diffusionbee-stable-diffusion-ui icon indicating copy to clipboard operation
diffusionbee-stable-diffusion-ui copied to clipboard

CoreML

Open Ka1der opened this issue 1 year ago • 4 comments

Please tell me - it is planned to support CoreML and ssplit-einsum (mlmodels)?

Ka1der avatar Dec 26 '23 21:12 Ka1der

Even if it's "planned" DiffusionBee development is slow compared to other (already CoreML supporting ) Apps like https://github.com/godly-devotion/MochiDiffusion f.e.

Schneppi avatar Jan 04 '24 07:01 Schneppi

Even if it's "planned" DiffusionBee development is slow compared to other (already CoreML supporting ) Apps like https://github.com/godly-devotion/MochiDiffusion f.e.

Thank you, but there are problems with MochiDiffusion. Poor functionality. And he doesn’t want to work with the latest models based on SDXL. Tries to load for a long time and unsuccessfully. Maybe there are other good applications that support CoreML?

Ka1der avatar Jan 13 '24 16:01 Ka1der

Latest version of diffusion bee ( 2.4.4 ) should be pretty fast. No need for CoreML

On Sat, Jan 13, 2024 at 11:12 AM ebyrotbrat @.***> wrote:

Even if it's "planned" DiffusionBee development is slow compared to other (already CoreML supporting ) Apps like https://github.com/godly-devotion/MochiDiffusion f.e.

Thank you, but there are problems with MochiDiffusion. Poor functionality. And he doesn’t want to work with the latest models based on SDXL. Tries to load for a long time and unsuccessfully. Maybe there are other good applications that support CoreML?

— Reply to this email directly, view it on GitHub https://github.com/divamgupta/diffusionbee-stable-diffusion-ui/issues/490#issuecomment-1890559085, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAONR5NSGEYWKQEE3JZPF7TYOKXAFAVCNFSM6AAAAABBDUTANWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQOJQGU2TSMBYGU . You are receiving this because you are subscribed to this thread.Message ID: <divamgupta/diffusionbee-stable-diffusion-ui/issues/490/1890559085@ github.com>

divamgupta avatar Jan 13 '24 17:01 divamgupta

It's not a question of speed. The diffusionbee works well. The point is to load the neural engine and free up memory. It's more functional

Ka1der avatar Jan 20 '24 22:01 Ka1der