prism-em.com

What It Is

Prismatic is an image and diffraction simulation software package for scanning transmission electron microscopy (STEM) and high-resolution transmission electron microscopy (HRTEM). It implements quantum mechanical image simulation algorithms in CUDA/C++/Python, aiming to address the performance bottlenecks in STEM scanning, where multislice calculations are required at every probe position and the number of pixels can be extremely large.

Core Capabilities

At its core, it implements two types of algorithms: Multislice and PRISM, with parallelization across multi-core CPUs and CUDA GPUs, as well as streaming data implementations. The text states that in some scenarios, it can achieve over 1000x acceleration compared with traditional methods. The toolset is fairly complete: it offers a command-line interface, a graphical interface called prismatic-gui, and the Python package PyPrismatic, which allows users to call the underlying C++/CUDA code from Python. Output uses the HDF5/EMD file format and is directly compatible with py4DSTEM. It can generate default 3D output, with options for 2D, 4D, DPC Center of Mass, and projected potential slices.

Pricing and Open Source

Prismatic is explicitly free and open-sourced, making it suitable for reproduction, review, and further development within the research community. The text does not mention a commercial version, subscription, cloud service, or paid support, so it appears more like an academic open-source project than a commercial SaaS product.

Pros and Cons

Its strengths are its highly targeted algorithms, deep performance optimization, and multiple access points via GUI, CLI, and Python. The documentation covers input and output formats, compilation, packaging, tutorials, and source code walkthroughs, allowing research developers to trace implementation details. The downsides are also clear: the project states that it will no longer be actively maintained starting January 2026, so future troubleshooting and updates may be limited. It also describes itself as beta software, meaning users need to validate convergence and accuracy themselves. The GPU version depends on NVIDIA CUDA, and the project involves dependencies such as CMake, Boost, FFTW, HDF5, and Qt, making environment setup relatively demanding.

Who It’s For

It is suitable for researchers in materials science, electron microscopy, and 4D-STEM data analysis, as well as developers who are willing to compile and debug scientific research software locally. It is not suitable for teams that require commercial SLAs, ongoing maintenance, or a general-purpose development toolchain.

Access from China

The text does not provide information on accessibility from mainland China. The actual availability of the domain and external resources such as GitHub and Conda needs to be tested directly, so it is currently rated as unknown.