dtan.xyz

[Okay-Raws] Monster (Box 1) (DVD 576p)
Size: 9.79 GiB
Uploaded: 10/16/2026, 1:48:49 AM
Trackers: 5

File List (16 files)

📄[Okay-Raws] Monster (Box 1) (DVD 576p) monster01.mkv (688.2 MiB) monster02.mkv (654.7 MiB) monster03.mkv (727.8 MiB) monster04.mkv (651.2 MiB) monster05.mkv (631.2 MiB) monster06.mkv (636.6 MiB) monster07.mkv (527.6 MiB) monster08.mkv (644.9 MiB) monster09.mkv (678.0 MiB) monster10.mkv (558.5 MiB) monster11.mkv (623.9 MiB) monster12.mkv (650.2 MiB) monster13.mkv (481.5 MiB) monster14.mkv (625.3 MiB) monster15.mkv
625.10 MiB
📄monster01.mkv
688.20 MiB
📄monster02.mkv
654.70 MiB
📄monster03.mkv
727.80 MiB
📄monster04.mkv
651.20 MiB

Description

Monster was produced at a resolution of roughly 720x480. The ~480p image was encoded onto Japanese DVDs after being blurred horizontally. There are many SD shows like this, where the most viable source is missing horizontal detail due to such a blur.

The natural question to ask, then, is whether there are any unblurred sources. In some cases, a DVD (or even better, an SDBDs) from some region or another will be encoded without a blur. For example, in Welcome to the NHK, the US DVDs were unblurred, and I was able to use those DVDs to recover detail missing from the JP DVD. Unfortunately, in Monster's case, all of the (shared) non-Japanese DVDs are truly awful. Because the Blu-Rays for this show both seem to use a blurred SD source, my best guess is that an unblurred version will never appear.

Is there a way to recover the detail lost to the blur without an unblurred source? The answer is "technically yes." One can use an AI model that looks at the image and makes a guess as to what the pre-blurred version looked like. This is the method that the [sam] release uses. Unfortunately, very small changes to the image (like temporal noise) can result in extremely large changes to the "guess" that the AI model makes. There are various methods I can think of that one could use to deal with this temporal instability, but they are all labor-intensive and probably outside the realistic scope of a 74-episode project. I myself did various experiments with AI unblurring models and was dissatisfied with the results.

A method that can create the illusion of additional detail is to use a very sharp AI upscaler. Unfortunately, this method has the same temporal stability problems as the deblurring AI model. The upscaler has to make a guess about where to put its very sharp edges in its upscale, and its guess may change drastically from frame to frame, resulting in visible flickering/shimmering. The US BD and GER BD appear to have used sharp upscalers, and both appear to have the same loss of horizontal detail as the JP DVD (obfuscated somewhat by the sharp upscales).

The GER BD is very noisy, making for a somewhat unpleasant viewing experience. You can see some instability under the noise. I have not been able to look at the US BD to see if it is temporally stable, but I am skeptical that it could be, given that it is even sharper than GER BD. In addition, it appears to have upscale-related problems in niche situations, like with the scanlines here.

(By the way, I saw a random alleged "BD encode" on U2 that was quite temporally stable and appeared to be a sharp, stable upscale. It looked pretty good except for some IVTC issues. I wonder what was up with that?)

[null] also appears to have used some form of sharpening, since I have observed sharp edges in that encode that flicker in and out of existence. Or perhaps that was due to its aggressive denoising?

I am not interested in competing against these various sharp sources and encodes. This means that if you make an image comparison against GER BD, US BD, [sam], or [null], my encode will be subjectively inferior. I'm making this disclaimer because I don't want people to get the wrong idea about this encode—it is not meant to be something that beats every other encode in every aspect; rather, its goals are to be temporally stable, have no IVTC errors, and be better-looking than the Japanese DVD. That's it.

But those goals alone are decent ones to shoot for. In this batch, two episodes are pure 30fps, but other encodes of this show have ignored that fact and tried to encode them as 24fps. This means that every 5th frame was skipped, leading to jitter. This and many other little IVTC-related annoyances are addressed in this release.

Audio has been edited to remove CRT whine.

Comments

No comments found.

Other Links