Little Lark No Trace Text | LTX2.3 Watermark and Subtitle Removal Video Repair Workflow

This is a seamless subtitle removal / watermark removal / garbled text video restoration application based on LTX2.3 video model + Edit Anything LoRA + Inpaint Masked LoRA + Masked Inpaint + LTX Video Condition Encoding + Video Frame Synthesis. After a user uploads a video with subtitles, watermarks, AI garbled text, logos, or screen obstructions, the system performs inpainting on the problem areas based on the masked region. It restores the text and watermark areas into natural visuals while preserving the original video's subject, motion, camera angles, and background continuity as much as possible.

The core advantage of this workflow is: Instead of simple cropping, blurring, or overlaying, it uses LTX2.3 to perform generative restoration on the masked region. It is particularly well-suited for handling residual subtitles, English garbled text, Chinese garbled text, watermarks, logos, sticker text, edge text, and localized occlusions commonly found in Kling / Seedance / AI video generation. Compared to traditional watermark removal, it is better suited for "seamless background completion," ensuring that the restored areas match the original video's lighting, texture, and motion as closely as possible. When using it, it is recommended to focus on controlling three elements: the input video determines the original content, the mask region determines what to fix, and the global prompt (used to write overall visual rules) determines the style the restored image should maintain.
🎁Claim your RH Coins first, then go experience the workflow! Click your avatar in the upper right corner → Invitation Code → Enter [rh-v1111] to instantly receive 1000 RH Coins, and log in daily to get another 100 RH Coins~ For more ComfyUI workflows, tutorials, and gameplay, please follow our official WeChat account (AIKSK), with simultaneous updates on Douyin / Bilibili / Xiaohongshu / YouTube (AI-KSK)✨