Our latest, most potent SD1.5 model is now live!
You can use it for free directly on Distillery in Discord chat. (link)
https://discord.gg/distillery-1105589864453394472
About the model
You can learn every detail about the model on our Substack (link). There, you can find the datasets used, training methodologies, the WIP versions of the model, and the thinking behind it - everything is open-sourced that led to the model.
The model performs the best when generated at around 768x resolution.
The model is versatile, and you can prompt it in almost any style. Whether it is MidJourney style prompts or anything from Civitai or Lexica, you should expect some interesting results in most cases.
Description
FAQ
Comments (3)
What's the point in creating a checkpoint then obfuscating the prompts on EVERY "sample" image you post with it? Shouldn't we be able to check and see if we can recreate what was supposedly created with this Checkpoint?
hey, you are absolutely right and that was my oversight actually, sorry for that. It was not intentional at all...
I used a generator for these images that seems to be not compatible with Civitai and the metadata was not automatically displayed.
I added all the fields that I could manually - prompt, negative prompt, sampler, steps, etc.
However, please also note that all of these images went through sd upscale pass. Sorry for that - hope you will revisit the model and give it a try
If you want you can check also my image, the model looks like Deliberate and worth a try IMO.