If you're like me, then you are the type of person who loves to record their gameplay not matter how insignificant it is. OBS allowed us all to take advantage of the power of our computers to save precious, potentially embarrassing moments with our friends online. However, using suboptimal settings can lead to headaches down the line, which this guide will hopefully alleviate.

This guide is not perfect. All the settings along with my substantiation as to why I have chosen them may not be the best for you. My settings have been chosen based on the following goals:

1. Balance between both efficiencies in file size and video quality
2. Maintain future compatibility and using best practices
3. Ensuring stability and a solid viewing experience

For context, my following settings have been optimised for the following desired output:

• A 1440p source/canvas resolution downscaling to an output of 1080p
• Recording at 60 fps
• Using the NVENC hardware accelerated encoder

With that out of the way, let us begin!

Output (advanced mode)

This section will cover the most important part of my settings. These options will drastically influence both file size, stability and compatibility so pay attention to my reasoning to determine if these options will suit your needs.

Recording

• Type: Standard

Recording Settings

Recording Format: Matroska Video (.mkv)

• This video container is not only the most versatile, supporting many audio and video codecs, it also supports multi-track audio and can be remuxed within OBS itself into an mp4.
• It also is protected against sudden crashes, meaning the recording will not be lost

Video Encoder: NVIDIA NVENC HEVC

• HEVC is the most space efficient codec which can be hardware accelerated
• This is compatible with most computers/systems, however requires either using HEVC Video Extensions from the Microsoft Store for Windows or VLC
• It will provide the best quality at the same bitrate compared to avc1/h264
• If you are concerned about compatibility, use NVENC H.264
• Using a CPU encoder will result in higher quality videos, but major performance impacts on your system (just don't).

Audio Encoder: AAC (if you want maximum compatibility such as with video editing)

Update: 27/5/2023 - Added AAC back for compatiblity.

Update: 8/8/2023 - Removed Opus recommendation for maximum compatibility.

• It is recommended to use the CoreAudio implementation. You can find the steps to use this in the post: "Boost your OBS AAC Audio Quality"
• The FFmpeg AAC encoder performs worse compared to Apple's own CoreAudio implementation.
• If you are going to use AAC in OBS which you are required to for streaming, ensure you use at least an audio bitrate of 192kbps, preferably 256kbps if using FFmpeg.

Encoder Settings

Rate Control: CQP

• Under NO CIRCUMSTANCE should this be set to anything else other than CQP.
• CQP is the most efficient option as it will maintain a constant quality level with a varying bitrate. This is excellent for recording as it allows the encoder to save similar frames by dynamically adjusting to the current scene using only the required bitrate, leading to reduced file sizes.
• Using CBR or Lossless will result in VERY LARGE FILE SIZES WITH NO BENEFIT.

CQ Level: 22/23

• Level 22 seems to be a good balance between quality and file size from my testing
• A lower number will have a higher quality target, but increase the file size exponentially
• A range of 16-30 is recommended if you are to change this setting

Keyframe Interval: 0s

• Simple, its automatic

Preset: P6: Slower (Better Quality)

• Nvidia recommends this setting for most graphics cards
• If you are running multiple encodes (such as streaming/recording at the same time) or simply want stability, using P5 is also a good option (the default option)
• Test this before choosing one as it can vary depending on encoder utilisation

Tuning: High Quality

• HQ is best for everyone
• The low latency options will increase encoder usage and probably larger file sizes.

Profile: main

• main is for SDR content
• main10 is for HDR content
• Change depending on what you are recording

Look-ahead: off

Psycho Visual Tuning: on

Max B-frames: 2

Audio

Audio Bitrate: 160-256

• If you are just recording with the Opus codec, 160 is more than enough
• 128kbps is generally perceived to be transparent in Opus
• If you are streaming as well, use 256 as AAC will sound better (if using FFmpeg), otherwise 192 is enough for CoreAudio.
• Anything higher is only necessary if you are focusing on audio quality, rather than video

Streaming

Streaming Settings

Audio Encoder: FFmpeg AAC

• The only option

Video Encoder: NVIDIA NVENC HEVC

• You don't need to care about compatibility if you are streaming
• However, if you are using Twitch you will NEED TO USE H.264

Encoder Settings

Rate Control: CBR

• Recommended by all streaming platforms

Bitrate: 9000kbps (varies)

• For a 1080p 60fps streaming, this is pretty good
• You can increase to 10000kbps but it isn't too important
• Dependant on what resolution and framerate you are streaming at

Keyframe Interval: 0s

• Its automatic

Preset: P5: Slow (Good Quality)

• Streams have shit quality anyways, P6 isn't going to make a big difference
• This is more stable than P6
• Recommended by major streaming platforms

Tuning: High Quality

Multipass Mode: Two Passes (Quarter Resolution)

Profile: main

• main for SDR content
• main10 for HDR content

Look-ahead: off

Psycho Visual Tuning: on

Max B-frames: 2

Audio

Sample Rate: 48 kHz

Channels: Stereo

• 48 kHz is best for audio with video, which in this case it is!
• Channel is obviously dependant if you want to record surround sound

Video

This part made a major difference in the perceived quality of my videos.

Base (Canvas) Resolution: your monitor's native resolution

Output (Scaled) Resolution: the output resolution you want

Downscale Filter (IF DOWNSCALING, THIS IS VERY IMPORTANT)

Bilinear (Fastest, but blurry if scaling)

Many people on YouTube will recommend one option or the other based on how the filter technically works, but I would like to give my reason why I chose Bilinear.

• Bilinear scaling provides the most even distribution of scaling to a video.
• All the other options are best operated when UPSCALING, however we are NOT upscaling, we are DOWNSCALING.
• From my understanding, using anything other than bilinear when downscaling will result in the filter taking FURTHER ESTIMATES ON MORE PIXELS, which is completely unnecessary, especially if the resulting image is of a higher quality already.
• Downscaling will ALWAYS RESULT IN A LOSS OF DETAIL, IT IS NOT POSSIBLE TO FIT A 1440P PICTURE IMAGE ON A 1080P SCREEN WITH ALL PIXELS ON SCREEN WITHOUT VISIBLY LOOSING DETAIL! IT IS PHYSICALLY IMPOSSIBLE!
• Other options make text and UI elements look inconsistent, over sharpened and unnatural.
• Bilinear provides the best performance
• If you don't believe me, try record with the other options. In my opinion, they're an eyesore as I notice the illegible typefaces on the screen varying in thickness and weight.

Conclusion

Those are my OBS options which I have picked. I have battle tested these options in a live environment and overall the experience has been positive. File sizes are much more manageable and I am happy with the resulting picture quality and better audio.