3.1.0rc1

We are happy to provide the first release candidate for GraXpert 3.1

Changes and improvements

• Added an ai model for object deconvolution. That is, deconvolving structures in nebulae or galaxies. Stars should not be affected by this model.
• GraXpert now links to onnxruntime 1.20. For Linux, this implies a dependency to CUDA 12 and cudnn 9.

Known issues

• Regression: opening .png files currently fails in GraXpert 3.1.0rc1
• We have received reports that onnxruntime might cause issues on certain Windows 10 systems ("ImportError: DLL load failed while importing onnxruntime_pybind11_state"). As a workaround you can disable Hardware Acceleration in the Advanced tab on the right.
• The macOS arm64 version may require a macOS version >= 14
• We cannot provide an Ubuntu .deb package at the moment due to issues with a build tool dependency

Support

If you have any questions or want to chat with other GraXpert users, you can find us in the Dark Matters Discord.

3.0.2

Changes and improvements

• Added cancel button for denoising
• Hardware Acceleration can now be turned off to avoid driver issues
• Prepared denoising for AI model 2.0.0

Videos

Frank aka astrophotocologne has created two introductory videos on GraXpert 3.0:

• English video
• German video

Known issues

• We have received reports that onnxruntime might cause issues on certain Windows 10 systems ("ImportError: DLL load failed while importing onnxruntime_pybind11_state"). As a workaround you can disable Hardware Acceleration in the Advanced tab on the right.
• The macOS arm64 version may require a macOS version >= 13.6
• We cannot provide an Ubuntu .deb package at the moment due to issues with a build tool dependency

Support

If you have any questions or want to chat with other GraXpert users, you can find us in the Dark Matters Discord.

3.0.1

Changes and improvements

• set onnxruntime execution_mode to sequential
• bugfix for and improve ai model handling
• set usable batch_size range to 1, 2, 4, 8, 16, 32

Videos

Frank aka astrophotocologne has created two introductory videos on GraXpert 3.0:

• English video
• German video

Known issues

• We have received reports that onnxruntime might cause issues on certain Windows 10 systems ("ImportError: DLL load failed while importing onnxruntime_pybind11_state")
• The macOS arm64 version may require a macOS version >= 13.6
• We cannot provide an Ubuntu .deb package at the moment due to issues with a build tool dependency

Support

If you have any questions or want to chat with other GraXpert users, you can find us in the Dark Matters Discord.

3.0.0

New Features

• Added an AI model for denoising linear images
• Redesigned the user interface for incorporating the denoising model
• Extended GraXpert's cli interface with a denoising command. Run graxpert -h from a terminal to view the cli documentation
• For gpu acceleration during denoising, please note: on Windows, we rely on DirectML which requires a DirectX 12 compatible graphics card. On Linux, we rely on CUDA 12 and libcudnn8. On macOS, we rely on CoreML
• Repackaged AI models in .onnx format. This simplifies to provide GPU acceleration on different hardware platforms for us. Thanks a lot to Riccardo for his pull requests!

Videos

Frank aka astrophotocologne has created two introductory videos on GraXpert 3.0:

• English video
• German video

Known issues

• We have received reports that onnxruntime might cause issues on certain Windows 10 systems ("ImportError: DLL load failed while importing onnxruntime_pybind11_state")
• The macOS arm64 version may require a macOS version >= 13.6
• We cannot provide an Ubuntu .deb package at the moment due to issues with a build tool dependency

Support

If you have any questions or want to chat with other GraXpert users, you can find us in the Dark Matters Discord.

3.0.0b1

Original and Background image do not use Gradient-Corrected image as …

3.0.0b0

Stretch options are now shown in the lower status bar

3.0.0.dev4

Umbriel

3.0.0.dev3

make cli backwards compatible; improve denoising cli

3.0.0.dev2

Fixed error in stretching after gradient correction and denoising

3.0.0.dev1

New Features

• Added an AI model for denoising linear images
• Repackaged AI models in .onnx format. This simplifies it to provide GPU acceleration on different hardware platforms for us. Thanks a lot to Riccardo for his pull requests!

Limitations

• GPU acceleration on Linux is available via CUDA 12.4 only for now. Please contact us via Discord if you want to get GPU acceleration for your configuration on this architecture. You would have to run GraXpert from sources though.

Known issues

• The macOS arm64 version may require a macOS version > 13.6
• We cannot provide an Ubuntu .deb package at the moment due to issues with our underlying build technology

Support

If you have any questions or want to chat with other GraXpert users, you can find us in the Dark Matters Discord.