Do I understand this aspect of Motion Estimation correctly?

[Compositor]

I was looking over some of the encoder settings of various video files I had when a thought occured to me. I have read that H254 and H265 compress video by throwing out details during movements that the eye would miss anyway. Therefore, a more accurate (and time consuming) ME setting would actually throw out more details than a less accurate ME (and faster) setting. This may be constrained by other limits (such as crf?) Is this correct?

If yes, then should I want to convert an existing H264/5 video, using a lower ME would allow me to preserve more existing details?

Which would have a greater effect on keeping details, lower ME or lower CRF?

[poisondeathray]

I was looking over some of the encoder settings of various video files I had when a thought occured to me. I have read that H254 and H265 compress video by throwing out details during movements that the eye would miss anyway. Therefore, a more accurate (and time consuming) ME setting would actually throw out more details than a less accurate ME (and faster) setting. This may be constrained by other limits (such as crf?) Is this correct?

If yes, then should I want to convert an existing H264/5 video, using a lower ME would allow me to preserve more existing details?

Which would have a greater effect on keeping details, lower ME or lower CRF?

Lower CRF will have greater effect

This will result in larger filesizes. At some point it will be larger in filesize than the original. For x264 crf 0 is lossless, for x265 crf 4 is lossless. Both will be many times larger than the original at that point (because the original is decoded to uncompressed data, before it's re-compressed)

To answer the ME part , a better ME analysis will yield higher quality at a given filesize or bitrate.