XplicitRecAI

What does the tool do? Can it process and shrink my existing video collection by automatically cutting and merging clips with action?

The tool is designed to process your existing videos, automatically detecting and cutting out clips with relevant action (such as "pussy" and/or "bating" based on "Process Only Specific Content" in the cutting settings), and optionally merging them. To shrink your current collection, use the cutting module. The only requirement is that your folder structure follows this format:

└───InputDirectory
    ├───Performer_01
    │       Performer_01-2025-06-18_02-00.ts
    │       Performer_01-2025-06-18_04-00.ts
    │
    └───Performer_02
            Performer_02-2025-06-18_04-00.ts
            Performer_02-2025-06-18_07-00.ts
      

Select the input directory where your recorded videos are stored, and the tool will handle the processing by scanning the input and placing the results into an output directory.

How does the merging of clips work?

Merging happens based on the video metadata, specifically the creation time and last changed time. For example, if a file is created at 12:00 and last edited at 12:30, the tool assumes the clip is 30 minutes long. If another clip is created at 12:35 and last edited at 12:45, and the merge gap in settings is set to 5 minutes (or 300 seconds), these two clips will be merged. The key factors are the creation time, last edited time, and the configurable gap. You can adjust this in the recording settings under "Video Merge Gap (Seconds)". Clips are only merged if they are from the same subfolder (e.g., Performer_01 or Performer_02).

How does the automatic cutting work?

There are two modes for automatic cutting:

• Mode 1: Uses both detections the AI model is trained on ("pussy" and "bating").
• Mode 2: Uses only the "bating" detection.

You can select this via "Process Only Specific Content" in the cutting settings.

The process works as follows:

• The tool scans each directory and file in the input directory, supporting *.mp4, *.ts, *.mkv, *.mov, *.avi, *.wmv, *.flv, *.m4v video files. While these file types are all supported, only .ts, .mp4, and .mkv are heavily tested and reliable—no guarantee for others like .mov and so on.
• It captures a frame every 10-15 seconds (configurable in settings under "Post-Detection Skip (Seconds)" and "No-Detection Skip (Seconds)").
• AI detection is performed on each frame; if detections are found, the frame is marked.
• After processing the entire video, clusters are created based on detections (e.g., if "pussy" or "bating" is detected in frames between 00:45:00 and 01:00:00).
• The tool cuts out the relevant clip (e.g., 15 minutes) and saves it, discarding the rest where no detections occurred.
• Detection threshold is 33% by default (configurable).
• A buffer can be added around the first and last detection (e.g., 120 seconds) to capture action that starts slightly earlier or ends later.
• You can perform a dry run where no original videos are deleted, just to test the process.

There are more settings to be used, but for starters, a user may only test with some files and an input and output directory to get a feeling for the working of the tool.

Is there a trial version available for the software?

Yes, we offer a 7-day trial version. You will be prompted to start the trial on the first run of the app. If a user would like an extended time or has questions, they may ask, but soon support will be on holidays for a long time (1-2 months), so no guarantee on coming straight back to emails.