Ranked in these QuestionsQuestion Ranking
Pro Will be very easy for developers already familiar with Python
Although it's possible to use only C++ to program in Panda3D, all its power is available to the Python scripting language, while not trading in performance since the performance-critical parts are implemented in C++.
It has a powerful binding layer that exposes the vast majority of the API via Python-based interfaces.
Pro Free, open-source, and permissive license
The liberal license allows use of the engine for any purpose without restrictions or royalties.
Pro Supports most popular physics engines
Panda3D has in-depth integration with industry standard physics engines such as Bullet, NVIDIA PhysX and ODE, but also offers a simpler built-in physics engines that will cover more basic needs.
Pro Flexible scene and object hierarchy system
Creating weird world constructs is generally a breeze. The node system the engine runs with allows to build self-looping worlds and, on large scale, non-Euclidean scenes without having to introduce a huge amount of custom code.
Pro Powerful profiling and debugging tools
Panda3D has a suite of powerful tools to help track down performance bottlenecks, memory leaks and examine internal state.
Pro Supports browser deployment
Panda3D offers web plug-ins that allow deployment of an application to all major browsers. A WebGL port is in the works as well.
Con No unified editing program
Unlike Unity and Unreal, Panda3D doesn't currently offer a single, unified editing program in which objects can simply be dragged in and assigned properties (although third-party solutions are available). Developers are expected to design their scenes in a modelling program like Maya or Blender instead, and import them into Panda3D using Python code.
Con Direct3D support is behind
Direct3D support not up to par with OpenGL support, only version 9 is supported
Con Loading Pandas3d will change your builtins to contain non explicit references to non-standard helper functions
A lot of the pollution comes from storing global state. Instead, you can store and update the global state of a namespace instead. As for the built-in pollution, you can make a wrapper that backs up builtins, imports pandas and then restores builtins, though this may not work as pandas almost certainly uses it's extra builtins to work. The best thing to do would be to explicitly import the same objects that are in the builtins over the top of the modified builtin namespace, although it doesn't remove the code smell, it helps to make things look less (if not at all) magic.