* skip impersonated VAAPI implementaions ("imposter detection") * place VAAPI priority below CUDA/NVDEC/CUVID