Idea for use of ChatGPT-4 to draw on Moho

[ImpMontezuma]

While I working in one of my drawings today, I was thinking of the potential to create a tool to automate Moho processes with the help of ChatgPT. As you can imagine the potential is immense. I tested an idea a few days ago to create a plugin to automate the creation of a more complex finger joint (still working on this). And the experience has shown me that chatgpt really is good for programming using LUA, you just need to teach the basic functions for it. And he is able to make all the code alone. Then I thought, I could take a step beyond that. Perhaps we can teach chatgpt to make the whole drawing process. And again, using his new ability it has to understand images the potential for this is huge. Take a look at the test I did. I asked for chatgpt to create the steps on how to create this drawing using basic 2D shapes. (We use an imaginary canvas of 1080x1920 as coordinates). And here is his answer:



"Absolutely, Let's try this! Given the provided canvas dimensions, here's a Rough Estimation for Your Character Silhouette Based on Your Previous Drawing:

Head:

Shape: Oval
Center: (540, 1700)
Dimensions: Approx. 80x100 (width x height)
Torso:

Shape: Oval Larger, Slightly ELONGATED VERTICALLY
Center: (540, 1500)
Dimensions: Approx. 250x350
Left Arm (Mirror For the Right Arm):

Shape: Oval (Upper Arm)

Center: (480, 1470)

Dimensions: Approx. 60x90

Shape: Slightly Smaller Oval (Forearm)

Center: (440, 1370)

Dimensions: Approx. 50x80

Left Leg (Mirror For the Right Leg):

Shape: Oval (Thigh)

Center: (520, 1250)

Dimensions: Approx. 110x180

Shape: Slightly Lonar Oval (Shin)

Center: (520, 1080)

Dimensions: Approx. 90x220

Tail:

Shape: Thin, ELONGATED OVAL
Center: (540, 1000)
Dimensions: Approx. 30x300
EARS:

Shape: Small Ovals
Left Ear Center: (500, 1750)
Right Ear Center: (580, 1750)
Dimensions: Approx. 50x70 EACH
Now, these are Rough Estimates and Will Likely Need Adjustments When Implemented. The Ovals' Rotations are not Accounted for Here, But with the Given Centers and Dimensions, It Would Be Possible to Rotate Them To the Desired Angles Later On. This is a Basic Mapping, and the nuances of her form can be adjusted as Needed.


Anyway, I followed the instructions it made, manually, on Photoshop, and this is the result.



Well, far from perfect, I know, but all the correct shapes are there. I mean, if he is able to create a code on LUA for a plugin that would automatically create those shapes that would already be very useful (at least for me). Anyway, again, the potential is huge.

So, the idea here would be to show ChatGPT a drawing and ask it to create a code on LUA to automatically recreate the basic shapes, and then use this code to generate those shapes on Moho. I'm pretty sure that can be done relatively easily, And obviously, this concept can be applied to make much more complex stuff. Like, using its API to create a module inside Moho itself, to automatically create those shapes based on a bitmap image of our choosing, and who knows, maybe one day this same principle can be used for ChatGPT to create the whole animation itself

PS: If you are afraid of losing your job with this, just know, that those AIs still need a lot of assistance to be able to do those things, so I'm pretty sure no one here would be seen as replaceable any time soon. In the end, those AIs are just another tool to help us to automate tedious work.