---
license: mit
library_name: coremltools
tags:
  - coreml
  - depth-estimation
  - stable-diffusion
  - image-to-image
  - macos
  - apple-silicon
---

# StreamRGBD-mac CoreML Models

Pre-converted CoreML `.mlpackage` models used by the [StreamRGBD-mac](https://github.com/Jup33Q/StreamRGBD-mac) project.

Mirrors: [Hugging Face](https://huggingface.co/Jup33QE/streamrgbd-mac-coreml) · [ModelScope](https://www.modelscope.cn/models/Jup33Q/streamrgbd-mac-coreml)

The repository code is released under the MIT License. Converted model artifacts retain the licenses of their respective original weights (see table below).

## Models

### Depth Estimation (Depth Anything V2 / V3)

| File | Source | License |
|------|--------|---------|
| `da2_small.mlpackage` | [depth-anything/Depth-Anything-V2-Small-hf](https://huggingface.co/depth-anything/Depth-Anything-V2-Small-hf) | Apache-2.0 |
| `da2_base.mlpackage` | [depth-anything/Depth-Anything-V2-Base-hf](https://huggingface.co/depth-anything/Depth-Anything-V2-Base-hf) | Apache-2.0 |
| `da2_large.mlpackage` | [depth-anything/Depth-Anything-V2-Large-hf](https://huggingface.co/depth-anything/Depth-Anything-V2-Large-hf) | Apache-2.0 |
| `da3_small.mlpackage` | [depth-anything/DA3-SMALL](https://huggingface.co/depth-anything/DA3-SMALL) | Apache-2.0 |
| `da3_base.mlpackage` | [depth-anything/DA3-BASE](https://huggingface.co/depth-anything/DA3-BASE) | Apache-2.0 |
| `da3_large.mlpackage` | [depth-anything/DA3-LARGE](https://huggingface.co/depth-anything/DA3-LARGE) | Apache-2.0 |

### Image Generation (UNet)

| File | Source | License |
|------|--------|---------|
| `unet_sdxs_512.mlpackage` | [IDKiro/sdxs-512-0.9](https://huggingface.co/IDKiro/sdxs-512-0.9) | Apache-2.0 |
| `unet_sdxs_768.mlpackage` | [IDKiro/sdxs-512-0.9](https://huggingface.co/IDKiro/sdxs-512-0.9) | Apache-2.0 |
| `unet_sd_1_5.mlpackage` | [runwayml/stable-diffusion-v1-5](https://huggingface.co/runwayml/stable-diffusion-v1-5) | CreativeML Open RAIL-M |

### TinyVAE (Encoder / Decoder)

| File | Source | License |
|------|--------|---------|
| `taesd_encoder_512.mlpackage` | [madebyollin/taesd](https://huggingface.co/madebyollin/taesd) | Apache-2.0 |
| `taesd_encoder_768.mlpackage` | [madebyollin/taesd](https://huggingface.co/madebyollin/taesd) | Apache-2.0 |
| `taesd_decoder.mlpackage` | [madebyollin/taesd](https://huggingface.co/madebyollin/taesd) | Apache-2.0 |
| `taesd_decoder_768.mlpackage` | [madebyollin/taesd](https://huggingface.co/madebyollin/taesd) | Apache-2.0 |

## Conversion Details

- Depth models are converted to CoreML with input resolution `504x504`.
- UNet models are converted at their respective render resolutions (`512x512` / `768x768`).
- VAE encoder/decoder models are converted at matching resolutions.

## Usage

Clone or download this repo and point your StreamRGBD-mac configuration at the `coreml_models` folder, or set `--coreml-dir` to the downloaded path.

```bash
# Hugging Face
hf download Jup33QE/streamrgbd-mac-coreml --local-dir coreml_models

# ModelScope (国内镜像)
modelscope download --model Jup33Q/streamrgbd-mac-coreml --local_dir coreml_models
```

## Disclaimer

These are converted artifacts derived from the original model weights listed above. Please refer to the original repositories for exact license terms and citation requirements.