These settings can be found in the “advanced” tab when creating a model. They allow you to experiment more and get the absolute best results when you find settings that work.
<aside> 💡 Training a good model with the right dataset is just as important as finding the right prompt in the studio
</aside>
The base model is the AI tech provider and version you wish to use for the training. We currently support 2 versions of stable diffusion, and in the future we’ll add other available image and text models.
This version of the model is currently the best to have results on :
Currently, we use this version for an overwhelming majority of applications.
This version has brought improvements on the following topics :
<aside> 💡 Overall, is is harder to get good results on version 2.1, partly become a lot more work is required entering antiprompts. In other words, you guide more by telling it what not to display rather than what it should display. You can get better results on specific generations, but it will take more experimentation.
</aside>