castledeng
castledeng
在观察教程的过程中发现作者使用了与SwapChain个数一致的Fence,这样的优势是在渲染过程中不必等待上一帧渲染结束就可以开始递交下一帧的数据,但以此为代价需要申请与SwapChain数量为倍数的UBO,并且在部分postprocess中无法使用sub pass的情况下需要额外申请SwapChain数量为倍数的FrameBuffer,内存需求大大扩大,感觉弊端比优势更大?想提问下有没有更优的解决方案呢?
**Describe the bug** Model skinned is no current. Problem show on model's hand and model's neck. Model skinned working current on PIE model. The problem only show on packaged project....
**Describe the bug** SkeletalMesh loaded by function LoadVRMFile, and I try to add a templete animation blueprint to set some morph targets. It work fine on PIE, but morph target...