Yuck. This is a good example of why you're supposed to reduce the dimensions of the video if you want to reduce bit rate with H.264. If you simply reduce the bit rate, you end up losing detail.
e.g. for a video with a typical level of motion, if you want to achieve 1 Mbps, encoding at 854x480 is going to look significantly cleaner & sharper than encoding at 1280x720 (even when upscaled).
This is why adaptive streaming uses quality levels with varying dimensions. e.g. on Netflix, the low bandwidth streams are not 1920x1080, as that would look like garbage.
e.g. for a video with a typical level of motion, if you want to achieve 1 Mbps, encoding at 854x480 is going to look significantly cleaner & sharper than encoding at 1280x720 (even when upscaled).
This is why adaptive streaming uses quality levels with varying dimensions. e.g. on Netflix, the low bandwidth streams are not 1920x1080, as that would look like garbage.