Welcome to Metaphysical
This model does just about everything, I've mixed it to be fairly photorealistic with some anime vibes. It's super versatile on its own but is also very compatible with embeddings/lora's without losing too much of it's curated look.
Please leave a review or upload your work of the model for others to see.
Description
FAQ
Comments (11)
6+gb SD1.5 model ?!?!?!?!?!
From A1111 Toolkit: Model is 6.20 GB. Multiple model types identified: SD-v1, SD-v1-ControlNet, LoRA-v1-CLIP. Model type SD-v1 will be used. Model components are: UNET-v1-SD, CLIP-v1-SD, VAE-v1-SD.
Contains 944.05 MB of junk data! Wastes 3.30 GB on precision.
I'm guessing it is FP32, not FP16, and it includes a VAE. I'm not sure how to figure out manually which FP type it is.
It did seem a bit large but i figured it was a merge with 10+ different models mixed, do you know of a way to cut it down during export for Auto1111?
@Kwheelie, In the model merge tab there should be an option to save as fp16. There is also a model converter extension that lets you convert to fp16 and prune existing models, which should get you down to 2gb. https://github.com/Akegarasu/sd-webui-model-converter
@woobly Thanks I'll be using this for v2!
@Kwheelie awaiting v2 <3
Is it trained or merged?
this is a merge, I've updated the tag, thanks for the heads up!
Looks good, some more information would be nice
Sure, what info would you like added?
Very good! Versatile use makes this one powerful.