SD 1.5 Inpainting - CoreML
CoreML-converted Stable Diffusion 1.5 Inpainting model optimized for iOS/macOS.
Model Details
- Base Model: stable-diffusion-v1-5/stable-diffusion-inpainting
- Format: CoreML (.mlmodelc compiled models)
- Resolution: 512x512
- UNet Channels: 9 (standard inpainting architecture)
- Attention: SPLIT_EINSUM_V2 (optimized for Apple Neural Engine)
Architecture
This is a true inpainting model with a 9-channel UNet:
- Channels 0-3: Noised latent
- Channel 4: Mask (1 = regenerate, 0 = keep)
- Channels 5-8: Masked image latent
Contents
| File | Description | Size |
|---|---|---|
| Unet.mlmodelc | 9-channel inpainting UNet | ~1.6GB |
| TextEncoder.mlmodelc | CLIP text encoder | ~235MB |
| VAEEncoder.mlmodelc | VAE encoder | ~65MB |
| VAEDecoder.mlmodelc | VAE decoder | ~95MB |
| vocab.json | Tokenizer vocabulary | ~1MB |
| merges.txt | BPE merges | ~516KB |
Usage
This model is designed for use with the Haplo AI iOS app for mask-based image inpainting.
License
This model is released under the CreativeML Open RAIL-M License.
Use Restrictions
You agree not to use the Model or Derivatives of the Model:
- In any way that violates any applicable national, federal, state, local or international law or regulation
- For the purpose of exploiting, harming or attempting to exploit or harm minors
- To generate or disseminate verifiably false information with the purpose of harming others
- To generate or disseminate personal identifiable information that can be used to harm an individual
- To generate or disseminate information for the purpose to be used for administration of justice, law enforcement, immigration or asylum processes
- To generate content that may be used to defame, harass or otherwise harm others
Credits
- Original model by RunwayML and Stability AI
- CoreML conversion using Apple's ml-stable-diffusion