In my tests, OpenFace 3.0 appears to place the chin-region landmarks consistently higher than expected, often closer to the area below the lower lip. I tested the same video frames with OpenFace 2.2 and did not observe this upward offset in the chin area. I would like to ask whether this could be caused by my deployment/runtime setup, preprocessing, or landmark visualization, or whether it may be related to the current landmark estimation behavior of OpenFace 3.0 under certain video conditions.
In my tests, OpenFace 3.0 appears to place the chin-region landmarks consistently higher than expected, often closer to the area below the lower lip. I tested the same video frames with OpenFace 2.2 and did not observe this upward offset in the chin area. I would like to ask whether this could be caused by my deployment/runtime setup, preprocessing, or landmark visualization, or whether it may be related to the current landmark estimation behavior of OpenFace 3.0 under certain video conditions.