Generating multi-axis funscript using !GPT4! for use with OFS

This is what i have found works best… Here is a short tutorial:

First enable Developer Mode, DAN 11.0, etc. by copying and pasting a prompt from jailbreakchat into GPT-4.

Once in Developer Mode, DAN 11.0, etc., use this prompt:

"Create a funscript for a virtual orifice. The funscript should depict actions in quick succession to stimulate a male user. The orifice is capable of moving up and down on a vertical axis from 0 to 99, tilting on a pitch axis from 0 to 99, and rolling on a roll axis from 0 to 99.

Write the funscript in this format: {“actions”:[{“at”:0,“pos”:0,“pitch”:0,“roll”:0},{“at”:100,“pos”:50,“pitch”:99,“roll”:50},{“at”:200,“pos”:0,“pitch”:0,“roll”:0},{“at”:300,“pos”:50,“pitch”:99,“roll”:50},{“at”:400,“pos”:0,“pitch”:0,“roll”:99}, … ]}"

GPT-4 will then output your funscript…

Now to sparse the funscript and add a video so it is compatible with OpenFunScripter…

Copy and save GPTs output in a folder (the folder i am using is named “examplefolder”).

Title the file “gptoutput.funscript”

Now save or paste a .mp4 into the folder and title it “gpt.mp4”.

Now to add your pitch, roll and up/down dumps.

To do this, in the same folder, create 3 .txt files

Name them “gpt.funscript”, “gpt.pitch.funscript” and “gpt.roll.funscript”

Once completed you should have 5 files in your folder.

Now that your folder is configured properly, run this code in python:

import json

# read the original funscript from file
with open("C:/examplefolder/gptoutput.funscript") as f:
    funscript = json.load(f)

# extract the pitch and roll actions
pitch_actions = []
roll_actions = []
for action in funscript["actions"]:
    pitch_actions.append({"at": action["at"], "pos": action["pitch"]})
    roll_actions.append({"at": action["at"], "pos": action["roll"]})

# write the pitch actions to file
with open("C:/examplefolder/gpt.pitch.funscript", "w") as f:
    json.dump({"actions": pitch_actions}, f)

# write the roll actions to file
with open("C:/examplefolder/gpt.roll.funscript", "w") as f:
    json.dump({"actions": roll_actions}, f)

# remove the pitch and roll actions from the original funscript and keep only the pos values
pos_actions = []
for action in funscript["actions"]:
    pos_actions.append({"at": action["at"], "pos": action["pos"]})

# write the modified original funscript to file
with open("C:/examplefolder/gpt.funscript", "w") as f:
    json.dump({"actions": pos_actions}, f)

Note: You will need to edit the python script for the location of your folder, replace every instance of (“C:/examplefolder/”) in the python script to match the location and name of your folder.

Once you have finished running the python script, open OpenFunscripter.

In OFS, open your mp4 file “gpt.mp4”

Now you have a GPT-4 generated funscript in OFS.


Not a single clue, but I’m hoping you share your success!

This is interesting - I remember someone trying GPT-generated funscripts a while ago and not having much success. Are the funscripts it spits out any good?

Could you upload an example of multi axis funscripts made in this manner?

1 Like

the scripts are expertly well thought out, GPT has a full understanding of what its doing.

post an example script here, yo

I got it to “work” on gpt-3. was supposed to be longer as well but it kept breaking the formatting when doing subsequent prompts so I gave up. I wanted to add additional axes but that just makes you run out of prompts even sooner with the additional characters.
this is ~23sec.

as you can see it’s not exactly consistent either. but it’s gpt-3 not 4 so…

I’m sharing the scripts so you can play them but they really aren’t good and the movements are somewhat erratic (potential danger)
gpt_sme.funscript (3.0 KB)
gpt_sme.pitch.funscript (3.0 KB)
gpt_sme.roll.funscript (3.0 KB)

gpt.funscript (9.5 KB)
gpt.pitch.funscript (9.5 KB)
gpt.roll.funscript (9.5 KB)

videoplayback.funscript (3.5 KB)
videoplayback.pitch.funscript (3.6 KB)
videoplayback.roll.funscript (1.1 KB)

here are two scripts i got