Classifying action script style

Life is too short to run scripts you don’t like.

I spend a lot of time downloading scripts to see if the match what I’m after in terms of style. Intensity is mostly covered by heatmaps but those don’t always give a great picture of what the script is like.

The idea
Find a way of classifying (action) scripts by some different parameters, to make it easier to find and run the scripts you like. Scripters could, if they want, provide the script style in the script post.

A suggestion on parameters:

Accuracy/effort: Low | Medium | High
How much effort has the scripter made to make the script match the vid? Low could be an untested on the fly script, High could be a frame by frame script where the scripter spent hours perfecting it.

Extra strokes: Low | Medium | High
How much extra (not depicted) strokes like extra wiggle or filler strokes?

Stroke length: Realistic | Exaggerated | Full stroke
How much is the stroke length extended compared to the vid?

Is this a good idea? Would it help you to judge which scripts to DL?

What do you think about my suggested parameters to describe action script style (along with heatmaps of course)? Anything missing/unclear?


Accuracy: meaningless value to apply as effort is relative for the creator. Someone who used AI for 80% of the script might still make a better script than the one who did everything manualy.

Extra strokes: This part is too vague as identification. Its not just fillers (which need their own tag anyway), but script style in general. Smooth strokes, shaking bumps, Movement detail etc are all things that matter here. Movement detail being a complex case here as when and how often are you going to apply a speed change midstroke. Script detail, target device etc all matter (some machines are better at handling small details, while a handy might stutter).

Stroke length: Again depends a bit on the machine. Realistic on an OSR or on a handy makes quite a diffirence. Target device is on that a key aspect. But at least realistic and exaggerated at least works for identifying needs. Full stroke is meaningless again as thats more about scripting style in general (in a video full strokes might still have a varying length, but in most cases that doesnt matter at all).

So here my suggestion to change extra strokes and stroke lenghts:

  • Stroke detail: Simple, Smooth (with node count for smoothing), Excessive
  • Stroke style: Realistic, Exaggerated, Shaking.
  • Target device

The smoothness is for example 4 nodes, which means that for a direction change, 4 nodes are generaly used. Simple in this case would be 1 or 2. Excessive in this case is generaly for devices that can handle like 20 operations per second smoothly.

The style differs between the types you can desire. Shakes are already amplified versions, so exagerrated or realistic are both excluded here anyway. And distance at that point isnt as relevant anymore. Other than that, it seperates the styles fine. Vibrations will fall under the shake part.

Target device is then the last supporting category, as that defines the scripts limitations. If it says OSR, you can be quite sure the script isnt going to be handy compatible (or it clips movement), while the other way around it might be fine (or have disruptive stutters). By knowing both the targeted and your own devices, the standard quirks between the 2 are often known enough, so you can make a prediction here.

1 Like

Good points you made here. But wouldn’t this classification have to be done by someone other than the scripter. We need something like script reviewers who, for example, receive a paid script for half the regular price and in return oblige to write an objective, uninfluenced test report. Maybe I should test that when I release my next paid script. I could do that with a gumroad coupon code.


I’d be into that, hmu when you release next…


Thanks for the comments, particularly @SomeoneRandom

I still think Accuracy has a place. Sometimes I do a fast motion tracked script without going through all points frame by frame, sometimes I spend hours and hours. But maybe Method could be better? A short description of how the script was produced, maybe some standard alternatives like On the fly | Frame by frame | Motion tracked

So I suggest:

  • Target device: Handy, OSR2, Launch, Keon, Vibrator, E-stim, …
  • Stroke detail: Simple, Smooth, Excessive
  • Stroke style: Realistic, Exaggerated, Shaking (can be a combo)
  • Method: On the fly, Frame by frame, Motion tracked (can a combo)

Tried using this format in this recent script

And this also can be provided in a very accurate way, like: 80% motion tracked, 15% finetuning (basicly motion tracked with some adjustments), 5% frame by frame. And the naming here, while some strandard terms are welcome, is quite flexible. Its mostly a description of the used method so whatever was done can be named.

And its not required to be very accurate here either. As its just a rough indication (where usualy that 5% frame by frame is the indication that matters a lot since that defines hard to script sections got the extra effort). If its 90% motion tracked, and 10% filler.

But still, for the standard terms there i would say:

  • Frame by frame/on the fly (its user input, 1 category will do fine here as scripting often doesnt even need frame perfection)
  • AI/Machine learning/Motion tracked
  • Input tracked (this is when users use a device to track position, for example a mouse, or any other method - seperated from motion tracked)
  • Finetuned (motion/input tracked script is used as base, but adjustments were made)
  • Filler (content that was added for sections without action)

More than this isnt realy needed as creators often have their own style and qualities. Some script in frame by frame style by default, but dont consider high accuracy to be worth here, while some will try to get every frame to perfection. So this detail would only be vague. The only reason input tracked is a diffirent one is because this often is at a lower quality, but higher position variety (less positions rounded to intervals of 10).

1 Like