Obscura Remova Wipes Away Visual Clutter From Video Scenes

Obscura Remova is a new video-to-video LoRA adapter that removes visual obstructions from existing footage. The model clears away haze, smoke, foreground objects, and partial occlusions to reveal the scene behind them. It works by reconstructing background details that are visible, even faintly, across multiple frames.

WepeNerd developed this adapter specifically for the LTX 2.3 video model. The goal was to reconstruct clean ground truth video from sparse or partially blocked data. Semi-transparent occlusions like smoke get cleaned up through retained scene information, while completely hidden areas are filled in through hallucination.

Scene cleanup without retakes

Video editors and content creators working with footage that has unwanted foreground elements will find this tool immediately useful. Privacy-conscious users can strip identifying objects from clips without manual frame-by-frame rotoscoping. A recommended starting strength of 1.3 gives good results, with higher values handling more complex removal tasks.

What to watch for

Fast-moving scenes still present a challenge for the underlying LTX model, though interpolating input footage to 50 frames per second before inference helps address this. The LoRA strength needs manual adjustment depending on how many occlusion layers must be removed from a clip. When background details never appear in any frame, the model must rely entirely on its own scene generation rather than real pixel data.

"If the data is completely invisible from any of the frames then it will have to be hallucinated, but if it is present anywhere, even semi-transparently, then it will be retained." — Source: Hugging Face