Boring Reality LoRA for Flux Dev 2

This LoRA is an early experimental training attempt at Flux Dev 2. The results are not perfect, but they should have at least limited improvement on Flux generations (distortion and disfigurement will increase with issues like text getting worse.)

A noticeable impact on this Lora is not just that it increases the "realism" of the images but that they tend to have better world knowledge and can produce better results in other styles such as cinematic shots and animation.

You will usually have to increase the guidance by a lot when raising the scale of the LoRA for better results. More descriptions in the prompt tend to help as well.

There is still a lot of potential on Flux Dev 2 from the creativity side of each generation that other recent models have not been able to achieve.

NOTE: The ComfyUI version may have not been converted properly and may now perform as well as the Diffusers version. There may need to be better improvements in the workflow provide and what not as well.