Skip to content
Community Research


At the fast evolving edge of the image synthesis frontier weights can be confusing. From the perspective of what makes a neural network do its thing, weights are one of the most important concepts to understand. In the crafting of elegant prompts and mods, the meaning of weights under the hood isn't all that different from, but fittingly enough their emphasis in context is. It's all a matter of perspective really, from heavy to light.

The only relevant difference between weights for models and for prompts / mods here is who's doing the weighting.

Either way, weights are how the significance of factors within a model or prompt are balanced in relation to each other. As you'd imagine, they can have a very large impact on synthesis results.

While you find, test, and contribute to weights matched with example results in the growing collection of open research here, we can't help but pass on the obvious Godot joke.