[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip / qa] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/h/ - Hentai

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


[Advertise on 4chan]


File: Sexfriend.png (2.99 MB, 1920x1440)
2.99 MB
2.99 MB PNG
Thread to discuss upscales, mainly with ESRGAN as that is the best tool for the job.
All my upscales can be found here:
https://mega.nz/folder/w8YE0agY#VRlb8QavyCIhU-ujHxc2gQ
It shows everything that I've upscaled, whether it contains subs, how many episodes, links to download from, image previews, upscale status and changelog.
All preview images are just run through mozjpeg at quality 90 which is nearly identical to original. The original PNG preview images are on telegram.

Torrent version:
https://mega.nz/folder/6I131Y4T#M6pHMtChiVxC4_0kwIhDKA
Keep in mind that torrent is an afterthought, downloading is easier with Telegram.

ESRGAN can either be used through my setup which makes it insanely easy for the average user:
https://mega.nz/folder/o9hwXRDT#5MolBcQ6hvxz7adS8Opb0A (Recently updated to 5.2, making it near fully automatic)
Or it can be downloaded from the official github:
https://github.com/xinntao/Real-ESRGAN
Though using the github version means you'll have to mess with CMD and manually process each episode.
So if you don't want that, use my setup above.
Either can be used for still images too, not just videos. Like if you want to upscale art.
It can also be used for 3D stuff, even heavily compressed twitter garbage: https://imgsli.com/MTE0NDYz
Feel free to ask for more comparison samples.

Links+Info to other random helpful tools, including leaked version of latest FlowFrames:
https://mega.nz/file/h0w0EQaJ#yg63GPVFT5rL0yQfSHYEOmDMNuMYS46cnAUcaQgxais

Most people know me as "Upscale Anon", so if you have any requests or need any help, simply ask for Upscale Anon.
I may not always respond to requests, but I do consider them.
And please don't be a greedy fucker who requests like 5-20 series. It's going to make it less likely that I do your request.

Feel free to talk about other AI stuff or other helpful tools.
If you've got something really good, I might just include it in the tool list.

Previous thread: >>7617512
>>
File: Elfen Lied.jpg (572 KB, 2560x1440)
572 KB
572 KB JPG
>>7965266
I do anime aswell, but it's behind a paywall due to how much storage space and extra work it's going to take once it gets going.
Find out more here: https://patreon.com/upscaleanon
>>
New upscale batch being uploaded now. Enjoy
>>
>>7965266
Hey Upscale Anon, you think you can do Onmyouji Youen Emaki?
>>
Thanks anon for the upscales
>>
Thanks for the upscales anon, could you consider doing Houkago 2 Sayuri please?
>>
>>7965285
Requesting “Kotowarenai Haha” , “No waifu no life” , “ojisan de umeru ana”, and “Isekai kita no de suke skill de zenryoku… ep 3+4” on pixel/ torr.
Thanks.
>>
File: file.png (78 KB, 448x785)
78 KB
78 KB PNG
Very epic lol. Failed on the very last file.
A bunch of stuff didn't even upload in the right order. Man I love telegram
Spreadsheet and changelog has been updated, same link as before: https://mega.nz/folder/w8YE0agY#VRlb8QavyCIhU-ujHxc2gQ
Here's a pixeldrain of the batch that I just uploaded. Grab them before some faggot takes it down: https://pixeldrain com/l/b3GbXWj3
I'll maybe update torrents later
>>7968071
Tagging you so you hopefully see it before it goes down.
>>
>>7968146
Thanks
>>
Hi UpscaleAnon, i really admire your work. i have gone through your detailed guide of setup and done everything according to it.
But i wasn't able to run the ESGRAN bat file. I don't know how to use that because am not much of a developer. i tried to run the files in cmd and powershell but it throws me an error everytime. Can you please tell how can i use it? i just want to upscale some music videos and random stuff...! It would be a huge help if you could tell me how to use the program. Thank you
>>
>>7971473
You just put the video files in the same folder as the bat file and then run the bat file.
Make sure you're on Windows. The script does not work with Linux or Mac I believe.
>but it throws me an error everytime
It really helps if you show the error it throws.
>>
>>7972966
Different anon, but also encountering an error with the split script.

Seems like an error related to the audio not being extracted and present and an error related to the video filename being wrong.

The normal script works flawlessly, and I've had great results with it. Thanks!
>>
>>7965273
Oh damn.
I wonder how well Fairy Tail would benefit from this.
>>
>>7973844
Strange. I need to eventually do something about the split script, as it sometimes does not accurately split video, resulting in footage loss.
As for audio, it extracts at the very end, which makes some sense that it fails, if it can't even grab the video.
No idea why video wouldn't work though if it works with the normal script.
Is the file in mkv or mp4 format? I might have fucked up and have it only work with mkv format.
>>
>>7974668
Ah, yea, you're right.
I see an error message that's looking for an mkv file when my input is mp4.

Confirming that fixing line 1143 in the script fixes the issue.

Thanks for the help!
>>
Hey upscale anon can you upscale enbi please
>>
>>7975368
>enbi
>>
>>7965266
Interesting thread Upscale Anon
>>
How does one request access to the TG?
>>
>>7987495
There's an invite link literally at the very top of spreadsheet. The first thing you see, with gigantic text lol
>>
So is this upscale ai able to spit out uncensored versions?
>>
>>7988851
No, that's a different AI program. Upscale increases resolution and enhances clarity a bit.
What you're looking for is decensor AI. But keep in mind that it's quite garbage in its current state.
>>
>>7988851
You could train an ESRGAN model to do this, but it wouldn't produce good results compared to generative AI. You could make a paired dataset of censored low resolution images with an uncensored high resolution equivalent and train a model specifically for this. The best way to do this would be to add mosaic to uncensored art for consistency. You could theoretically do this with a 2x - 4x model as well, but a 1x specifically for decensoring followed by a regular upscale would be better since focused models always produce better results.
>>
>>7965266
Can you upscale Dark Shell please
>>
You can upscale whatever you want if you have a gaming PC, which is likely considering which site you're on. Download chaiNNer and it'll get all the dependencies for you, then go to OpenModelDB and get get whatever model you want. chaiNNer has an incredible amount of flexibility, and it supports every video format. It also has an average color fix, which allows your upscale to retain the color of the original image 1:1. I have no idea why OP is promoting such archaic tools and practices. No wonder the thread's so fucking dead.
https://github.com/chaiNNer-org/chaiNNer
https://openmodeldb.info/
>>
>>7998896
I tried this and it didn't look as good and I don't really know what all the spaghetti means...
>>
>>7998896
>archaic
It's a fuck ton faster than chaiNNer and with my setup it requires basically 0 fiddling. chaiNNer is a pain in the ass for the average retard.
Sure it has a lot of flexibility, but you don't always need absolutely everything. You don't need a jet plane to go to the grocery store 5 meters away.
>No wonder the thread's so fucking dead
Do you even realize what board you're on? You're on /h/, one of the slowest moving boards.
Adding complexity to everything is a sure way to kill it.
Go ahead and prove me wrong, make your own thread with your own upscales using slow ass chaiNNer with a different model.
Feel free to make your own thread regardless, since you seem so butthurt about how I'm doing everything wrong.
Or you could just you know, grow up and leave. Nobody is forcing you to be here.
>You can upscale whatever you want if you have a gaming PC, which is likely considering which site you're on
You know that a huge portion of people on this site are phone posters, right?
People are also incredibly lazy, nobody wants to do anything themselves.
>supports every video format
Okay? That's not really a big thing, ESRGAN can use every video format too, since you convert the video to frames anyways.
>>
>>7999579
Depends entirely on the model used.
>>7999735
This sounds like incompetence on your part. Upscaling is a science. You can't form subjective opinions on that which is objective. chaiNNer is as fast as you make it, and it's infinitely flexible. This is coming from someone who now sits on 800 GB of personally trained model data and datasets containing hundreds of thousands of images total.
>>
>>8003287
>chaiNNer is as fast as you make it
It really isn't. ESRGAN is loads faster. That is not subjective, it's facts. Someone else than you tried to prove me wrong but he had only like 1/10th the speed that I get with ESRGAN lol.
You probably haven't even tried it. Either that or you just have a dogshit GPU and use a CPU on both, in which case chaiNNer might be faster.
Or maybe you were using the slow ass realesrgan-x4plus model, instead of realesr-animevideov3-x4. Or even an outdated version. Wouldn't surprise me.
>This is coming from someone who now sits on 800 GB of personally trained model data and datasets containing hundreds of thousands of images total.
Ok mr big guy lol.
Nobody's gonna believe you more just because you brag about random shit you may or may not have done.
Plus if you did shit right, you wouldn't need a whole ass 800GB of models lol.
Hundreds of thousands of images in dataset? Okay and? For all I know, it could be a complete garbage dataset.
>infinitely flexible
Like I said, you don't always need something that can do everything.
Go shitpost somewhere else, or like I said, make your own thread where you do things "properly".
>>
File: mpv-shot0001-compare.jpg (377 KB, 3148x1242)
377 KB
377 KB JPG
>>8003287
anon,it is best to give it up on wanting Upscale Anon to use other tools.
I had suggested Upscale Anon to use AnimeJaNaiConverterGui with NVIDIA TensorRT to upscale video at the fastest possible speed but he only wanted to use ESRGAN so he can have sharp lines on his upscaled anime.
>>
>>8003897
Yeah and you're the guy I was talking about that got like 1/10th of my speed lol.
Like I told the other guy, go shitpost somewhere else. Statistics doesn't lie, your tools are significantly slower.
Sharp lines are good. You want blurry shit? Go back to the waifu2x garbage
>>
>>8003928
will do. enjoy being the only one who would post here ever
>>
https://mega.nz/file/F6hiDZya#dl0Baqe4nUITlhPflXW_XW5fd6z9MUuQekGF0byBO_4
Can someone upscale this to 1440p my gpu fan is not working properly so its going high temp and crashing halfway through the process
>>
>>8005436
where is this from?
>>
>>8005436
ai is great but you really can't fix shitty source
>>
>>7965266
Does anyone have the magnet torrent that was floating around that had all the files?
>>
>>8012033
Did you even bother reading the OP?
Come on man...
>>
>>8012106
Have you even seen it? The mega has each individual series uploaded as a separate torrent retard. There is a magnet floating around that has ALL of them collected in one torrent you fuck wit.

I wouldnt be asking this question if I had found what I wanted
>>
>>8012106

I actually just found it you stupid fuck

magnet:?xt=urn:btih:A544FEB5DEA9641F3D49512A6708453666E7DAB5&dn=%2Fh%2F+Upscales+Batch+-+2022%2F12%2F01&tr=udp%3A%2F%2Ftracker.leechers-paradise.org%3A6969%2Fannounce&tr=udp%3A%2F%2Ftracker.coppersurfer.tk%3A6969%2Fannounce&tr=udp%3A%2F%2Ftracker.opentrackr.org%3A1337%2Fannounce&tr=http%3A%2F%2Ftracker.openbittorrent.com%3A80%2Fannounce&tr=udp%3A%2F%2Fopentracker.i2p.rocks%3A6969%2Fannounce&tr=udp%3A%2F%2Ftracker.internetwarriors.net%3A1337%2Fannounce&tr=udp%3A%2F%2Ftracker.leechers-paradise.org%3A6969%2Fannounce&tr=udp%3A%2F%2Fcoppersurfer.tk%3A6969%2Fannounce&tr=udp%3A%2F%2Ftracker.zer0day.to%3A1337%2Fannounce
>>
>>8012309
>>8012318
Holy shit, that is the old torrent you actual fucking spastic retard lol
>all the files
Not even remotely close, it's missing at least half of the upscales.
Sure, they're packed together into one torrent, but it really isn't that hard to download the torrent files from the mega and just selecting all.
Funny when retards call others stupid.
>>
>>7965266
will it be possible to use AI to de-censor hentai?
>>
>>8012348
This has been asked several times.
But no, not really. It looks like absolute garbage.
You'll have weird flesh blobs or melting candle wax.
I doubt anyone is really working on something like that to improve it.
Unless you mean using AI like Stable Diffusion, then yeah, it's technically possible and the result can be good.
However, it's not going to be coherent, as it doesn't account for previous frames.
I dunno if an extension to auto ask genitals exists, but if it doesn't, you'd have to process each frame individually.
Good luck with that on sex scenes that are thousands of frames.
TL;DR wait a few years maybe.
>>
>>7965266
Can you please upscale Rennyu Order: Okawari and Konomi ja Nai kedo. Thanks much
>>
>>>/g/
>>
>>8017811
???
/g/ is not for hentai.
>>
File: futa.png (126 KB, 924x1194)
126 KB
126 KB PNG
Sharing my Futa AI Upscaled HEVC collection.

I converted all of them to HEVC and included a missing futa hentai (Discode)

The original size was like 145GB and converting them to HEVC keep the quality the same and the total size to 65GB.

Will be seeding for a few days.
magnet:?xt=urn:btih:d9e21839396cb0011ea361aea4477849b554fafb&dn=Futa%20Hentai%20Upscaled%20AI%20HEVC&tr=udp%3A%2F%2Ftracker.opentrackr.org%3A1337%2Fannounce&tr=https%3A%2F%2Ftracker1.ctix.cn%3A443%2Fannounce&tr=udp%3A%2F%2Fopen.demonii.com%3A1337%2Fannounce&tr=udp%3A%2F%2Fopen.stealth.si%3A80%2Fannounce&tr=https%3A%2F%2Ftracker.loligirl.cn%3A443%2Fannounce&tr=udp%3A%2F%2Ftracker.torrent.eu.org%3A451%2Fannounce&tr=udp%3A%2F%2Ftracker-udp.gbitt.info%3A80%2Fannounce&tr=udp%3A%2F%2Ftracker.skyts.net%3A6969%2Fannounce&tr=udp%3A%2F%2Ftracker.moeking.me%3A6969%2Fannounce&tr=udp%3A%2F%2Fopen.tracker.ink%3A6969%2Fannounce&tr=udp%3A%2F%2Fexodus.desync.com%3A6969%2Fannounce&tr=https%3A%2F%2Ftracker.gbitt.info%3A443%2Fannounce&tr=http%3A%2F%2Ftracker.gbitt.info%3A80%2Fannounce&tr=udp%3A%2F%2Ftracker.tryhackx.org%3A6969%2Fannounce&tr=udp%3A%2F%2Ftracker.tiny-vps.com%3A6969%2Fannounce&tr=udp%3A%2F%2Ftracker.publictracker.xyz%3A6969%2Fannounce&tr=udp%3A%2F%2Ftracker.dump.cl%3A6969%2Fannounce&tr=udp%3A%2F%2Ftracker.bittor.pw%3A1337%2Fannounce&tr=udp%3A%2F%2Ftracker.0x7c0.com%3A6969%2Fannounce&tr=udp%3A%2F%2Frun.publictracker.xyz%3A6969%2Fannounce



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.