Hello out there!
I'm looking for guidance on how to squeeze the most quality out of lowish spec devices for a very specific application. I've already been doing a lot of testing, but need some expert advice. This is going to be a long post, so please bear with me.
I'm using OBS to record dual streams of usb3 cameras 1280x1920 @ 60fps for facial motion capture (so really 2160x1920) with audio via one of the cameras.
The goal is to record on a device small enough to be mounted somewhere on the body for mocap, but low powered as to give good v-mount battery life
The system I'm currently testing is this one https://www.ecs.com.tw/en/Product/Mini-PC/LIVA_Q3_Plus/overview which is 15w Ryzen v1605 embedded 4 core with Radeon Vega 8 graphics, 8gb RAM, running a Tiny10 windows 10 image (so none of the fluff), which according to specs, should be able to do the job. After recording, I'm splitting the streams out using Davinci Resolve. I've also tested a i7 NUC with Iris XE GPU, but although it has no issues whatsoever encoding, it really sucks v-mount batteries dry in about 1 hr, so not suitable
The application I'm using to process the video for facial mocap is Epic games Metahuman Animator, in which the quality of the video recorded directly influences the depth map generated for facial mocap processing, so file size does not matter.
In my testing, I've already discarded codecs such as x.264, which quickly overloads the encoder using CQP, so I'm currently testing AMD HW x.264 and AMD HW h.265 HEVC. Simple mode, which I've read that everyone normally recommends, also has a tendency to overload it so I'm using advanced mode to tweak the settings
Understanding that CQP is normally the best quality for recording, I've been testing this setting in both flavors of HW accelerated codecs without overloading the encoder, but I still see various degrees of skipped frames while recording in the stats, with the lowest being using h.265 HEVC. Although after processing the video in Metahuman Animator I'm getting ok results, I'd like to see if there are better settings. (One interesting finding, is that using Websockets to enable triggering of OBS recording from a remote PC really increases the # of skipped frames)
What I can't find a lot of information on, is QVBR, HQVBR and HQCBR, especially the last two as I'm assuming that by the name, they might have something to do with quality? what setting should I use with these to get the best results?
Your advice is greatly appreciated
Thanks
I'm looking for guidance on how to squeeze the most quality out of lowish spec devices for a very specific application. I've already been doing a lot of testing, but need some expert advice. This is going to be a long post, so please bear with me.
I'm using OBS to record dual streams of usb3 cameras 1280x1920 @ 60fps for facial motion capture (so really 2160x1920) with audio via one of the cameras.
The goal is to record on a device small enough to be mounted somewhere on the body for mocap, but low powered as to give good v-mount battery life
The system I'm currently testing is this one https://www.ecs.com.tw/en/Product/Mini-PC/LIVA_Q3_Plus/overview which is 15w Ryzen v1605 embedded 4 core with Radeon Vega 8 graphics, 8gb RAM, running a Tiny10 windows 10 image (so none of the fluff), which according to specs, should be able to do the job. After recording, I'm splitting the streams out using Davinci Resolve. I've also tested a i7 NUC with Iris XE GPU, but although it has no issues whatsoever encoding, it really sucks v-mount batteries dry in about 1 hr, so not suitable
The application I'm using to process the video for facial mocap is Epic games Metahuman Animator, in which the quality of the video recorded directly influences the depth map generated for facial mocap processing, so file size does not matter.
In my testing, I've already discarded codecs such as x.264, which quickly overloads the encoder using CQP, so I'm currently testing AMD HW x.264 and AMD HW h.265 HEVC. Simple mode, which I've read that everyone normally recommends, also has a tendency to overload it so I'm using advanced mode to tweak the settings
Understanding that CQP is normally the best quality for recording, I've been testing this setting in both flavors of HW accelerated codecs without overloading the encoder, but I still see various degrees of skipped frames while recording in the stats, with the lowest being using h.265 HEVC. Although after processing the video in Metahuman Animator I'm getting ok results, I'd like to see if there are better settings. (One interesting finding, is that using Websockets to enable triggering of OBS recording from a remote PC really increases the # of skipped frames)
What I can't find a lot of information on, is QVBR, HQVBR and HQCBR, especially the last two as I'm assuming that by the name, they might have something to do with quality? what setting should I use with these to get the best results?
Your advice is greatly appreciated
Thanks