Admin update: this model is depreciated and will not receive any future updates.

To be able to generate AI Scripts yourself, keep an eye on our social media and forum thread for SLR Drive

The manual scripting process requires a lot of time and effort on the side of scripters. With production of VR content ramping up across the board, the gap between the size of SLR’s video catalog and the number of available funscripts is ever increasing. In order to help deliver scripts for as many videos as possible, our computer vision team has been developing an AI assisted system to help speed up this process. This is a preview of the current state of the application, which is still under active development.

DOWNLOAD

[ScriptAI App Preview Version 1.1.3.0]
The app is available for Windows and requires a free SLR account.

For the current preview, pre-generated AI tracks are available for the following videos (right click the "Data" link and select "Save As"):
SLR Free Videos

  • 194 [Video] [Data]

SLR Premium Videos

  • 11384 [Video][Data]
  • 21475 [Video][Data]
  • 26753 [Video][Data]
  • 29192 [Video][Data]
  • 31571 [Video] [Data]
  • 31641 [Video] [Data]

QUICKSTART

  1. Download the application and extract the contents.
  2. Select a video from the list. Download both the track and the video.
  3. Launch the application.
  4. From the File section of the menu open the video and load the tracking data.
  5. Switch to the Other_1 track in the dropdown menu to the right of the video controls.
  6. Select the entire secondary track and insert it into the main editor track with the right mouse button context menu.
  7. For all controls and the intended workflow see the instructions in the Help section of the menu.

Video overview

ABOUT

ScriptsAI is a new system for AI assisted funscript creation. It’s the successor to our previous automation tool, which was based on visual tracking. It’s intended to enhance the script creation workflow, by automating the tedious and repetitive parts of scripting. It consists of two parts:

  • Our AI model automatically generates multiple action tracks that follow the actor’s motion from the video, as well as a recommended primary action track.
  • The application contains tools for bulk point editing as well as manual scripting. It enables scripters to combine, tweak and clean up the generated tracks into a complete funscript, as well as incorporate their own personal style and flair.

FEEDBACK

Both the AI generated tracks and the app are in active development and are continually being updated and improved. We welcome any feedback you may have. We will be reading the replies. Let us know what works well, and what we could improve in future releases. Also let us know regarding features you would want for manual editing and scripting within the app.

AI PREVIEWS

The following previews visualize our AI in action. The body part detection overlays are not visible when editing scripts in the release version of the app.

[Blowjob]
[Cowgirl]
[Doggy]
[Missionary]

APP PREVI



AI NOTES

Our AI model detects the motion of different actions the actors are performing in the video. It currently covers handjobs, blowjobs and hip thrusts. It also creates a track that automatically switches between the actions, based on what’s most appropriate. The model currently excels at tracking clear continuous movement, such as cowgirl, missionary and extended blowjob sections. It currently doesn’t perform as well with certain parts of the action that are not clearly visible, highly overlapping interactions, and transitions between different actions.

Changelog

1.1.3.0

New features

-Preferences menu for configuring user settings.
-Configurable keyboard shortcuts for scripting and navigation actions.
-Invalid stroke detection. Invalid strokes are colored red in the timeline.
-Configurable min stroke length and max stroke speed values for invalid stroke detection.
-Buttons underneath the stroke statistics for jumping to the next or previous invalid stroke.

New Keyboard Actions

-Adjust the amplitude of the closest point or the entire selection.
-Set the amplitude of the closest point or the entire selection.
-Shift points backward or forward frame by frame.
-Shift points and moving backward or forward by one frame.
-Shift and deselecting the next or previous point.
-Shift and deselecting points in the next or previous frame.
-Move the selected point or active selection.
-Select and deselecting all points.
-Increase and decreasing playback speed.

Improvements

-Smoother position indicator movement during playback. Position is refreshed during UI render.
-Funscript no longer jitters when panning the timeline at lower zoom levels.

Changes

-Left & right keys now move to next frame instead of 100ms.
-Jumping to next point is now done with up & down keys by default.

1.1.1.1

  • Fixed an issue where some users were unable to log in.
  • Fixed incorrect funscript range calculation.

1.1.1.0:

  • Added support for loading .mov video files.
  • Added option to lock point movement to the horizontal axis.
  • Added amplitude indicators to stroke visualizer.
  • Added amplitude and timing indicators to point timelines.
  • Added precise timeline panning with mouse.
  • Added video zooming and panning.
  • Added SLR free account login.
  • Fixed slight drift when moving points with the mouse.

FAQ

Can I use this tool to automatically script my own videos?
Currently only pre-generated tracks are available. Due to the nature of the AI model, the generation is performed on our servers. In the future, we plan to give access to users through our service.

vztfu changed the title to SLR Scripts AI Preview: AI Assisted Funscript Creation .

Very interesting. Will definitely test it and give some feedback after some usage.

How do I download AI tracks? When clicking on them it opens just a tab, with all the information in the file. Nothing downloads.

    sl210
    I will see if anything can be done about that. For the time being, I will add your comment to the instructions.

        sl210
        Thank you, worked.

        vztfu
        Is this tool intended to actually script in like OpenFunScripter? Or purely for the AI component?

              Danny88
              Its primary purpose is to combine and edit the AI generated tracks, as well as fill in the gaps by manually scripting. You can absolutely create fully manual scripts with it as well, but it currently doesn't cover all the features that dedicated manual scripting apps have. If you would like to have features added, even for purely manual scripting, let us know. Looking forward to your feedback.

                  vztfu
                  Okay. I will do some testing and probably leave some feedback tomorrow. Wanted to know what the intention of the tool is first.

                    Okay, I did some testing. I don't want to sound harsh or rude, but in it's current state the program is not usable at all. Only having the double pane view and no way to precisely skim through the video / stroke points make it impossible to correct the ai generated script. I made a list for the most important changes we need to make it usable.

                    General:

                    • Option to only show the view like in vr (single view instead of both panes). It's not possible to correctly check if the positions are correct with the current view.
                    • Ability to proper skim to one point to another. Make it like OFS / JFS, where you use arrow keys up and down to go to the next / last point.
                    • A better view on the timeline to know the stroke length. Make it like OFS, where the timeline has small squares and a line in the middle. By just looking at the timeline you know the stroke length.
                    • Also ability to hold middle mouse button and drag the timeline to the left and right.
                    • Stroke visualizer also needs to show the current stroke length.
                    • Colors to indicate the speed of the strokes.

                    Some qol sutff:

                    • Ability to drag a video in instead of file -> open
                    • Option to set max speed of strokes. If stroke is higher, highlight it.
                    • Ability to change size and color to the stroke visualizer. Also having lines for the visualizer.

                    In it's current state too rough to use it properly. But with these changes I can see it being used together with the ai scripting.
                    But to be quite honest, I don't see a reason to use this over OFS. I would just save it as a funscript and then load it into OFS and correct it there.

                      Danny88
                      Thank you for trying out the app and all the feedback. It's very valuable to us. This is the very fist preview of the app, so we are still working to improve it and add necessary features.

                      • We are planning to add the option to zoom and pan around the video view. We can add the option to hide half of the view as well.
                      • Jumping between points is already present. If you missed it in the instructions, the shortcut is shift + right/left, or the "|<" and ">|" buttons in the player controls. Tooltips are also planned for all UI elements, so it's more clear which controls do what.
                      • All the other features you mentioned are also already planned. Based on your response, we'll increase their priority.

                      As for the comment regarding just using OFS. Having a dedicated app allows us to be much more flexible with all the features we plan to add in the future. In the current version it gives you easier access to all the different tracks that the AI generated. And having different apps available is always good, as it gives you more choice to use whatever works best for you.

                          vztfu
                          Thank you for considering my feedback!

                          vztfu Jumping between points is already present. If you missed it in the instructions, the shortcut is shift + right/left, or the "|<" and ">|" buttons in the player controls. Tooltips are also planned for all UI elements, so it's more clear which controls do what.

                          I am that type of person who never reads instructions and then wonders why certain things are not working ^^ That's on me then, didn't see it. This is also probably something you are plan to add in as well, but having the ability to change all the hotkeys how we want them would be good to have.

                          vztfu And having different apps available is always good, as it gives you more choice to use whatever works best for you.

                          Absolutely. I can see you guys making a very good scripting tool. You have the skill and the resources to do so. I am very much looking forward to the next version and will be happy to give you more feedback on the upcoming versions.

                                Is this just a demo to show what you have done? Or is there some way to use this tool on a video of our choosing?

                                Also " We can add the option to hide half of the view as well." is not what is needed, your tool would need a VR view like OFS to be useful for actual funscript generation, although I suspect I would use this only to generate a draft script, then pull it into OFS for final editing.

                                I have a lot of 29192 scripted and I can a lot of similarities between your work and mine.

                                  Rakly3

                                  Ok, so can someone post some steps on how to process a video from scratch?

                                  I don't see anything in what was posted or the tool that would help me get started to do that.

                                  Thanks.

                                    @vztfu Can you explain what the idea is with our AI and scriptAI app?
                                    I don't know what I'm allowed to say, nor am I 100% I have all the details correct.

                                      Rakly3

                                      If you can't post more information but are allowed to help me, my contact info should be available to you (or you can get it from Elena or Realcumber).

                                        scrptr9
                                        This is a preview showing the current state of our AI generation model and the app for integrating that data into a funscript. It's mainly intended to get feedback on both the quality of the AI generated tracks as well as the user experience of the editor, so we can improve both. The AI generator is not a part of this preview, only the six generated data files can be pulled. All of them are unscripted on SLR at the time of this release.

                                        Thanks for clarifying about the VR view. We will look into adding that feature as well.

                                          7 days later

                                          We have added a new AI generated script for a free video on SLR:

                                          The app has also been updated:
                                          ScriptAI App Preview Version 1.1.1.0.
                                          This version includes a number of requested changes.
                                          Changelog:

                                          • Added support for loading .mov video files.
                                          • Added option to lock point movement to the horizontal axis.
                                          • Added amplitude indicators to stroke visualizer.
                                          • Added amplitude and timing indicators to point timelines.
                                          • Added precise timeline panning with mouse.
                                          • Added video zooming and panning.
                                          • Added SLR free account login.
                                          • Fixed slight drift when moving points with the mouse.

                                          New features can be seen in this gif:

                                          Hope you enjoy the new version.
                                          We are working on implementing other requested features.

                                          9 days later

                                          mnmlikereal Works fine for me, although this version requires your slr website login (the previous version posted did not require a login).

                                          Also when you reply, you need to reply to the post so the other poster gets notified you replied, i.e. @vztfu should look at the post above.