Comment by nottorp 2 months ago But world building and graphic assets are repetitive, time-consuming work. 7 comments nottorp Reply nasmorn 2 months ago The industry tries to use generative tools for assets since forever. Textures, terrain, foliage and eventually even parametric human models. I don’t see transformers much different. nottorp 2 months ago Oh I do. Because with a tree generation library the trees do look random (they are because the algorithm has a random component)."AI" assets all look the same. Clamchop 2 months ago This observation seems innately vulnerable to sampling error and confirmation bias. "Good" generative AI almost by definition is difficult to notice. 3 replies → dheatov 2 months ago The stochastic nature, minimal amount of care required to get something going, and the inefficiency, just to name a few.
nasmorn 2 months ago The industry tries to use generative tools for assets since forever. Textures, terrain, foliage and eventually even parametric human models. I don’t see transformers much different. nottorp 2 months ago Oh I do. Because with a tree generation library the trees do look random (they are because the algorithm has a random component)."AI" assets all look the same. Clamchop 2 months ago This observation seems innately vulnerable to sampling error and confirmation bias. "Good" generative AI almost by definition is difficult to notice. 3 replies → dheatov 2 months ago The stochastic nature, minimal amount of care required to get something going, and the inefficiency, just to name a few.
nottorp 2 months ago Oh I do. Because with a tree generation library the trees do look random (they are because the algorithm has a random component)."AI" assets all look the same. Clamchop 2 months ago This observation seems innately vulnerable to sampling error and confirmation bias. "Good" generative AI almost by definition is difficult to notice. 3 replies →
Clamchop 2 months ago This observation seems innately vulnerable to sampling error and confirmation bias. "Good" generative AI almost by definition is difficult to notice. 3 replies →
dheatov 2 months ago The stochastic nature, minimal amount of care required to get something going, and the inefficiency, just to name a few.
The industry tries to use generative tools for assets since forever. Textures, terrain, foliage and eventually even parametric human models. I don’t see transformers much different.
Oh I do. Because with a tree generation library the trees do look random (they are because the algorithm has a random component).
"AI" assets all look the same.
This observation seems innately vulnerable to sampling error and confirmation bias. "Good" generative AI almost by definition is difficult to notice.
3 replies →
The stochastic nature, minimal amount of care required to get something going, and the inefficiency, just to name a few.