I appreciate that this has been discussed multiple times on the forum and have read every single post related to the subject, but I am still a little hazy on the exact best preset to use to CQP recording.
I am recording from an Xbox Series X using a live gamer 4K in HDR.
I have been testing literally every day for the last couple of weeks as I have ran into a few issues but am currently 99% of the way there with one last issue I am trying to resolve.
My current CQP setting are below. Multiple tests on Doom Eternal with 10 minute clip times and no issues at all.
CQ Level: 12-15
Keyframe: 2
Preset: Medium (P4)
Tuning: High Quality
Multipass: Two Pass (Quarter Res)
Profile: Main 10
Look Ahead: Checked
Psycho: Checked
Max B-Frames: 4
The problem came when I did further testing using Skyrim as this is basically locked to 60FPS and open world. Doom with the settings above was clocking in at a fairly modest 150-200mb/s a second on the encode. Skyrim on the other hand is clocking in at around 400-500 mb/s and can easily go up to 800 mb/s and occasionally over 1Gb/s.
Testing 10 minute clip times would at some point during the 10 minutes cause the encoder to shutdown with the message
[jim-nvenc] get_encoded_packet: nv.nvEncLockBitstream(s, &lock) failed: 8 (NV_ENC_ERR_INVALID_PARAM):
1 encoder(s) were remaining
The only way I can consistently resolve this in Skyrim is to either use the below settings with the encoder set to fast (P3)
Encoder: Nvidia NVENC HEVC
CQ Level: 15
Keyframe: 2
Preset: Fast (P3)
Tuning: High Quality
Multipass: Two Pass (Quarter Res)
Profile: Main 10
Look Ahead: Checked
Psycho: Checked
Max B-Frames: 4
Or the below, still at Medium but single pass and no look ahead.
CQ Level: 15
Keyframe: 2
Preset: Medium (P4)
Tuning: High Quality
Multipass: Single
Profile: Main 10
Look Ahead: Unchecked
Psycho: Checked
Max B-Frames: 2
I appreciate the answer is kind of staring me in the face, but am trying to avoid having to change my settings for every game I play if I can.
The main question I would like to ask is, which settings would generally be better for image quality, regardless of file size?
Would I have a better image quality with preset to Medium (P4) with Single Pass, Psycho and no Look Ahead? Or would I be better off using Fast (P3) Two Pass Quarter or Full (haven't tested full yet) with both Look Ahead and Psycho checked? And also possibly lowering the CQ level below 15 if all is okay with the testing.
What is additionally confusing is that, checking the Nvidia migration table for Nvenc, it appears that for CQP at 4K the only recommended presets are P3 or P5 with no Multipass (single)? But I have seen posts suggesting to put Multipass to Quarter Res using CQP? And what would the recommended CQ level be with regard to the migration table CQ profiles, as in what CQ level are they relating to the Lossless Default profile and Lossless High Performance profile? Or is that just down to user trial and error?
If anyone can shed any light on this I would be hugely appreciative. I have done a ridiculous amount of testing and solved a number of issues. If I can hopefully get some clarification on the above I will create a post that will hopefully assist a number of people as I have definitely resolved a number of issues that others seem to be reporting at the moment.
I am recording from an Xbox Series X using a live gamer 4K in HDR.
I have been testing literally every day for the last couple of weeks as I have ran into a few issues but am currently 99% of the way there with one last issue I am trying to resolve.
My current CQP setting are below. Multiple tests on Doom Eternal with 10 minute clip times and no issues at all.
CQ Level: 12-15
Keyframe: 2
Preset: Medium (P4)
Tuning: High Quality
Multipass: Two Pass (Quarter Res)
Profile: Main 10
Look Ahead: Checked
Psycho: Checked
Max B-Frames: 4
The problem came when I did further testing using Skyrim as this is basically locked to 60FPS and open world. Doom with the settings above was clocking in at a fairly modest 150-200mb/s a second on the encode. Skyrim on the other hand is clocking in at around 400-500 mb/s and can easily go up to 800 mb/s and occasionally over 1Gb/s.
Testing 10 minute clip times would at some point during the 10 minutes cause the encoder to shutdown with the message
[jim-nvenc] get_encoded_packet: nv.nvEncLockBitstream(s, &lock) failed: 8 (NV_ENC_ERR_INVALID_PARAM):
1 encoder(s) were remaining
The only way I can consistently resolve this in Skyrim is to either use the below settings with the encoder set to fast (P3)
Encoder: Nvidia NVENC HEVC
CQ Level: 15
Keyframe: 2
Preset: Fast (P3)
Tuning: High Quality
Multipass: Two Pass (Quarter Res)
Profile: Main 10
Look Ahead: Checked
Psycho: Checked
Max B-Frames: 4
Or the below, still at Medium but single pass and no look ahead.
CQ Level: 15
Keyframe: 2
Preset: Medium (P4)
Tuning: High Quality
Multipass: Single
Profile: Main 10
Look Ahead: Unchecked
Psycho: Checked
Max B-Frames: 2
I appreciate the answer is kind of staring me in the face, but am trying to avoid having to change my settings for every game I play if I can.
The main question I would like to ask is, which settings would generally be better for image quality, regardless of file size?
Would I have a better image quality with preset to Medium (P4) with Single Pass, Psycho and no Look Ahead? Or would I be better off using Fast (P3) Two Pass Quarter or Full (haven't tested full yet) with both Look Ahead and Psycho checked? And also possibly lowering the CQ level below 15 if all is okay with the testing.
What is additionally confusing is that, checking the Nvidia migration table for Nvenc, it appears that for CQP at 4K the only recommended presets are P3 or P5 with no Multipass (single)? But I have seen posts suggesting to put Multipass to Quarter Res using CQP? And what would the recommended CQ level be with regard to the migration table CQ profiles, as in what CQ level are they relating to the Lossless Default profile and Lossless High Performance profile? Or is that just down to user trial and error?
If anyone can shed any light on this I would be hugely appreciative. I have done a ridiculous amount of testing and solved a number of issues. If I can hopefully get some clarification on the above I will create a post that will hopefully assist a number of people as I have definitely resolved a number of issues that others seem to be reporting at the moment.