For the release profile go here.
Polemic:
There
has been a lot of trash working on this title. Where fansubs are more
worried on popularity and download count than making quality
releases/restorations, I always keep humble on behalf of the project.
Part 1. My first course of action was to talk to the encoder of the last release available of this title on nyaa. This (link) was my first approach, you can see his snarky harshness at its best, and this the second (1st page),
or maybe I got the order wrong... Anyway I made my way through tests and
self-learning and made a release that was well received. The guy had me
stuck in his craw and despite his no English ability he was brave enough
to trash talk in my upload comment section. This is part one of the story.
Part 2. After the Perfect Blue
restoration got some coverage, I got this title offered to BakaBT by
someone else. Many people agreed and supported the offer (all the four
that commented on the offer and the at-the-time BakaBT release). Soon after,
the offer was put down in its most ridiculous form. Please, just let me
upload a screenshot of the situation.
Me thinking loud: "It's funny that my colors are worse than the old one when the old ones weren't even addressed...
Also
now the "richer" colors seems to be a "poor" approach to color
correction, since it removes so much detail (whatever he meant with
"detail").
Oh! the houses, the faces. Now I understand, the guy
is talking about the negative ringing you get when overshooting
(oversharpen). The source (and old release) have this so I guess that's
what he refers to "details". On the other hand it's funny he doesn't
spot REAL detail loss on sky, ground, and rock textures on the old
release. I wonder... /ironic."
The critic was so trolling and infuriating that I had at least to make a post on it. So I had my last thread from part one, and there
I put all things in place. Soon all the troll friends showed up like
rabbits, without arguments. It just goes to show what is BakaBT all
about, a very well designed web/tracker for anime which its purpose
doesn't hold up to its design. In reality its just a place for some
kiddies or maybe in their early twenties that don't even know what goes
on restoring an animation work as it clearly shows in this thread. Their mouth is full of terms they don't even have a grasp of. Oh, why such a blunt statement, you ask? Just read here ;)
Part 3. Damn, this is like
going to the cinema without leaving home, fuck it, this is way better!
Because of this, this time the "Kaze wo Mita Shonen" polemic section is
worth its own blog post.
If in part 1 and 2 ranting about
quality restoration work wasn't trolling enough, we got a newbie (but
you know... a staff colleague newbie) in the party. It wouldn't be
nothing to call home about if this noob wasn't the one that provoked my ban
on the BakaBT forums, "der richter" (aka "Judgment" -uh, I bet the guy
has complexes-). He decided that this was a great -trolling- moment to
offer his crappy version, the guy seems to don't pour any love to his
works, it's a matter of quantity over quality. I found his web where his main task seems to do typsetting (some snob subtitling task), his first touch with avisynth is as recent as October 2013, less than 2 years in which he seemed to stick with tools that I created like SMDegrain (basically an advanced but entry friendly denoiser tool), yet he is careless enough to spit things like "1 hour filtering + 1 hour testing encoding", "lazy mode encode", or "hopefully i will be spared with possible upcoming drama",
as if he was calling me out to cause some stir where it could be easy for his "lover" Chiya to ban me again (yeah, that pathetic). Later one guy gave him
some constructive critic but it was in vain, his offer was ready to be blindly pushed up to legible and replace the sono1998 release in just a matter of days (FYI, that's rare, I have seen offers wait well 6 months or more, common is weeks).
Let's
talk about the quality, SMDegrain is scalable, that is as great for
newcomers as for more advanced users, what this means is that you will
see many noobs like "der richter" misuse it (and hence give bad
impression about the filter). I myself used SMDegrain for my release
with better results, as you can see hims lacks stabilization and the
temporal denoising isn't strong enough so it gets that disgusting
swarming worms effect, processing things with Dehalo (for the ringing)
doesn't help the image look less like a moving oil-paint. It doesn't
seem he did anything else, other than sharpening (bundled inside
SMDegrain). It *can* look fine in screenshots due to his conservative
approach (other than strong dehalo), but the crappy truth comes in the
moving medium.
Kaze wo Mita Shonen [Judgment] Kaze wo Mita Shonen [Dogway]
Other issues besides lack of stabilization and temporal denoising, is lack of
color correction, either levels/colors/saturation, everything is
practically untouched. Also the strong Dehalo produced a very soft look
of the lineart which gives it an even thicker look, the bounce back
ringing is not handled either. Dehalo also caused some dimming of bright
spots and highlights reducing the detail richness of the original and
some smearing of low frequency detail. To make things worse, he used a PAL source (always an awful idea) as depicted by the 576 width size, and to give a false impression of quality not only he encoded at 10-bit to preserve the beautiful
swarming worms effect of the above image but ensured to grant it with a generous and at the same time wasteful 720p-worthy 3200kbps, so you don't miss ever a worm.
Overall it looks very poor and
lazy, and I feel bad for the sono1998 release which at least was half
good.
lmaoooooooo this is gold
ReplyDelete