DE-censoring JAV with Free Open source LADA application
- Thread starter avatarthe
- Start date
-
Akiba-Online is sponsored by FileJoker.
FileJoker is a required filehost for all new posts and content replies in the Direct Downloads subforums.
Failure to include FileJoker links for Direct Download posts will result in deletion of your posts or worse.
For more information see this thread.
Hm...
(1) Chitramaya is not meant to be a real-time player - though you can trick it to behave like one
- it uses HTML 5.0 player for display.(2) It is meant to serve as a conversion (mosaic removal) studio. You can play around with restoration and detection settings, visualize your changes and decide if you want to process the whole video with those settings. It lets you compare upto 5 different changes
(3) It is geared toward people who have large collections of high-res and VR videos
(4) The only software that natively supports 3D-SBS split L/R ->process -> merge
(5) All features are free to use - including SuperRes scaling and temporal stabilization of scaled regions ( It does have button that if clicked will take you to a charity and you can donate to SaveTheChildren.org).
(7) Batch Processing capability
(8) We do not ship or train models - for mosaic removal and addition, it depends on ladapp's models, another reason why ChitraMaya will never go behind a paywall of any sort
(9) The only open source software that I know of that lets you add mosaic by graphically selecting a region or multiple regions and visualize it
(10) The only software that lets you add mosaics to non-mosaic'd NSFW content (relies on ladaapp's model)
(11) Lets you download and install models of your choice from huggingface, compile to TRT engines and run
(12) Recently started releasing Intel and AMD discrete GPU versions
You can always install and find out how it differs. I have not used Jasna much, started building my own once lada disappeared from github, never to be caught flat footed.
Few additions from my side for people that are wondering about the differences outlined above.
1. Jasna supports real time playback (via stash, browser or built-in player). On 5090 it can handle 4K@60fps with 2nd restoration.
2. Jasna does allow you to do the same via Segment Processing window - you can pick detection settings, see results and then process only selected fragments or entire video.
3. Jasna provides the highest processing speed so its perfect for people with large collections to restore
4. Jasna does support natively 3D-SBS split -> process -> merge so Chitramaya is not the only one.
5. All features of Jasna are free except proprietary 2nd restoration model (unet4x) but free users have access to TVAI and RTX Super Res
7. Jasna does support batch processing (advanced queue system)
8. Jasna has SD 1.5 for still images restoration but core restoration still depends on lada model.
9. Not sure whats the use case here, allowing user to select region probably doesn't translate well to the actual restoration process. Jasna allows user to draw their own mosaic which is then used to train next-gen rf-detr model
10. OK
11. Jasna supports all YOLO/Lada models so you can download whatever you want from whatever place
12. Jasna supports Nvidia and AMD.
Bonus:
Chitramaya high processing speed (bascivrspp sub-engines logic is directly copied from Jasna source code).
Although Chitramaya still uses an older version which consumes a lot of VRAM.
1. Jasna supports real time playback (via stash, browser or built-in player). On 5090 it can handle 4K@60fps with 2nd restoration.
2. Jasna does allow you to do the same via Segment Processing window - you can pick detection settings, see results and then process only selected fragments or entire video.
3. Jasna provides the highest processing speed so its perfect for people with large collections to restore
4. Jasna does support natively 3D-SBS split -> process -> merge so Chitramaya is not the only one.
5. All features of Jasna are free except proprietary 2nd restoration model (unet4x) but free users have access to TVAI and RTX Super Res
7. Jasna does support batch processing (advanced queue system)
8. Jasna has SD 1.5 for still images restoration but core restoration still depends on lada model.
9. Not sure whats the use case here, allowing user to select region probably doesn't translate well to the actual restoration process. Jasna allows user to draw their own mosaic which is then used to train next-gen rf-detr model
10. OK
11. Jasna supports all YOLO/Lada models so you can download whatever you want from whatever place
12. Jasna supports Nvidia and AMD.
Bonus:
Chitramaya high processing speed (bascivrspp sub-engines logic is directly copied from Jasna source code).
Although Chitramaya still uses an older version which consumes a lot of VRAM.
Thanks for posting that. I'm a jasna fan, and after trying both over the last few days, jasna is still the preference. I'm also a supporter, so I get access to unet4x. Love the program!
Kruk2 is correct, and it's not a secret — ChitraMaya's BasicVSR++ sub-engine implementation began as a port of Jasna's, and says so in the code (line #12) from the time we used it. ("Faithful port of Jasna's basicvsrpp_sub_engines.py"). Both projects are AGPL-3.0 — the license that exists precisely so projects can build on each other openly, the same way Jasna builds on ladaapp's models. What I should have done is credit kruk2/Jasna in the README Acknowledgements alongside HypoX, ladaapp, zelefans, and pifroggi, not just in the code. That has been fixed as of today — thank you for the prompt. And you're right that our port is from an older revision; if the newer version's VRAM improvements are as good as you say, that's a compliment to your work and I'll happily study it — that exchange is what the AGPL-3.0 license is for.
If you used ChitraMaya for a few days, please consider donating to my favorite charity
Similar threads
