Thread by @johnstark_ai

To feed on this flow, how about a small thread on vector testing?
First of all, what is a vector? An image vector is a scaffold that aims to consolidate as much unbiased information as possible into a single image. Think of it like a ball of clay. Take the below example.

I can take this ball of clay, and shape it into whatever I want even given the most basic of prompts. The clay analogy is apt because the ball can only be manipulated within the scope of its properties. #aiart #midjourney

This would of course get more interesting with more complicated prompts, and you can take the vector and experiment as you like, but where it really gets interesting is its use for precision.

Say for example, you wanted to use the properties of this vector but would like if you could mold those with a character portrait so you could use the combination of both as a scaffold?
i3=(i1+i2), a visualization of the imagined scaffold.

Then I will start a new tree with this vector scaffold, with a disrupting prompt that I can use to manipulate it: i2=p+i1 given p=a werewolf king holds the globe in his hands, Frazetta study --niji --ar 4:5

I will then continuously reinforce the prompt with more an more polished upscales, and maybe disrupting the scaffold with additions that cater to that specific lineage of iterations. i3=dp+i2 given the disrupted prompt adds "ornate armor, long flowing cape"

This can be taken down an infinite number of reinforcements, and the image will come out more cohesive the further down it goes.
Final: i4=p+i3
#AIart #AIArtwork #midjourney