RealESRGANx4_fp16
RealESRGANx4_fp16 is an ONNX export of a Real-ESRGAN-based super-resolution model for 4× image upscaling.
Model Description
This repository provides a float16 ONNX version of a Real-ESRGAN super-resolution model designed to enhance image detail and visual quality across a variety of content.
This variant follows the standard Real-ESRGAN approach and may produce sharper, more stylized results compared to more balanced variants.
File
RealESRGANx4_fp16.onnx
Intended Use
This model is intended for:
- 4× image upscaling
- video frame enhancement
- detail enhancement in images and video pipelines
- ONNX Runtime based applications
Input / Output
- Input: low-resolution image tensor
- Output: 4× upscaled image
Exact preprocessing and normalization depend on the host application.
Usage Notes
- Emphasizes sharper and more detailed output
- FP16 export for improved performance on supported hardware
- Compatible with ONNX Runtime (CUDA, TensorRT, DirectML, CPU)
- May produce more aggressive detail enhancement than general-purpose models
Source / Attribution
This repository contains an ONNX-packaged version of a Real-ESRGAN-based model.
The upstream Real-ESRGAN project is licensed under the BSD 3-Clause License.
Limitations
- GAN-based models may introduce artificial or exaggerated detail
- May oversharpen certain textures or edges
- Output quality depends on input resolution and preprocessing
License
BSD 3-Clause