AI-Assisted-Map-Planning

From MapSpawn
Revision as of 08:28, 7 February 2024 by Zastels (talk | contribs)
Jump to navigationJump to search

Midjourney and ChatGPT examples of map planning


Blanket is my first map made from start to finish using AI to help plan ideas. I used AI to enhance my workflow, it did not replace any aspect of my workflow. Midjourney taught me that what I desired was Low Key Lighting and incandescent lighting; or that is to say full spectrum lighting in contrast to the limited spectrum of fluorescent bulbs.


I knew going in I wanted an office look and feel, but there was going to be a level design shakeup for my officer paradigm. After a few tries on Midjourney, I found what I was looking for. I think AI is sometimes used improperly, and as you can see I have used it to enhance my workflow and to more efficiently find my direction. The way the AI images look and my map Blanket looks are not at all the same, that is because I am always working from my memory and not trying to copy an image. AI images are also not very spatially intelligent, a careful observation can sometimes reveal there are physical impossibilities or improbabilities.


When it came to texture creation, AI came in handy again. Goldsrc relies heavily on texture expression, and what I do is take an AI image I like and then build the image out into 3D in the scene. This breaks away from it being just a generated image. If you consider where we come from, taking photorealistic pictures and trying to downscale them to Goldsrc, this is the same principle that always existed. You want images without predefined shadows, and that are not overly detailed and present in a large exaggerated frame. For example a window in Goldsrc Counter-Strike realistically is always at least 10 feet in the air at the highest. This is because a player will jump on the windowcill, and standup. You take into account what the player needs and bend the world around it, so you get the aesthetic of everything being enlarged.


This is where Counter-Strike 2 falls on its face, the developers simply do not know how to make Counter-Strike maps any longer, and perceive 'realism' to be some sort of guiding light -- and it's not. I'm teaching you what Counter-Strike mapping really is, and it bends to the players reality and needs, producing in the end an abstraction of reality, not realism.




Temp