Thanks for releasing the excellent work and the training/testing code!
We evaluated the DCVC-UF models. The RD performance of HT-L/S models aligns well with the results reported in the paper. For the LD model, we still observe a gap of approximately 0.5%–1.0% BD-rate compared with the reported result.
Could you please clarify the exact evaluation settings (such as command parameter, platform) used for the LD results in the paper? This information would help a lot.
Thanks for releasing the excellent work and the training/testing code!
We evaluated the DCVC-UF models. The RD performance of HT-L/S models aligns well with the results reported in the paper. For the LD model, we still observe a gap of approximately 0.5%–1.0% BD-rate compared with the reported result.
Could you please clarify the exact evaluation settings (such as command parameter, platform) used for the LD results in the paper? This information would help a lot.