This ComfyUI workflow restores old or low-quality footage using an LTX 2.3 Video-to-Video LoRA. It splits your video into frames with VHS_LoadVideo, enhances each frame with the custom LTX 2.3 V2V processor (class_id: 2e37ecc9-770a-4ac2-972c-e953901f49d3), then reassembles the result with VHS_VideoCombine. A MarkdownNote in the graph provides inline tips so you can adjust settings without leaving the canvas.

Technically, VHS_LoadVideo handles decoding and frame range control: use skip_first_frames to trim the head of a clip and frame_count to limit how many frames are processed. Those frames are passed to the LTX 2.3 Video-to-Video node, which applies the LoRA-driven restoration to recover texture, lines, and fine detail common in archival film and anime sources. You’ll typically balance LoRA strength and denoise to preserve motion while lifting detail. Finally, VHS_VideoCombine encodes the restored frames back into a video, where you can set the output fps and codec to match your delivery needs. The result is a repeatable, frame-accurate pipeline for bringing older footage back to life without hand-tuning every shot.

FAQ

よくある質問

すべてのワークフローを見る
Character
Cinematic
Image to Video
Lip Sync
Multiple Angles
Portrait
Style Reference
Style Transfer
Text to Video
Video Generation
Video
Showing 30 of 579 templates