| license: bsd-3-clause | |
| library_name: mlx | |
| pipeline_tag: image-to-image | |
| tags: | |
| - mlx | |
| - super-resolution | |
| - real-esrgan | |
| - image-to-image | |
| - apple-silicon | |
| # Real-ESRGAN-x2plus (MLX) | |
| Apple **MLX** fp16 port of Real-ESRGAN **RealESRGAN_x2plus** (RRDBNet, ×2), | |
| for super-resolution on Apple Silicon. Converted from the official | |
| [xinntao/Real-ESRGAN](https://github.com/xinntao/Real-ESRGAN) release checkpoint | |
| (BSD-3-Clause). | |
| ## Usage | |
| ```bash | |
| pip install realesrgan-mlx # https://github.com/xocialize/realesrgan-mlx | |
| realesrgan-mlx -i input.png -o out/ -n RealESRGAN_x2plus | |
| ``` | |
| ```python | |
| from realesrgan_mlx.pipeline_mlx import make_upsampler, upscale_image | |
| up = make_upsampler("RealESRGAN_x2plus", tile=256) # tile>0 caps memory on large images | |
| out = upscale_image("input.png", up) | |
| ``` | |
| `make_upsampler` downloads these weights automatically. | |
| ## Details | |
| - **Architecture**: RRDBNet (`num_feat=64`, `num_block=23`, `num_grow_ch=32`) | |
| - **Scale**: ×2 · **Precision**: fp16 | |
| - **Parity vs PyTorch**: full-forward 2.4e-6 (CPU fp32); fp16 vs fp32 golden 5.5e-4. | |
| ## License | |
| BSD-3-Clause (upstream Real-ESRGAN, Xintao Wang et al.). | |