This model is all Cyborg's. It runs on 1.5 512x512 but if running at 512x512 don't run it with high res fix at 512x512 or outputs look jacked. Sorry I scared everyone this is not a 2.0 or 2.1 model. It is 1.5 and runs great in automatic 1111. Outputs for me have been really nice. Token is Cyborgdiffusion. So in a prompt Cyborgdiffusion, portrait of a cyborg.
Description
FAQ
Comments (11)
Hi. Like the model, but seem to get more robot subjects as opposed to "cyborg". Any thoughts on how to influence this behavior?
No after using it a lot I thought the same thing. So I may revisit this and try more human and less robotish images to train a model. Coming back from a windows wipe and having covid so ready to make some more models.
i see an issue i see with many robot/android/cyborg models. all models tend to come out very human/bipedal. i've only recently started looking into embeddings and how to train them, but i'd like to generate robotic animals too. do you have any suggestions for me here? maybe im just prompting wrong...
the scope of these trained checkpoints is sort of limited Too many examples during training can lead to a muddled effect.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.