Exact same seed & SD setup (no starter img) but 2 = SDXL & 2 = SD1.5 with image size changes
Here's the Seed and set up info the Model (or SD version) is the change for most of the examples shown in this thread while the SDXL only changed Model & Aspect Size (square to landscape)
I took a 17-Grid to reduce the presented shape down from a max-17-point number to a max-4 point-number by making all singular positions stuck to a sub-section of "0" dead center.
0 === single box
Reading L2R, place per dec pos (1|10|...)
Reading T2B, Add per column
x1,x...,xn
So if we have any x-linear obj it'll be a string of same numbers ie 111, 11, 1111, etc (the number corresponding to the row)
while top-layered objects ads confusion and obstruction ie 3223 is two duo-stacked objects with a single object high pole in between w/2 blocks between em
I know there's people trying to figure out how to determine all shapes without repeating possible-shapes for connected shapes and this may be an efficient way to do that besides hashing matrix-mapping of hit & misses.
AI Speech is something I rarely hear people talk about.
AI has a way to "talk" including a language of "sounds" instead of words. Spacing is strange but more related to thee entire content not the phrase or sentence.
Let's "Show Mental Ores" to "Overt the Teenage Hack"
0/--
While trying to take an image and get AI to just change the character from male to female while keeping the art style of the image used, I noticed the text in the sign
I've wondered about AI text before but when I asked AI to, "Show me what you are thinking" I got this one
1/--
The Blonde Hair Women all came from the one in the prev tweet so here's the info:
"Live Still or Epic" "Dice Down in Pieces"
"Inherit Dominate Perk" "Pervert Pour In Leach"
Maybe terms to live by, maybe truthful information...you decide.
How does AI handle Ray Tracing and why is so unlike game RT but same results?
Here's a micro thread on the results of Stable Diffusion creating images with Ray tracing and how, things are getting better for the imagery with AI art like SD.
Today's subjects:
First we will start light with the room (inside view) and how it tracks the light pretty well and how there may be more going on upstairs that we can't see.
IMG A: All possiblities at once
IMG B: Light range for highlights based on imagery
IMG C: Shadow finding based on traced light ranges for highlights
SD tracks the verticies based on the light source to find highlight ranges, then uses crossing bubble ranges as shadow area