inferred on the cloud or local?

[cwy0110]

The inference is completed on the cloud server or local smart phone? It is import to me.

[Foul-Tarnished]

you can do both

local doesn't have a lot of settings, so no LoRa, no embeddings, ...

You can't even use another checkpoint, unless you edit a hardcoded link in docs/models.json (a link to download a converted model from .safetensors to .onnx to .ort) and rebuild the whole app with Android Studio..

You can find preconverted checkpoints here : https://huggingface.co/Androidonnxfork/test/tree/main/fp16fullonnxsdquantized_in_ort

Easier to use this instead
https://github.com/ZTMIDGO/Android-Stable-diffusion-ONNX with a checkpoint from link above

[ShiftHackZ]

The inference is done on server side for A1111 and Horde AI configurations. For the Local Diffusion configuration everything is done on local android device (you can even use Local Diffusion in airplane mode if you pre-downloaded local checkpoint model in configuration settings).