Custom autoencoder: The popular WebUI has the script to load separate first_stage_model parameters.

[cibernicola]

From trinart_characters_19.2m_stable_diffusion_v1

`Custom autoencoder
We also provide a separate checkpoint for the custom KL autoencoder. As suggested by the Latent Diffusion paper, we found that training the autoencoder and the latent diffusion model separately improves the result. Since the official stable diffusion script does not support loading the other VAE, in order to run it in your script, you'll need to override state_dict for first_stage_model.

The popular WebUI has the script to load separate first_stage_model parameters.`

https://huggingface.co/naclbit/trinart_characters_19.2m_stable_diffusion_v1

So.... any idea about how?

Found: the VAE trick, same as NAI.

[ilazarte]

@cibernicola what is the vae trick? the renaming of model to .vae.pt?

[cibernicola]

Yes, also, in the latest updates there is a VAE selector, so now it's an option :D