I've tried training a Lora that would generate images in the allegria / corporate art style. Why? To make the world a better place of course!

It did turn out better than I expected, and seems to be able to draw a simplified representations of objects. Possibly, the style is just able to extract the knowledge of simple icons etc. from the diffusion model. Though, it's obviously not able to create a perfect image.

To fully apply it I remommend using both corpoart and flat colors in the prompt.