For a 24/7 YouTube gaming VOD stream, choose FFmpeg hardware encoding settings only after confirming that your installed build and driver can use the encoder you intend to rely on. Then match the output codec, resolution, frame rate, bitrate and keyframe interval to YouTube’s current ingest guidance, and keep a separate local recording as your preservation copy.
There is no single command or bitrate that suits every GPU, FFmpeg build and gameplay source. Treat YouTube’s live page as the distribution endpoint, test the whole recording and streaming workflow with representative gameplay, and do not assume a live broadcast will become a complete archive.
Start with your hardware and FFmpeg build
Write down the details that determine whether a hardware encoder is usable: operating system, GPU or integrated graphics, driver, FFmpeg version and build, and the codec and output mode you want. Resolution, frame rate, audio format, upload capacity and the local-recording plan matter too. A setting copied from another creator is not evidence that your machine exposes the same encoder or options.
FFmpeg’s -hwaccels output can tell you which hardware acceleration components were enabled when that build was compiled. It does not prove that a compatible device and driver are available at runtime. FFmpeg’s hardware acceleration documentation makes that distinction: actual availability depends on suitable hardware and a suitable driver.
Also distinguish input decoding from output encoding. In FFmpeg’s command-line documentation, -hwaccel concerns hardware acceleration for decoding input; adding it alone does not select a hardware video encoder. The encoder is chosen by the output codec implementation, and its own available options depend on the build and runtime. Confirm that the intended encoder appears in your installation’s encoder list and inspect its help output before building a long-running workflow around it.
A practical check is to make a short test encode on the actual machine, using the same resolution, frame rate, codec and audio arrangement you intend to stream. Confirm that the encoder initialises, the output plays, and the system can run the encoder and local recorder at the same time. If the test fails, the failure is useful information: resolve the missing runtime, driver, or unsupported option before you configure an overnight broadcast.
Choose an encoder your setup actually supports
Start with the codecs YouTube accepts for live video—H.264, HEVC or AV1—and narrow the choice to what your installed hardware encoder and YouTube ingest path support. The availability of one codec does not imply that another is exposed by the same device, driver and FFmpeg build. Check the exact encoder’s help rather than assuming options carry over between vendors or generations.
Hardware encoding can reduce the work done by the CPU, but the result depends on the machine and workload. You still need to test game capture or source playback, video encoding, audio handling, network sending and local recording together. A machine that encodes a short static clip may behave differently with moving gameplay and a continuous recording running beside the live output.
Do not select a preset solely because its name sounds faster or higher quality. FFmpeg’s QSV documentation, for example, describes implementation-specific presets, a low_power option and a scenario hint that includes livestreaming; it also notes version and mode limitations for low-power operation. AMD’s AMF AV1 documentation exposes its own rate-control and latency options. These are not interchangeable controls, and they do not establish a universal quality or performance winner across NVIDIA, Intel, AMD or software encoding.
Compare alternatives against your actual constraints: encoder availability, supported codec, target resolution and frame rate, the controls exposed by that implementation, and whether the machine can sustain both streaming and local recording. If your current encoder cannot produce a supported combination reliably, test a different supported encoder or software encoding rather than concealing the mismatch with copied flags. For a wider look at trade-offs in a modest setup, see budget-friendly streaming choices that survive a full-night test.
Match YouTube ingest settings to the output
YouTube’s live encoder settings and bitrates are the target for your outgoing stream. The page recommends RTMPS, supports H.264, HEVC and AV1 video, and specifies progressive video up to 60 frames per second. It also recommends constant bitrate (CBR) and a two-second keyframe interval, with an interval no longer than four seconds. Check the current official page before going live, because its complete codec and format guidance is more useful than a generic FFmpeg example.
Choose one output combination first: codec, resolution, frame rate and SDR or HDR. For ordinary SDR gaming output, YouTube’s guidance specifies Rec. 709 and 8-bit video. It lists square pixels, progressive scan, two B-frames, one reference frame and CABAC among the advanced settings. Those details are targets to compare with the controls your chosen encoder actually exposes, not proof that every hardware encoder can be configured identically.
Audio needs its own plan. YouTube lists stereo AAC at 128 Kbps and 44.1 kHz; for 5.1 it lists AAC at 384 Kbps and 48 kHz. Use the arrangement your source and audience need, then verify that audio stays in sync and remains present in the stream test. A sound check should include game audio, commentary if present, and any music you are entitled to use; video encoding alone does not validate the full output.
For HDR, YouTube recommends HEVC and says AV1 is not supported for HDR. Do not select an HDR workflow merely because an encoder offers an HDR-related option: check the platform requirements, source colour format and hardware encoder support as a set. If your material is SDR, keeping the path in SDR avoids an unnecessary format conversion and makes the output easier to validate.
Select bitrate for resolution and frame rate
Use the row for your selected codec, resolution and frame rate in YouTube’s table. The recommendations below are platform ingest targets, not comparative encoder tests or promises about image quality. You may need to adjust the output to fit your upload connection and the consistency of that connection, but changing codec or frame rate means you should recheck the corresponding row.
| Output | H.264 recommended bitrate | AV1 or HEVC recommended bitrate |
|---|---|---|
| 2160p (4K), 60 fps | 50 Mbps | 35 Mbps |
| 2160p (4K), 30 fps | 42 Mbps | 30 Mbps |
| 1440p, 60 fps | 34 Mbps | 24 Mbps |
| 1440p, 30 fps | 21 Mbps | 15 Mbps |
| 1080p, 60 fps | 17 Mbps | 12 Mbps |
| 1080p, 30 fps | 14 Mbps | 10 Mbps |
| 720p, 60 fps | 8 Mbps | 6 Mbps |
| 720p, 30 fps | 8 Mbps | 6 Mbps |
For example, a 1080p60 H.264 stream has a different recommended target from a 1080p60 HEVC or AV1 stream. A 1440p60 output has a different target again. Do not take the H.264 number from one row and apply it to another codec or frame rate by habit; use the current official table for the combination you will actually send.
A connection that can reach the target briefly may not sustain it through a 24/7 run. Leave network headroom for ordinary variation and other household or business traffic, and test at the intended output rate. If the connection cannot sustain the chosen target, consider a lower resolution or frame rate and use that combination’s listed recommendation. Troubleshooting a stream that buffers or drops frames is easier when the cause is isolated; the YouTube buffering checklist can help separate connection symptoms from encoding problems.
Set CBR and the keyframe interval
For YouTube live ingest, configure the output for CBR where the chosen hardware encoder supports it, and aim for a two-second keyframe interval. YouTube says the interval should not exceed four seconds. These values are ingest guidance, not a guarantee that an encoder remains active or that the stream will stay connected. Check the exact option names and units for your encoder before applying them.
Frame rate affects how a two-second interval translates into frames. At 30 frames per second, two seconds spans 60 frames; at 60 frames per second, it spans 120. Use the relevant interval control and verify its behaviour in the output rather than assuming every implementation interprets a copied value in the same way. Some encoders expose keyframe controls differently, and settings can vary by codec and FFmpeg version.
CBR describes how the encoder manages output rate over time; it does not remove the need for a suitable bitrate target or a stable upload path. If the encoder’s rate-control menu uses different names, consult that implementation’s documentation and installed help output. For instance, the FFmpeg QSV and AMF controls have implementation-specific options; do not treat a mode documented for one encoder as a setting that applies to another.
Before relying on a configuration, inspect a short output and the YouTube live preview or stream-health information. Confirm that the intended resolution, frame rate, codec, audio and keyframe behaviour are present. YouTube’s tips for live streaming advise testing and monitoring stream health. A clean local test file is useful, but the YouTube preview checks the actual ingest path as well.
Keep a separate local recording
A 24/7 broadcast should not be your only copy of the programme. YouTube says streams shorter than 12 hours can be automatically archived, but warns that a stream exceeding 12 hours may not be captured at all; it recommends a local archive as a backup. Read its archive livestreams guidance before deciding how to preserve a continuous VOD. A live page distributes the stream, but it is not a substitute for a recording you control.
A separate local recording also protects against a failed upload or a stream interruption. YouTube’s archive guidance notes that archive integrity should be checked, and its stream tips recommend confirming that the local recording file is growing. Check the file while the broadcast runs, then open and verify the recording after a test. Do not assume that a file exists just because the live output looked healthy.
Plan the recording independently from the ingest stream. Its bitrate may differ from the live bitrate if you configure it separately, so estimate storage from the actual local recording bitrate, overhead and intended retention time. The official guidance does not prescribe a drive type, capacity or RAID scheme. An external drive is one possible destination, but the right choice depends on how much material you retain and how you verify a second copy.
Avoid relying on one continuously growing file without testing how your recorder handles a long session, interruption or storage limit. You can consider dividing the programme into shorter recording segments or shorter live sessions if that suits your archive workflow. That may make individual files easier to inspect, but it does not guarantee that YouTube will preserve every segment. Keep and verify the local recording regardless. YouTube also notes that rewind or DVR capability may be limited or unavailable on streams longer than 12 hours, which is a separate issue from having a finished local archive.
Validate before the 24/7 run
Test the exact system you plan to leave running: source material with representative gameplay motion, selected encoder and codec, chosen output resolution and frame rate, audio path, live ingest and local recording. A moving game scene is a more useful test than a static title card for finding encoder or system pressure. Watch the YouTube preview and stream-health status, and check that the local file grows and plays correctly.
Record what you tested: FFmpeg version and build, hardware and driver, encoder name, output settings, and any warnings or dropped frames. This makes a later change easier to diagnose. If you update a driver, change the FFmpeg build, switch encoder or alter resolution, repeat the test; runtime availability and exposed options can change with those components.
Check upload stability under the conditions the channel will actually face. If the stream becomes unhealthy, change one variable at a time where possible: first confirm the connection and ingest status, then inspect encoder load, frame-rate consistency and audio. If the system cannot maintain the chosen resolution and rate while saving a recording, lower the target or revise the recording setup before leaving it unattended. Advice for keeping an FFmpeg playlist from stopping at its end covers a different failure mode, but the same principle applies: test the complete playback-to-output path, not just a command’s launch.
A 24/7 stream also needs someone to notice a problem. Define how you will check preview, stream health, storage and the local archive during the run, and decide what you will do if the encoder or connection stops. No hardware setting can guarantee uninterrupted encoding or network delivery. A measured test and a verified local copy reduce the cost of a failure; they do not eliminate it.
Before committing, compare the operating options on the pricing page. When the file and channel are ready, start free — 24-hour trial, no card.
FAQ
Which FFmpeg hardware encoding settings should I use for YouTube?
Use settings that your installed encoder supports, then match codec, resolution, frame rate and bitrate to YouTube’s current live ingest guidance. For live output, YouTube recommends CBR and a two-second keyframe interval, not longer than four seconds. There is no universal command because encoder options and runtime support differ by machine and FFmpeg build.
Does -hwaccel enable hardware encoding?
No. In FFmpeg’s command-line documentation, -hwaccel refers to hardware acceleration for input decoding; it does not by itself select an output encoder. Choose a hardware encoder implementation for the output and check that it is available and works with your driver on the actual machine.
Will YouTube archive a 24-hour livestream?
Do not rely on it as your only recording. YouTube says streams exceeding 12 hours may not be captured at all, so keep a local recording and verify that it is growing and complete. DVR rewind is separate and may also be limited or unavailable on streams longer than 12 hours.
Can I record locally at a different quality from the live stream?
You can configure local recording separately, but its bitrate and storage needs depend on the actual recording settings. Estimate capacity from that recording bitrate, overhead and how long you intend to retain files; the live ingest bitrate is not automatically the local file bitrate. Test streaming and recording concurrently before a long run.