LocalMuseAI/coreml-dreamshaper-8-inpaint-fp16

DreamShaper 8 Inpaint — Core ML FP16 for LocalMuse

This repository contains an unquantized FP16 Core ML conversion of DreamShaper 8 Inpainting by Lykon for the dedicated Inpaint workflow in the LocalMuse iOS app.

Provenance

• Source repository: Lykon/dreamshaper-8-inpainting
• Pinned source revision: a09e9d0bb9849aaf46c8fc59dfa604796da15a39
• Architecture: Stable Diffusion 1.5 Inpaint
• UNet input/output: [2, 9, 64, 64] → [2, 4, 64, 64]
• Fixed image resolution: 512×512
• Storage precision: FP16
• Quantization/palettization: none
• UNet attention graph: SPLIT_EINSUM_V2
• Core ML deployment target: iOS 17
• Core ML execution policy in LocalMuse: CPU + GPU
• Conversion tooling: Apple ml-stable-diffusion commit e12202c1f6405b83918b58a5d097cd61e3e1f702, Core ML Tools 8.3.0

The UNet is split into two compiled graphs so every weight file remains below 1 GB. The package includes the matching CLIP text encoder, VAE decoder, VAE encoder and tokenizer. PROVENANCE.json records the exact authenticated source files, conversion dependencies and SHA-256 of every compiled runtime artifact.

This package is only compatible with a true nine-channel Inpaint runtime. It must not replace a normal four-channel SD 1.5 model.

Validation

The pinned PyTorch source was compared component-by-component with the Core ML conversion before publication:

• UNet parity: 74.2 dB PSNR
• CLIP text encoder parity: 82.4 dB PSNR
• VAE decoder parity: 61.3 dB PSNR
• VAE encoder parity: 80.7 dB PSNR
• Compiled UNet input: [2, 9, 64, 64]
• Compiled UNet output: [2, 4, 64, 64]

Recommended settings

• Scheduler: DPM-Solver++
• Steps: 25
• Guidance scale: 7.0
• Resolution: 512×512
• Batch size: 1 on iOS
• Device memory class: 8 GB minimum in LocalMuse

License and attribution

DreamShaper 8 Inpainting is authored by Lykon and distributed under the CreativeML Open RAIL-M license. The files here are modified by conversion to compiled Core ML FP16 format and by splitting the UNet graph. No weights were retrained or quantized, and no additional restrictions are imposed. The original license and its use-based restrictions continue to apply.

This repository does not imply endorsement by Lykon, Stability AI, CompVis, Runway, Apple or Hugging Face.