DevicedialWhat does your stream have to send, and can your upload carry it?
The settings are chosen one at a time and paid for all at once, by the same upload.
One gear check a month
MonthlyOne thing a month worth measuring before it fails properly โ a mic that reset itself, a stick that drifts, a switch sending two clicks. Two sentences and the test.
Free to use โ Pro is for keeping it
This tool, like every other one here, is free and needs no account. A free account keeps three setups and three email alerts. Pro takes the ads off every site we run and lifts that to 200 setups and 50 alerts, adds a public page for any result and lets you embed the tool without our badge.
Try also
How much upload does a stream or a call actually need?
The Devicedial stream check sheet turns each source you are sending into megabits a second and adds them into the figure your upload has to carry. A row is a resolution, a frame rate and a bits per pixel setting, which is the honest way to compare a camera against a screen share: multiply the pixels by the frames and by how many bits each pixel is worth.
The two example rows, a 1080p60 camera at 0.07 and a 1080p30 screen share at 0.05, come to 11.8 Mbps. Stereo audio adds 0.16, and twenty percent headroom for scene changes takes it to 14.3 Mbps, or 12.9 gigabytes across a two-hour stream. Headroom is the part people leave out: an encoder aiming at an average will exceed it every time the picture changes, and the upload it exceeds it into is shared with everything else in the building.
What each resolution and frame rate costs, before anyone argues about codecs
Pixels a second is width times height times frame rate. The three columns are the same pixels at three different bits per pixel, which is the only knob that separates a soft picture from a sharp one at a given size.
| Source | Pixels a second | At 0.05 bpp | At 0.07 bpp | At 0.10 bpp |
|---|---|---|---|---|
| 2560 x 1440 at 60 | 221.2 million | 11.1 Mbps | 15.5 Mbps | 22.1 Mbps |
| 1920 x 1080 at 60 | 124.4 million | 6.2 Mbps | 8.7 Mbps | 12.4 Mbps |
| 1920 x 1080 at 30 | 62.2 million | 3.1 Mbps | 4.4 Mbps | 6.2 Mbps |
| 1280 x 720 at 60 | 55.3 million | 2.8 Mbps | 3.9 Mbps | 5.5 Mbps |
| 1280 x 720 at 30 | 27.6 million | 1.4 Mbps | 1.9 Mbps | 2.8 Mbps |
| 640 x 480 at 30 | 9.2 million | 0.5 Mbps | 0.6 Mbps | 0.9 Mbps |
Doubling the frame rate costs exactly what doubling the pixel count costs โ the arithmetic does not care which one you change. That is why 1080p30 and 720p60 land within a few percent of each other.
The checks worth doing before the number matters
Two of the three inputs in this sheet are things you should measure rather than assume. The camera test reports the resolution and frame rate your webcam is genuinely sending, which in a dim room is often half of what it claims, and there is no point budgeting for 1080p60 when the camera has quietly settled on 720p15. The mic test settles the audio side: the bitrate is fixed, but a level of โ30 dBFS is not fixed by any amount of bandwidth.
And the two nobody checks until it is live
Whether they can hear the room, and whether you look like you are lagging. The echo test plays five seconds of you back so you hear the room the way the audience does, and the webcam mirror is for the framing and the light before you go on rather than during. If the complaint is that you feel behind, that is a different measurement entirely โ the input lag budget is where the milliseconds between your hand and the screen get added up, and none of them is bandwidth.
For choosing which microphone goes into the stream in the first place, the headset and mic comparison ranks them by how little of the room they bring with them.
Frequently asked questions
What bits per pixel should I put in?
Between 0.05 and 0.10 for modern encoders, and the number depends on the picture rather than the hardware. A largely static screen share, a slide deck or a coding session compresses far better than a camera, so 0.04 to 0.05 is realistic there. A handheld camera, a busy game or anything with grass, water, confetti or smoke in it wants 0.08 to 0.10. If a platform has already given you a bitrate to aim at, work backwards instead: divide it by the pixels a second in the table and you have the figure that platform is assuming.
My upload is faster than this number and the stream still stutters. Why?
Because a speed test measures a good moment on an empty line, and a stream needs a mediocre moment on a shared one. Two things usually explain it: everything else in the building is uploading at the same time, and the encoder’s peaks run well above its average whenever the scene changes. That is what the headroom line is for. The other common cause is not bandwidth at all but the machine โ if the encoder cannot keep up, frames are dropped before they ever reach the network.
Does a higher frame rate cost as much as a higher resolution?
Exactly as much, which surprises people. The arithmetic is pixels times frames, so going from 30 to 60 frames a second doubles the bill in the same way that going from 720p to 1080p roughly doubles it. That is why 1080p30 and 720p60 sit within a few percent of each other in the table. Which one to spend on depends on the content: motion needs frames, text and detail need pixels, and a screen share full of small type is the one case where resolution wins outright.
How the test works
Sources: