No overclocking and BeQuiet case + air cooling
Device | Name / Description |
---|---|
OS | Windows 10 Pro (x64) |
CPU | Intel i7-10700K @ 3.8GHz |
GPU | NVIDIA GeForce GTX 1070 (ASUS ROG OC version) |
Mainboard | Asus ROG STRIX Z490-E Gaming |
PSU | BeQuiet 750W ATX |
RAM | 48GB (2x8GB @ 3200C16 + 2x16GB @ 3000C15) DDR4 @ 2133MHz |
Storage | Samsung M.2 SSD 970 EVO for C:\ and WD Black HDD for everything else |
Monitors | 2x old EIZO FlexScan @ 60Hz vertical (1200px high - 16:10 and 4:3) |
Keyboard | Full-sized Mechanical QWERTZ keyboard from ASUS ROG (GK2000 Horus RGB) |
Mouse | Mini Wireless Mouse @ 3100P Invisible Optical (powered by 2x AA Batteries) |
Speakers | 2x JBL control one speaker and an amplifier (+ a Cinch to AUX converter) |
Headphones | AKG K-371 closed ear headphones (rarely use it, but it's great) |
currently using OBS Studio version 30.1.0 (64 bit).
- Input Overlay with my overlays.
- win-capture-audio to only capture the audio I need (all other sources are deactivated in settings).
Output Mode: Advanced
Setting | Value |
---|---|
Encoder | NVIDIA NVENC H.264 ; CoreAudio AAC |
Scale output | (Bicubic) 1664x936 (for better bitrate on Twitch) |
Bitrate | CBR @ 6000Kbps |
Keyframe Interval | 2 sec. |
Preset | P5 / Slow (Good Quality) |
Tuning | Low Latency |
Multipass Mode | Single Pass |
Profile | main |
Look-ahead | OFF |
Psycho Visual Tuning | ON |
Max B-frames | 2 |
Also, see the Twitch VOD Track Guide for how to set up a separate audio track for VOD.
Setting | Value |
---|---|
Type | Standard |
Format | mkv (see why further below) |
Encoder | NVIDIA NVENC H.264 ; CoreAudio AAC |
Scale output | no scaling (1920x1080) |
auto split files | ON ; split by time ; 120 min. |
Compression | CQP @ 4 |
Keyframe Interval | 0 (auto) |
Preset | P7 / Slowest (Best Quality) |
Tuning | High Quality |
Multipass Mode | Single Pass |
Profile | high |
Look-ahead | OFF |
Psycho Visual Tuning | ON |
Max B-frames | 2 |
All tracks have bitrate set to 128; nothing else is configured.
Setting | Value |
---|---|
Sample Rate | 48 kHz |
Channles | Stereo |
Decay Rate | Fast |
Peak Meter Type | Sample Peak |
I've also deactivated all global audio devices that I don't need.
Setting | Value |
---|---|
Resolutions | 1920x1080 (16:9) |
Common FPS Values | 60 |
Setting | Value |
---|---|
Process Priority | Above Normal (and always run OBS as Administrator) |
Renderer | Direct3D 11 |
Color Format | NV12 (8-bit, 4:2:0, 2 planes) |
Color Space | sRGB ; Limited |
SDR White Level | 300 nits |
HDR Nominal Peak Level | 1000 nits |
automatic remuxing to mp4 | OFF (see below) |
automatic reconnecting | ON ; 2s ; 10x |
network optimizations | ON |
browser source hardware acceleration | ON |
Record to MKV (easy to restore if something happens), then convert to MP4 via FFmpeg as follows
# (Audio codec already is AAC, so it can be copied to save some time)
# Using NVIDIA CUDA for faster encoding (with 4 CPU threads ; more is not necessarily faster)
# Compression strength (qp) 4 & Bitrate: average 8Mbits ; min 500Kbits ; max 16Mbits ; buffer 16Mbits for nearly no quality loss
ffmpeg.exe -v level+warning -stats -threads 4 -hwaccel cuda -hwaccel_output_format cuda -i INPUT.mkv -map 0 -c copy -c:v:0 h264_nvenc -preset p7 -tune hq -profile:v:0 high -level:v:0 auto -rc vbr -b:v:0 8M -minrate:v:0 500k -maxrate:v:0 16M -bufsize:v:0 16M -multipass disabled -fps_mode passthrough -b_ref_mode:v:0 disabled -rc-lookahead:v:0 32 -qp 4 OUTPUT.mp4
# Copies every stream except the first video stream, which is compressed and converted to mp4 via NVIDIA h264_nvenc (+CUDA hardware acceleration)
Then, it can be used in any video editing software (delete the MKV version only if you tested the MP4 version for errors first).
A simpler, more general command can be found here.
One could also add -metadata comment="VIDEO DESCRIPTION" -metadata title="VIDEO TITLE"
(before the output) to insert some metadata into the MP4 file in one command.
also see my list of useful FFmpeg commands and the official FFmpeg wiki for some inspiration.