and Crannies

4 nitpick nooks and crannies

This isn’t really much of an issue, but I’d still like to address it regardless.

Here, I would actually typeset the line vertically with “I” at the top and “did!” on the bottom to better maximize the space inside the speech bubble.

This also has the added bonus of not making the text look like it was jammed in, making this section of the page look a bit less crowded.

(Nitpick) Not Enough Word Emphasis

5 nitpick not enough emphasis

This one actually looks like the result of poor font choice (capital letters and punctuation marks seem the most emphasized here).

Normally, I’d tell you that a good scanlator never blames the font for their mistakes because it’s all in the way we use them. However, the only workaround that I see here is to actually just change the font. I say this because this is a mixed-case font, meaning that the first letters of the sentence need to be capitalized and this font is going to emphasize them needlessly.

If I had to use mixed-casing (such fonts just work better for some lines), I would either make use of the font’s different faces (regular, italic, and bold) to emphasize each word or, if the font doesn’t have other faces, I would emphasize my words with a larger font size and all-caps LIKE SO.

Emphasis done right gives lines much-needed life in this visual medium, which is crucial because we want the reader to be as immersed in our works as possible – and giving them some brain candy (kinda like eye candy, but for the mind) in the form of well-delivered lines is one of the best ways to do so.

Missing Extra Pages

6 did not include pages after last chapter A

6 did not include pages after last chapter B

If we didn’t have the RAWs, these pages would’ve been lost forever. Doesn’t that scare you?

6 did not include pages after last chapter C

Okay, no one really cares about this last page, but it’s still part of the manga!

This one is a particularly big issue because even when you have no plans to translate extra pages, you can at least leave them in for completeness’ sake. At the very least, your readers will get to see more of the author’s art.

I will actually give TDX the benefit of the doubt here and assume that they meant to work on these and just somehow forgot. They were able to work on the covers and other extra pages just fine, after all.

Still, it’s pretty glaring because it should’ve been caught immediately in their QC phase (provided that they actually bothered to compare their page count with the RAWs).

Thankfully, this only happened in Volume 2 and I was actually obsessive enough to double-check all existing releases with the RAWs.

A good way to avoid this issue is to put all pages that belong in a release in the same folder, regardless of whether you plan to translate them or not. This way, you can be sure that you’ll never miss any pages in your releases.

Untranslated Speech Bubble SFX

8 untranslated bubble SFX

Don’t you just hate it when your belly starts speaking Japanese?

This is a prime example of untranslated SFX breaking the reader’s immersion because the speech bubble, while slightly different than regular ones due to its longer-than-average tail, can still easily be perceived as character dialogue to the untrained eye.

You can argue that readers who actually know their Furigana might realize that there are no everyday-speech Japanese words for such a combination, but if they already knew Japanese they probably wouldn’t be reading scanlations anyway.

As such, we can safely assume that our readers will be completely dependent on our translation efforts. This means we need to translate SFX especially when they’re part of the panel’s focus. Even if you have no plans to translate SFX, you’ll always need to translate these ones because your readers will always get drawn to them.

(Nitpick) Misleading Text Placement

9 nitpick misleading text placement A

9 nitpick misleading text placement B

For those who are new to this series: Left: Yoshioka. Right: Sugisaki.

We’ll discuss this issue a bit more in-depth in a future post, so consider this a preview.

You never want to ask your readers to figure out who’s saying which line because it will always break their immersion.

Looking at the RAWs, we can clearly see that the outer line belongs to Sugisaki. However, in TDX’s version, it looks like it belongs to Yoshioka at first glance due to how near it is to her. The only way to identify who the line belongs to is to read it and figure out that Yoshioka would never say such mean-spirited things (but such a detail will definitely be lost on some readers, which makes this “solution” invalid).

While I commend TDX’s efforts to point readers in the right direction by tilting the line towards Sugisaki, it’s still too close to Yoshioka and will probably confuse some readers.

The solution is to actually push the line upwards – because there’s no other person for it to belong to, the line will “anchor” itself to Sugisaki’s speech bubble (thereby hinting that this line belongs to her) and will distance itself far away from any other potential speaker.

You always want to minimize ambiguity in your releases, because even the smallest amount can mess up the reader’s enjoyment.

Didn’t Use Outer Glow

10 didn't use outer glow

Yeah… I’d be pretty upset too if my text was easy to miss.

For those unfamiliar, Outer Glow is what Photoshop calls one of its effects where you surround any visual element with a color of your choice. Think of it as a “halo” of sorts that you can customize (e.g. color, size, blurriness, etc).

Outer Glows are one of the most used effects in scanlation because of how useful they are. They’re usually used to help separate your text from the background so they’re easier to read and spot.

In TDX’s version, while the background is light enough that you can still read the text, the lack of an Outer Glow makes the line blend in with the background, making it easy to miss unless you’re paying close attention to everything in the panel (which most readers sadly do not).

It doesn’t matter how “insignificant” you feel a line is – if you already bothered to translate and typeset it, you might as well go all the way because it will be wasted effort if your reader doesn’t see them. I’d even go as far as saying that you should add Outer Glows even if the RAWs didn’t have them if you feel that they’ll make your text more readable.

Text Feels Clogged

13 text feels clogged

This one is a particularly bad example of poor space management, as can be seen in the lack of breathing room on both sides (which gives off a claustrophobic feeling when being read). While not much can be done with the text’s basic form (having each word on its own line is pretty much mandatory here due to space constraints), there are still some improvements that can be made to help readability and line delivery.

For one thing, I would re-evaluate the decision to use “Mama” instead of “Mom” because the latter would certainly make typesetting easier. However, I would also consider the age and personality of the speaker, including what emotion they’re feeling in the current panel (more on this in a future post). Given how the speaker is an emotional child (they look and act like they still go to kindergarten), I would probably stick with “Mama” here, but at least I thought about the decision more carefully.

The time spent reviewing such decisions is never wasted because you will be reaffirming them, which will make you more confident in your translation and typesetting because you’ll know that you’re going with the best possible option, which in turn will help you make even finer decisions later on.

Another possible improvement would be to make the text larger than the speech bubble itself to give it more breathing room (which also makes it feel easier to read), and to give it an outer glow so it doesn’t blend in with the art. Such an effect might make the line sound too loud though, so care must be used when resizing the text. We want to make it larger so it looks good, but not so large that it feels like the speaker is screaming their head off.

Poor use of free space can lead to awkward and uncomfortable reading sessions, so we must always try to make reading our works as effortless and as pleasant as possible.

(Nitpick) Didn’t Overlap Text

14a nitpick didn't overlap text

While this is not that big of an issue (the reader should still get the feeling of Futaba’s line being cut off because her speech bubble is still getting covered), a part of the author’s original intention is being lost in the adaptation.

Compared with ydwtt’s version of the same panel, TDX’s feels a bit flatter, while ydwtt’s has a bit more personality.

Here is ydwtt’s version again for comparison:

3 overlapped TL A

Personally, I would’ve stuck with TDX’s vertical typesetting of “Why?”, but made the font size bigger while clipping a small portion of the letters for a more dynamic feeling.

Used Overly-Eccentric Fonts

15 used eccentric fonts

This one doesn’t look like much of an issue at first glance, but just keep reading the line above a couple of times and you’ll hopefully see the issue.

While I now fully support having variety in our font choices, I would still advise against choosing fonts with awkward-looking or difficult to read characters.

Here, the offending character is the “D.” The extra lines make it look like a backwards “C,” and the general shape also makes it look like an incomplete “O,” or a crab claw of some kind. This will throw off readers who’ve never seen the font before, and those that have will still need some time to process that the strange-looking character is in fact a “D.”

It doesn’t matter how good the font is if it has at least one character that makes reading every line it appears in difficult.

Poor Font Choice for Rotated Translations

16 confusing font at an angle

“MMMI–?!” or “EEEH–?!” You decide!

Another consideration when choosing the right font for the job is how that font will look like in the final version. While the font they chose for the outer line here actually looks decent with zero rotation, it becomes confusing to look at when rotated in a (somewhat) 45-degree angle.

It doesn’t matter if the “?!” at the end will clue readers in that the line is rotated. By the time they see it, they will have already read the sideways E and H as an M and an Ɪ.

If they had to use the font, they would’ve been better off just tilting the text slightly instead of rotating it outright.

Underwhelming Typesetting

17 underwhelming sfx A

17 underwhelming sfx B

My biggest concern with TDX’s version here is actually with how flat it looks. Not even having three exclamation points could give it emotion (as a scream, it feels very flat and underwhelming). I’d say it has something to do with the text size (it feels more like a shout than a scream) and the lack of motion (and thus emotion) in the text.

I’m guessing they chose to go with that font size to better stay in line with the RAWs, but sometimes you need to go beyond the RAWs to get the same feeling across to an English audience.

If you don’t feel like hand-positioning each letter and varying their font sizes to simulate a blood-curdling scream (which is usually a good idea, but not always), the least you can do is make the text large enough to simulate loudness and maybe tilt it outwards to give off the feeling of being screamed towards the distance. I’d probably even transform the text into an arc and make each letter go from biggest to smallest to really sell the intensity.

The text direction of the RAWs where the SFX emanates from the gutter and goes up the sky doesn’t matter, as it’s not always possible to mimic the original text direction and still have the typesetting look good. What matters is that the feeling of the SFX is sent across to the reader.

While hand-crafting your SFX like this can sometimes take some time and effort (and some experimentation to see what looks good for a particular panel), the results are usually always worth it because, just like emphases, they help bring your translations and typesetting to life.

Here’s another example:

17 underwhelming sfx C

17 underwhelming sfx D

First off, the stencil-like font is a very weird choice for any SFX as it looks so unnatural just floating in the air like that. I would personally use such a horror font for something more normal like menacing dialogue, not something that will tend to repeat certain characters like elongated SFX.

Once again, varying the font sizes for each character from biggest to smallest probably would’ve gone a long way for this panel.

Props to them for trying harder this time with tilted and arcing text, however. While it’s still not as intense as the RAWs, at least we know they tried.

And now for the last one:

17 underwhelming sfx E

17 underwhelming sfx F

Here, much of the impact from the RAWs has been lost due to poor font and size choice.

In the original, the line “From head to toe!” (I’m adding an exclamation point here because it’s going to be more impactful that way) is delivered as a very scathing remark. Note how strongly and roughly the text is hand-written to emphasize the speaker’s intent. The larger characters even help simulate emphasis, at least visually (as far as I know, the Japanese don’t emphasize syllables the way an English speaker would emphasize whole words in a sentence).

Contrast with TDX’s version where the line is delivered matter-of-factly in a relatively classy font. The line still has somewhat of an edge given how snappy it is, but its small size undermines the effect and even makes it blend in with the background, making it a bit difficult to spot.

While I personally would’ve gone for a rougher font to better simulate the RAWs, if I had to go with TDX’s approach I would at least make the line bigger to give it more impact.

(Nitpick) Rotated VS Vertical Typesetting

18a nitpick rotated vs vertical

Now that I look at it, I just don’t really like the font I guess…

When you typeset non-dialogue lines like this (e.g. SFX and other special-purpose text), rotated perfectly straight (usually due to space constraints), you’re usually better off typesetting it vertically with zero rotation like so:

18b right

The reason for this is obvious: It’s always easier to read non-rotated text, and it looks better in general too!

Also, your text is going to consume about the same space anyway (one character per line), so you might as well make things easier for your readers. I suggest using Photoshop’s text properties to fine-tune how close you want each character to be to each other.

Try to minimize rotated text as much as possible because (let’s be honest) no one really wants to rotate the image or their devices just to read some line. Only do so when you have absolutely no other option.

Used Hard-to-Read Fonts

19 hard-to-read fonts

This one is another no-brainer: As awesome as it is to vary our font choices to better mimic the RAWs and get the feeling of each line across, we must be careful not to choose fonts that don’t read very well, especially in smaller font sizes.

On the left panel, the letters are too bunched up together which makes them hard to read. The capital i’s are also written in a wobbly style that makes them look like another H when grouped together. Thus, a more readable font would’ve been a better choice.

On the right panel, the letters seem to blend together in a somewhat handwritten, semi-cursive style. While this font would have no issues being used on a handwritten letter, preferably with a voice over or narration of its contents written in a more readable font, using it for spoken dialogue is going to act like a speed bump for the reader because of how difficult it is to read.

Remember: For a seamless reading experience, we never want to stop or slow the reader down. We want them to blitz through our releases (not literally of course, because speedreaders could never fully appreciate anything they’re reading, let alone the scanlator’s efforts) as uninterrupted as possible, sort of like riding a bike on a long, flat road. It’s that feeling of riding against the wind unhindered that we’re after.

Missed Some Things on Quality Check

20 Missed QC A

Boy, I’m getting a lot of mileage out of this panel, aren’t I?

Be honest now, how many of you noticed that little error while reading this article?

These brush marks tend to happen whenever Photoshop spazzes out and simulates a brush stroke wherever our mouse cursors happened to be at the time (trust me, I know). Or someone made a misclick and never noticed. One of the two.

Either way, such mistakes can only be caught through rigorous quality checking, so multiple QC phases are a must.

Here’s a more glaring one:

20 missed QC B

No, there’s nothing wrong with the art this time…

Here, the chapter title was typeset in an extremely small font size.

While I spotted a lot more QC issues, this one stood out to me the most because it’s the very first line on the very first page. I mean, how on Earth do you miss that?

Also, while I assume that TDX does have quality-checkers on their staff and that they do have QC phases, such errors make me question how diligent they really were with their duties. I don’t hate them, I just feel that they could’ve done a much better job.

(Nitpick) Didn’t Tweak the RAWs

21 nitpick RAW tweaks

I’ve mentioned before that while our goal is to try and mimic the RAWs as closely as possible, sometimes the RAWs just look extra lacking in some places.

Here, I personally would’ve replaced the handwritten line with a typeset one placed on the center of the speech bubble to better fit the rest of the page.

Whenever I see things like this that raise tiny red flags in my head (the “!!” inside the speech balloon just doesn’t look right here), I go into editor mode and make all the changes that I feel are necessary, because it never looks good to leave things that look bad as is.

Whether something that looks wrong is from the author or you, that’s on you because you had the power to change it and chose not to.

Didn’t Make Full Use of Free Space

22 nitpick not big enough A

22 nitpick not big enough B

Come on, it’s literally free real estate!

When it comes to down to typesetting, bigger is usually better. Not always, but on most occasions readers will want bigger characters because they’re easier to read and also maximize use of free space.

While the RAWs will usually help you figure out how to maximize the space available to you, you’ll sometimes have to think a bit to find how to best use them. Trust your eyes and your gut.

Needless Tilting

23a nitpick bad tilt

23b raw

Here we see a line that’s been needlessly tilted which led to a lot of wasted space.

Comparing TDX’s version with the RAWs, we see that the line was actually typeset normally.

I assume the typesetter chose to tilt the line because of how awkward the free space is (due to the shopping bag overlapping it a bit). Another possibility is that they were going for a “turning” effect (here, Mitsuba is turning her head to talk to someone behind her).

Either way it feels poorly executed because, by tilting the text, they were forced to use a smaller font, which led to too much breathing room, which in turn led to the line being awkward to look at.

Personally, I would’ve just gone for a basic typesetting here (no fancy tilting), but with the line italicized to give off excitement. You can usually tell which lines need such treatment due to the cloudy shape of the speech bubble (this varies by author and is Norio-sensei’s preferred style).

While I’d have to position the words around the shopping bag, the end result would still look better because the inside of the speech bubble will be a lot more balanced.

Do note that while I generally try to reserve tilting for extremely emotional or excited lines, I have used tilting for mildly excited lines as well – it all depends on the vibe I get from the speaker in the panel (don’t worry, you’ll also get a feel for it as you go along).

If you’re not yet confident in your ability to sense lines that would feel better with tilting, I suggest that you only tilt where the RAWs do, like so (at least until you get more confident):

23d good tilt

23c good tilt

Used Bad RAWs

24 bad raws

(Click on image to enlarge.)

At first, this may seem like “Compressed the Images: Part 2,” but I assure you that this issue is completely different.

We already know that TDX was using the magazine RAWs, but it’s the decision to use them that I want to discuss this time.

But first, I want to talk about each comparison in the image above:

1. In the first, we can immediately tell that TDX’s RAWs have a lot of dirt (it’s especially bad in panel 3).

2. In the second, we can see that their cleaning attempts have warped the screentones into something that resembles an impressionist painting.

3. In the third, we can actually see that some lines and textures have gone missing, having been smoothed out during the cleaning process.

While I will not argue that the world could always use more Mitsudomoe, it is my strong opinion that if you cannot salvage the RAWs in your cleaning phase, then it would be in you and your readers’ best interests to wait for better RAWs to become available. Perhaps you could even provide them yourself.

My reasons are as follows:

1. It’s not gonna look good for you if your releases end up looking like garbage. Whether it’s your cleaner’s or the RAWs’ fault (the reader won’t know or care), they’re going to tie said terrible release with your group. This is because your quality-checkers approved said pages despite the awful quality. This will make readers assume that you either don’t know what you’re doing or just don’t care (or both).

2. Regardless of how much a reader loves a specific series, reading such low-quality releases is going to taint the experience considerably. It’s like eating moldy chocolate cake – you can still taste the chocolate in there, but it’s just tainted somehow. Naturally, such releases will never be enjoyed in their entirety, and you won’t just be doing a great disservice to your readers, but to the original author as well. This is because you’ll be presenting their works in a horribly mangled state.

I really wish TDX had better RAWs to work with for their releases, or at least chose to buy or wait for better RAWs to become available. Their translation work has always been pretty spot on and enjoyable, but their later releases really suffered due to the poor quality of their RAWs.

This is why I advise everyone not to waste their efforts with poor source material. But if you really feel that those better RAWs are never gonna come, then by all means, feel free to take the bullet. We’re still in the Wild West age of scanlation, after all!

Made a Huge Gap

25 huge gap

This page still haunts me to this day…

Perhaps the biggest issue that all Mitsudomoe fans are familiar with is the inexplicably huge gap between Volumes 4 and 11. I wanted to double-check if TDX ever explained why they chose to do so, but their site seems to have been terminated, which is a darn shame because I forgot to archive their Mitsudomoe posts.

While Mitsudomoe has always been an episodic slice-of-life series and could be picked up from any chapter, there’s still some continuity throughout the series despite the lack of a traditional plot.

Why they chose to make such a large gap, perhaps we’ll never know – although I did send their group leader a message on MangaDex and will update this post if I ever get a reply.

With all that said, I highly advise against making jumps of any kind with your releases, regardless of whether they’re episodic or not. This is because there will always be the possibility of references to previous chapters (even with something relatively plotless like a gag series), so there really isn’t an argument for making gaps in your works.

Still, if TDX never made that gap to begin with, Sunfish Scans probably never would’ve existed, and I never would’ve met all the wonderful people and fellow fans that I did over the years. I want to thank TDX and every other group that’s ever worked on this series for that, so from all of us here at Sunfish Scans: Thank you!

Alright, that’s enough getting sappy for one post.

If you think I’ve been pretty ruthless with every other group’s releases so far, you haven’t seen anything yet because we’re gonna have a field day tearing our own releases apart and pointing out all of my failures – I definitely made a lot of the same mistakes I complained about in these posts and it will be glorious.

But before we do so, we need to make a quick detour for some Photoshop tutorials because I’ve been itching to share these with you guys for the last few years. Don’t raise your expectations too high, though. I don’t have any advanced tricks to share because I only use the basics – but it’s a sophisticated use of the basics that really matters in the long run.

See you guys on the next post! Same Mitsudomoe time, same Mitsudomoe channel!

6 thoughts on “Scanlating the Sunfish Scans Way Part 4: /a/nonymous, ydwtt, and TDX

  1. OshimaAzusa says:

    So, I finally had the time to compare your examples here with some of my practice work, and it was pretty brutal, I must admit.Typesetting is definitely a lot harder than it looks. I feel good about the font choices and the subtle tilting I did here and there, but my work can mostly be seen in the ‘did wrong’ sections here.

    I’ll have to come back to this post in the future when I try again, since it seems like you’ve listed my most common mistakes all right here.

    Liked by 1 person

  2. Unfilledflag says:

    So, I’m in a… similar boat as you: I heard about Mitsudomoe but for some reason never got to watch it, and at that time I didn’t read manga (I didn’t even know it was a thing, tbh).

    Fast forward a LONG time and one thing leads to another, I own a comic book store, and out of many thing I happen to start reading Mitsudomoe (funny enough, because someone recommended a VERY different manga: Onani Master Kurosawa) and loved every second of happines you and the other scanlators provided.

    Now, I did take some japanese lessons WAY WAY long ago and remember very little, and I felt curious about how some things were said in Japanese, so I searched for the raws and found this beautiful place. And then I realized… oh, no, I’ve been robbed: There’s so many missing jokes, yon-koma, and details. I want it _all_, dammit!

    But I trudge on, and keep reading what is available… and again I am betrayed! I searched far and wide just to have my heart broken: there’s a gap. Book 10 does not exist. And 12+ aren’t translated either…

    But I’m an idiot. And a dumb and subborn idiot. And now on my hard drive there reside translations (probably awful and full of mistakes, but I’m learning!) for chapters 172.5 and 173.

    So, please, はやく, type up those posts and help me learn how I can help.

    Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s