WoLNamesBlackedOut for OBS

WoLNamesBlackedOut for OBS v1.0.6

Go to download
Overview Updates (3) History Discussion
Calocen Rieti

Calocen Rieti

New Member
Calocen Rieti submitted a new resource:

WoLNamesBlackedOut for OBS - This is a filter plugin that blacks out character names on the FF14 screen when streaming via OBS.

WoLNamesBlackedOut for OBS

This plugin hides character names from the FF14 screen when streaming (capturing) with OBS.
This is an OBS plugin port of the Windows application "WoLNamesBlackedOut".
It detects and filters usernames using DirectML.

The original is here:
https://github.com/royshil/obs-detect

This fork adds the YOLODetector class and supports custom YOLO26 models.
It also improves FPS by threading inference.

Changes from the Original...
Click to expand...

Read more about this resource...
 
Calocen Rieti updated WoLNamesBlackedOut for OBS with a new update entry:

We've added the ability to set areas to exclude from masking.

If you want to mask other characters but not your own, you can set the exclusion range to only the area around your character's name using "Exclude Options". This will exclude the area around your character's name from masking.
If the name detection area overlaps even slightly with the mask target area, it will be masked.

While this is a simple implementation, it was chosen because implementing countermeasures using detection methods is difficult due to processing speed and other factors.
Click to expand...

Read the rest of this update entry...
 
Calocen Rieti updated WoLNamesBlackedOut for OBS with a new update entry:

v1.0.6

  • The inpaint radius has been changed from a fixed value to a variable value using a slider.
  • The initial values for blur and mosaic have been changed from 0 to 3.
  • The "None" mask type has been removed.
    If you do not want to use a mask, please uncheck the mask option.
  • Multilingual support has been added.
    The following language files have been added:
    Except for Japanese, translations are done by AI, so please provide feedback if you have a more appropriate translation...
Click to expand...

Read the rest of this update entry...
 
I've completed all the relatively minor improvements I could think of at the moment.
Please post any suggestions for further improvements. I'll consider them.
The current model is a repurposed version from the Windows app and is primarily designed for in-content detection. I believe it's insufficiently trained for areas like urban environments, making it unsuitable for streaming.
The next update will involve updating the detection model, but the annotation process is time-consuming, so there might be a gap of several months between updates.
Of course, I'd like to implement minor improvements, so please feel free to provide feedback if you have any suggestions.

私が今思いついた比較的軽い改善はやり切りました。
何か改善して欲しい内容があれば投稿ください。まずは検討をしてみます。
現在のmodelはWindowsアプリ版を流用しており主にコンテンツ内での対応を目指したもので、配信向けとしては街中など、学習が不十分と考えています。
次のアップデートは検出モデルの更新を考えていますが、アノテーション作業に時間がかかるため、数カ月、アップデートの間隔があく可能性があると考えています。
もちろん軽微な改善は実施したいと考えていますので、何かありましたらフィードバックのほどよろしくお願いいたします。
 
You must log in or register to reply here.
Back
Top