If you consider it a "failure" when the codec resorts to fall-back mode to avoid crashing, then even the factory (un-hacked) settings fail at times. That's why fall-back mode exists. Fall-back mode temporarily uses much reduced encoding quality for several frames when it kicks in.
I'm with lpowell on this issue. I have found that not only does flowmotion improve quality - it is at least as reliable as un-hacked settings. Lee has posted exhaustive evidence that quality is much improved with flowmotion - if you haven't seen such evidence, then you haven't looked for it.
driftwood's settings encompass more of a spectrum, including some settings that sacrifice stability in favor of extreme motion rendering capabilities. He discloses which settings are stable and which aren't - just pick the ones that suit your needs. I'm a fan of many of his settings.
There are several circumstances where evaluations don't make much sense:
1 - When no post processing is done. The main value of many patches is improvement in gradability - which isn't visible until you grade something.
2 - Still frames when the setting is designed to improve motion rendering. When motion is poorly renderd the detail in a single frame may look fine, but if it isn't correct it will look bad in sequence with other frames.
3 - When video captured at, say 150Mb, is re-rendered to something like, say 16Mb (e.g. Vimeo). When final output bitrate is even lower than the camera's factory default bitrate not only would I not expect to see an improvement, I wouldn't even expect the final rendering to necessarily be as good as the camera's default (22Mb) rendering.
I'm with lpowell on this issue. I have found that not only does flowmotion improve quality - it is at least as reliable as un-hacked settings. Lee has posted exhaustive evidence that quality is much improved with flowmotion - if you haven't seen such evidence, then you haven't looked for it.
driftwood's settings encompass more of a spectrum, including some settings that sacrifice stability in favor of extreme motion rendering capabilities. He discloses which settings are stable and which aren't - just pick the ones that suit your needs. I'm a fan of many of his settings.
There are several circumstances where evaluations don't make much sense:
1 - When no post processing is done. The main value of many patches is improvement in gradability - which isn't visible until you grade something.
2 - Still frames when the setting is designed to improve motion rendering. When motion is poorly renderd the detail in a single frame may look fine, but if it isn't correct it will look bad in sequence with other frames.
3 - When video captured at, say 150Mb, is re-rendered to something like, say 16Mb (e.g. Vimeo). When final output bitrate is even lower than the camera's factory default bitrate not only would I not expect to see an improvement, I wouldn't even expect the final rendering to necessarily be as good as the camera's default (22Mb) rendering.
Last edited:
