search

fuxihao66 / ODI-TestProject

MIT License
2 stars 0 forks source link

Failed to compile #1

Closed ss0912 closed 3 months ago

ss0912 commented 3 months ago

Snipaste_2024-07-03_15-52-44

Snipaste_2024-07-03_15-50-02 I'm using Unreal Engine version 5.2.1 and VS version 2019. The project fails to be compiled, as shown in the following figure.

fuxihao66 commented 3 months ago

Before opening this project, you should right click on the .uproject file to "switch Unreal Engine Version" and then "Generate Visual Studio project files". Then you can open the .sln file and build the project.

ss0912 commented 3 months ago

Thank you very much for your answer. I tried to use VS2019 to generate the project, but the error as shown in the picture occurred. Do you know how to solve it? Snipaste_2024-07-04_14-27-02

fuxihao66 commented 3 months ago

If official Unreal Engine is used (downloaded from epic games launcher), please download protobuf from here, then copy BASE_DIR/protobuf-3.18.0/src/google/protobuf to UE_BASE_DIR/Engine/Source/ThirdParty/Protobuf/3.18.0/include/google.

ss0912 commented 3 months ago

Thank you very much for your answer! I am now able to run the project in Unreal Engine, but when I execute the r.RealtimeStyleTransfer.Enable 1 command, the stylized inference result is called. I would like to ask, how to use your extrapolation frame inference model?

fuxihao66 commented 3 months ago

The "ExtraNet" branch is used for extrapolation frame inference. But currently it can only be used for gpu time testing, as I don't have time to implement the prerequisite passes (buffer preparation, image warping, etc). If you want to use it for extrapolation, you need to implement these parts yourselves.

ss0912 commented 3 months ago

Are you saying that ExtraNet is not currently being used in the Unreal Engine rendering pipeline? So the "ODI" plugin you provided doesn't implement the "ExtraNet" functionality in Unreal Engine? Thanks again for your patience in answering.

fuxihao66 commented 3 months ago

Yes, it only has the "inference" step of ExtraNet Neural Network, but not the entire pipeline of frame extrapolation (including buffer preparation, image warping, etc).