Fansub Reviews In-Depth

This post was written by Dark_Sage. He is Dark_Sage.


This is going to be rather long and boring. But if you were ever confused as to why my reviews are set up the way they are, I’ll be explaining the logic behind it all. Don’t expect a comprehensive insight into my sagey mind, though.


Table of Contents


How many of you even know this exists? (The Table of Contents, that is. Not Whomp.)

Check it out… You don’t even need to use the scroll bar if you wanna go to a certain section. Think of all the valuable time you could save!

Of course, you may miss out on a joke or two by using this haphazardly, since I occasionally write those into these reviews. Prepare yourself for the possibility.

Release Information

Visual Quality

Script Quality


And if you’re OCD, try not to think about how I have sections labeled “Visual Quality” and “Script Quality” in the Table of Contents and “Visual Review” and “Script Review” as the sections’ header titles.

It’s probably unintentional.




Release Information


Episode details.

Release format:

If you have any special needs in terms of encodes, this is a pretty good way of finding out if they’ll be met. I put down whatever format groups release in here along with the file size. And if a file’s really big (like over 500 MB for a single episode), I’ll probably highlight that by bolding it or something.

The primary value of this section will be letting those of you who watch anime on devices that can’t handle 10-bit whether or not you’re gonna need to re-encode the release yourself or if you’re gonna need to find an alternate group’s version. I mean, a few seconds of searching on Tokyo-Toshokan will also let you know that, but TT ain’t got the snark I do.


For lack of a better term, this section will let you know how “liberal” or “literal” the subs are. Yes, I’ve been over how “liberal” and “literal” are bad terms to use, but they’re the easiest way to explain what I’m talking about here.

I’ll basically let you know whether a release uses honorifics or not, and what they decided on for localizing tricky terms. If you’re of the opinion that a group’s decision when it comes to honorifics is of the utmost importance, well lucky you, you can just check here instead of watching a release yourself and being unfortunately surprised.

English style:

British English or American English. Most groups use American English, but a few stick with British English as their standard. The funny thing is a lot of groups who use American English as their standard have editors who speak British English variants, and a lot of the groups that use British English are faced with the same scenario, but with their editors speaking American English instead.

What you end up with is more hybrid English styles. For example, very few American English groups will use American-style quotation marks, since the American method makes little sense. Fansubbers are strewn across the globe, so you’re actually getting quite a bit of cultural mixing in the final subtitle product you watch.

Encoding details:

Oh what, you want some stats as to who the fuck cares about this? Fine. 300 people click the encoding details link a month. What they get to see is a whole bunch of fun data that… I’m not sure what they do with.

This information gets pulled from the video properties (File->Properties->MediaInfo) in Media Player Classic and is a simple copy/paste of what you can find in there. If you really like numbers, this is a link for you. Maybe.


Typically, this only operates on in a binary sense: subs are “quick” if they are released within 48 hours of airing in Japan and “slow” if they come after that.

In practice, I’m not actually going to do much math here. Values are determined by what time the first RAW was put up on Tokyo-Tosho as the “start date” and what time groups released their episode on Tokyo-Tosho as the “end date”.

The 48-hour mark can be considered arbitrary, though I did put some thought into it. 24-hour measurements are easiest to keep track of, so it had to be a multiple of that. 3 days was too slow to be considered “fast” and 24 hours alone would be argued as speedsub territory rather than just “quick” releases. So 48 hours was a good compromise. Fansubbers still find ways to get mad about it, though.

Oh, and I should mention this only measures the group’s speed for the episode I reviewed. No, I’m not nice enough to find all their speeds and then average them out. I don’t have the time for that shit.



External links.

Nankou's Nekos

Group website:

IRC channel:

And whatever fansub image comparison site is still running at the time:

While I don’t believe people usually click these links, people generally do click on links in the comments section and in the review section. According to the Site Stats, Crymore readers click on over 13,000 links a month. Not too bad considering there’s nothing worth visiting on the internet aside from Crymore.

Getting away from those numbers, I guess you could say this section is more for decoration than anything. But don’t you think it just ties the whole review together?





Visual Review





It’s easiest to think of my ratings as a five-point scale:

1: Terrible

2: Bad

3: Okay (+/- are often used for more specificity, since this is what the majority of my grades are)

4: Good

5: Great


I’ve gotten a lot of flak for apparently trying to make karaoke more boring with my reviews, and I figure here’s as good a place to address that as any:


So what exactly am I looking for in karaoke? It’s not just matching colors to the credits and calling it a day. Rather, I want karaoke to enable a song’s translations to enhance the viewer’s experience. (That’s a fancy way of saying “I just want it to not suck.”)

Yes, one easy way to do this is matching the font and its colors to the credits. This ensures a consistent color scheme throughout the OP/ED and makes the karaoke appear a natural part of the release.

Another way to do it is to match the karaoke to a theme or concept you picked up on. Are there a lot of hearts in the OP? Would adding them as a karaoke effect subtract or detract from the quality of the visuals? Is there a show about airsoft? How about an airsoft gun effect for the karaoke there? If you’ve been paying attention recently, groups have done just that (Hadena’s Love Lab and Anime-Koi’s C3, respectively).

[Hadena] Love Lab - 08 [720p] [AF9C419C].mkv_snapshot_01.35_[2013.09.14_00.15.38]
Hadena’s Love Lab
[Anime-Koi] Stella Jogakuin Koutouka C3-bu - 06 [h264-720p][52EED87C].mkv_snapshot_22.40_[2013.09.12_11.35.08]
Anime-Koi’s C3
I also try to be conscious of groups’ decisions when it comes to their karaoke preferences. Groups that take themselves real seriously tend to avoid Japanese words in their karaoke. But if they can match, for example, the quirky mood of an ED with interesting effects (like Commie’s Mondaji release), then there’s no problem, since the viewer experience is impacted positively overall.

[Commie] Mondaiji-tachi ga Isekai kara Kuru Sou Desu yo - 04 [18B50CAF].mkv_snapshot_23.21_[2013.02.05_22.18.02]
Commie’s Mondaiji
It is more difficult for them to hit top-tier scores, however, just because if they and a few other groups run into positioning issues with karaoke, they’re not gonna be able to recover with the bonus points that other groups might. (For example, with C3, Anime-Koi, and Commie each ran into placement issues with the karaoke. They had similar color schemes, but Anime-Koi came out of it with a better karaoke since they weren’t providing only a cute pink color with their subs — their karaoke offered more to the viewer by way of karaoke effect.)


(No, these images don’t exactly get across my point, but uploading Youtube videos of fansub karaoke got me banned from Youtube last time, so it’s the best you’re gonna get.)

What I don’t want to see is black & white Arial text that shows up for its allotted time and then trudges off with no emotion. Nor do I want to see “random” colors and fonts and effects being thrown out. And if I can’t read your subs because the karaoke credits overlap them, I may even call your mother bad words. Let’s stop doing that, please.

Real talk: if you give me rainbow karaoke for a show like Ergo Proxy, I will moderately dislike you for as long as my attention span allows.






Ideally, you shouldn’t be able to tell the difference between studio-made English signs and fansub typesetting — so the goal subbers strive for is matching typesets perfectly.

I’m not so picky about this. In fact, compared to my very critical eye when it comes to other facets of fansubbing, I really only care about typesetting so long as everything of value is typeset, and typeset well enough that it doesn’t negatively impact viewer experience. Yes, I do care more about quantity than quality when it comes to typesets at this stage of the fansub game.

That being said, “good typesetting” is not something I don’t recognize. And if I’m impressed, that means groups are getting bonus points thrown their way. It’s just that it’s easier to impress me by typesetting everything in an episode than typesetting half of the signs, but doing them all nice and pretty. Same goes for a script that’s translated nicely, but only half of it is done. I hope that my preferences are reflective of the standard fansub-viewing populace. If not, then that’s something for y’all to figure in when you check my scores. I provide all the typesets I can find to allow you to come to your own conclusions as well.




This is where I put all the extra stuff in that doesn’t necessarily show up in every review.

You’ll see encode comparisons here, images where groups fucked up fonts/positioning, bad encode screenshots, etc. Anything that makes the visuals suck outside of karaoke/typesetting is gonna show up in this space.





Script Review



I used to really analyze the shit out of this, but the only other fansubber that cares about karaoke scripts as much as I do is fotc, so I’ve held back a bit. Generally, you’ll only see screenshots pop up in this section if someone screwed up with spelling or really fucked up a line with gibberish.

Karaoke mistakes do factor into the final script score, but to a lesser degree than mistakes in the main script.


Main Script.


This is what you came to read, right? Fuck. All right, let’s see if I can come up with something…

Every anime release comes with between 200 and 500 lines in the script. It would be untenable for me to pick out every line and comment on it. So to compromise, I try to pick out representative issues to highlight in this section. Those are generally mistakes, but I’ll occasionally point out areas where groups did really well. I then use these to build a case for a final score to you AND myself (after finishing an episode, I rarely have a final score in mind, unless the subs were A-tier or F-tier).

With the mistakes, if I don’t think the errors are inherently obvious, I will try to explain them. And if I don’t think the solutions are clear enough from the explanation, I will offer alternative ways to handle that specific line. The difficulty with doing that is editors edit across an entire product — their personality will affect the flow of the entire script. Therefore, any corrections will inherently just be patches over the problem, and should serve more as examples of ways to handle it rather than the end-all and be-all final solution.

Mostly relevant

Because the content of my reviews is primarily determined by how many issues I can find, I have an incentive to find as many issues as possible (though I do look forward to a day when I can’t find any and can move on to other ventures like reviewing the bindings of manga or something). I also have to not complain about irrelevant things or you as the reader will just tell me to fuck off and not read a single thing I write ever again. This also works the other way — by keeping my complaints above a “spelling mistakes are the only mistakes” level, we won’t end up with a community of idiots. (Well, ideally. Turns out some of them like to be heard.)

All in all, bias is bad for business with this format (let’s pretend I make money off the site). If I intentionally leave mistakes out, not only does the content of my reviews suffer, my reputation will significantly drop if anyone points it out. Similarly, inventing issues with a release will get me crucified. And with the open way I conduct the reviews, it’s very easy to do either. Yeah, I tied my own hands here, and I think it’s one of the reasons why this site has lasted so long — I don’t have any room to fail you.

So what am I looking for? Spelling mistakes, grammar fail, and phrasing issues are the top of my list. Those are the easiest ways to turn native English speakers off a release. But essentially anything that seems “off” to me will get pulled down as a screenshot. And only if I’m sure it’s a mistake will I include it in a review. I can’t go into more detail than that, since then I’d basically be going through the entire English language and pointing out how it works. 20,000 articles to get across a lifetime’s worth of experience is not exactly something I’m interested in writing.

Well, I’m kind of out of ideas on what else to put here and I think it’s starting to get a little circlejerky, so I’m gonna stop. If you have any legit questions about this section, the comments section is open.







This is actually only here by popular demand. I still don’t know why so many of you love it as much as you do.

Anyway, this piece tells you whether or not the release I reviewed is technically “watchable”. Meaning, if you have incredibly low standards, will you be pleased? The answer is “yes, as long as Dark_Sage doesn’t give the release an ‘F'”. There, that’s the distinguisher. You don’t need a separate section for that, why do you want this?

Well, to save my sanity, there’s more to it than just Watchable/Unwatchable. Here’s the breakdown:

F: Unwatchable.

D- to B: Watchable.

B+: Quite watchable.

A-, A, A+: Oh fuck, now I have to rack my brain for some flattering words.

And that’s all there is.

Visual grade:

What mostly factors into this is the karaoke and typesetting. If you really screw something up with encodes/fonts/basic positioning, that will also affect the score, though. Karaoke generally dominates the grade unless there’s a lot of typesetting. The average scores for visual grades trend higher than the average marks for scripts. So think of this section as a good way for groups to increase their scores.

Script grade:

This is generally determined by a combination of how bad the representative issues are, how many of them there are, how many “minor” complaints I had that I didn’t put into the review (if these significantly impacted the score or could help fansub groups improve, I’d put them in my reviews; this is more just my brain working to analyze the shit than me being sneaky sneak with scores), the difficulty/length of the script, and the general impression I got while watching the show/writing the review.

If I were using a formula, I’d give it to you, but this is about as good as it gets for when it comes to me analyzing how my brain works to come up with the final value you see.

In case you were wondering, I did some analysis last season and the average score, almost on a season-by-season basis, is a C+. This ain’t high school. Bs actually mean something here.

Overall grade:

This used to be determined by evening out the visual grade and the script grade, but that caused a lot of “But the script is more important than karaoke, you baka!” Which, y’know, is a legitimate point. So I dropped down the effect of the visual grade until I finally reached a healthy balance (read: everyone generally being pleased enough by the weights that complaints have been brought to a minimum). So what you get here is primarily the script score, possibly impacted by the visual grade. I don’t have any numbers to give, since it’s really on a show-by-show basis, but let me give you some examples of how it plays out:

Group A gets an A for visuals on a release with a fuckton of typesetting and the karaoke really, really impresses me. Their script score is a C+. The visual grade boosts them up to a “B”.

Group B gets an A for visuals on a release with one sign of typesetting and the karaoke really, really impresses me. They also got a C+ for the script. The visual grade boosts them up to a “B-“.

Group C gets a B for visuals on a release with few signs and karaoke that was good, but didn’t blow my dick off. They got a C+ for their script. The visual grade doesn’t do a damn thing and they’re stuck with a C+ and now they have to mortgage their house again just to feed their children. Fucking hell, a C+ is a curse from god. Why, Dark_Sage, why?

So you can see it’s all dependent on a number of factors. Again, I’m not beholden to a formula for this.



At this bottom section, I’ll give some closing words — generally a recommendation for which group people should go with for this show, or a bit of advice to a group. Sometimes I’ll even sell the space out for sweet advertising money. I mostly spend it on condoms and Dape Rate Duct Tape™. Don’t ask why; there’s probably no specific reason.


Nope, we’re not done yet.

Back to top


You ever see this thing before? Yeah, it’s like a replacement for scrolling all the way to the top manually. Like a fucking elevator so you don’t have to take the stairs. Enjoy it.

40 thoughts on “Fansub Reviews In-Depth”

  1. Why’d I read that like it’s a manga :(

    That. Epic. Commie. Mondaiji. ED.
    And you could have added UTW & Mazui’s bestest OP from Railgun S. I know you saw it, and liked it.

    • There was a lot I could have talked about, yes. But 3,000 words is quite a bit and I had to draw the line somewhere. And I know if I started looking back at all the kara I’ve seen, I could end up with a similar-length post just about kara.

        • I could write on and on about anything. I don’t like to, though, because I’m under the impression very long articles are difficult to get into.

          Once I start long articles I tend to finish them too, but when I’m faced with 8 pages in Microsoft Word, I’m just as likely to hit the back button and find a more digestible article.

          And no, I don’t know the owner. I just have a few comics saved from him and thought it’d be worth sharing.

  2. >Spelling mistakes, grammar fail
    Spelling/typos in particular, and even certain grammar errors these days, are utterly unforgivable. Fansubs aren’t being done on typewriters or Memexes animated by dark magic, they’re done using computers and computers have spellcheck and can spot at least a few grammar fuck-ups also. Something like this:
    demonstrates fundamental disregard for viewers. “We couldn’t even be bothered to click the Check Spelling button that’s how much we care.”

    Granted, I’m not sure if anyone but Hadena does that and whether they’d actually pass the Turing test is a matter of debate. It would be a bit exciting if it turns out modern “Hadena” is actually an early stage AI resulting from the cross of an emergent spam bot on the computer of Arashi (since deceased), his old scripts, and a 419 scam.

  3. >Yes, I do care more about quality than quantity when it comes to typesets at this stage of the fansub game.
    >It’s just that it’s easier to impress me by typesetting everything in an episode than typesetting half of the signs, but doing them all nice and pretty.

    This isn’t a dig at you; I’m just honestly confused by these two seemingly contradictory statements.

  4. ‘Kay, was a nice read; good to know!
    And since you probably don’t get enough of these:

    Thank You very much for taking the time and providing info and stuff for us!! (goes for all included making Crymore possible as well)
    Blabla. But really, though… Thx. :D

  5. Why don’t you use a formula for the overall score? Would’n it be fun to put it together like *some blogs* do?

    Here, lemme show you: “This article got a score of 17965/25000. That makes it 435th in the global ranking!”.

  6. >Values are determined by what time the first RAW was put up on Tokyo-Tosho as the “start date” and what time groups released their episode on Toko-Tosho as the “end date”.



Leave a Comment