Ask HN: Why are AI generated images so shiny/glossy?

Many AI-generated images you encounter are low-effort creations without much prompt tuning, created using something like DALL-E or Llama 3.1. For whatever reason, the default style of DALL-E, Llama 3.1, and base Stable Diffusion seems to lean towards a glossy "photorealism" that people can instantly tell isn't real. By contrast, Midjourney's style is a bit more painted, like the cover of a fantasy novel.

All that being said, it's very possible to prompt these generators to create images in a particular style. I usually include "flat vector art" in image generation prompts to get something less photorealistic that I've found is closer to the style I want when generating images.

If you really want to go down the rabbit hole, click through the styles on this Stable Diffusion model to see the range that's possible with finetuning (the tags like "Watercolor Anime" above the images): https://civitai.com/models/264290/styles-for-pony-diffusion-...

Uhm, Llama 3.1 is an LLM.

whats a good prompt if i want a schematic/engineering/wireframe look for all my objects ?

Most models seem very reluctant to do that. (Historically, rendering full 3D was also easier than rendering wireframes. Art imitating life.)

DALL-E's style is intentional to prevent misuse of fake but near-undetectable highly realistic images.

Ah, my mistake. "Meta AI" can generate both text and images, but apparently text prompts are handled by Llama 3.1 while image prompts are handled by Emu. I initially struggled to find the name of the image generation model.

Oh, I didn't even realize Facebook had a text to image model.