Explanation: The return value of ExtractData is never checked anywhere in code. However if it returns false, outData remains uninitialized. All code using ExtractData assumes outData is initialized and proceeds to using it. I haven't encountered a real-life case where this goes wrong - but the simple fact that it can go wrong is a red flag. Instead of relying on every bit of code checking the return value and handling this properly, I think it makes much more sense to have ExtractData throw an exception. It obviously is an exceptional situation, and throwing makes sure that no code that doesn't explicitly handle such a scenario continues running and potentially causing harm.
Changed copyright end year to 2020 in every reference "Copyright (c) 2006-XXXX, assimp team".
Changed copyright end year to 2020 in every reference "Copyright (c) 2006-XXXX, ASSIMP Development Team".
Changed copyright end year to 2020 in LICENCE.rtf.
Changed copyright end year in CMakeFiles.txt files and any other places referencing Assimp with a copyright start and end year.
Importing of lights according to glTF 2.0 extension KHR_lights_punctual https://github.com/KhronosGroup/glTF/tree/master/extensions/2.0/Khronos/KHR_lights_punctual
Since glTF lights are based on PBR they use different attenuation model than conventional lights supported by assimp. It is possible to use attenuation factors in assimp to describe inverse square law fallof. But the light structure does not provide means to save range property. Therefore I resorted to use of metadata. When range parameter is present, I put it into 'PBR_LightRange' metadata of light's node. Please, see comment in glTF2Importer file.
Max Vollmer
Hui.Du
Marc-Antoine Lortie
Malcolm Tyrrell
Cory Fabre
Paul Arden
Kim Kulling
petrmohelnik