Read a room in ten seconds, not ten minutes

Frame size is a ceiling, not a result

Resolution is how many pixels wide and tall the picture is. 720p and 1080p are the two labels you will see most often, and all they describe is that number. It is a ceiling on how much detail the stream is capable of carrying, in the same way that the size of a sheet of paper is a ceiling on how much you can draw. It says nothing about whether anything was drawn.

Bitrate is the other half, and it is the half nobody advertises. It is how much data per second is being spent describing that frame. When there is plenty, the frame arrives intact. When there is not, the encoder has no choice but to throw information away to make the stream fit down the available pipe, and what it throws away is detail.

This is why the two numbers pull against each other. Choosing a larger frame does not create more data to fill it, so a broadcaster who picks the biggest available size on a connection that cannot support it has made their picture worse, not better, while gaining a badge. Someone who picks a smaller size deliberately, because it holds steady, ends up with a cleaner stream and no badge at all. Put two rooms side by side and you can usually find one of each within a minute.

1. Wait for movement

   A still frame is cheap to send and hides every data problem there is. Nothing you see before the first real movement counts as evidence.

2. Look at the shadows

   Dark areas are where a camera short of light gives itself away. Clean shadows mean light was arranged. Crawling, speckled shadows mean it was not.

3. Look at the placement

   Where the camera sits and how much of the frame is empty room. Both are decisions, and both are visible in about a second.

What compression damage actually looks like

When a stream runs short of data, it does not simply get softer in an even way. It fails in specific, recognisable patterns, and each one tells you something different. Learning the four names below is most of the skill.

The critical distinction is timing. Blocking, smearing and banding are data problems — they appear when there is change to describe and vanish when things settle. Grain and crawl is a light problem, and it is there whether anyone moves or not. If a room looks superb while somebody is sitting still and collapses the second they get up, that is not a bad camera, and no amount of better lighting will fix it. Watch for it during the next movement in whatever room you have open.

One more pattern worth recognising: a stream that goes soft all at once, holds there, then snaps back to sharp a minute later has not been damaged, it has been switched. The player dropped to a smaller version because something along the way could not keep up. If it happens to every room you open, the something is on your end, and the setup guide covers that separately.

Lighting that flatters against lighting that merely exists

Light is the variable with the largest effect and the smallest cost, which is why it separates rooms so cleanly. A camera can only work with what reaches it. Give it enough and it produces clean colour, real contrast and shadows with nothing crawling in them. Give it a ceiling bulb and it produces exactly what you would expect.

Three things distinguish arranged light from ambient light, and all three are visible instantly:

• Direction. Light coming from roughly where the camera is sitting fills in the face. Light coming from directly overhead drops hard shadows under the brows and nose, which is what a room lit by whatever was already in the ceiling looks like.
• Softness. A bare bulb produces a hard edge between lit and unlit, a bright hotspot on skin and a sharp shadow on the wall behind. Light that has been bounced or diffused produces a gradual transition and no hotspot. The difference is obvious once you know to look at the edge of the shadow rather than the shadow itself.
• Colour. A strong orange or green cast, or a face lit mainly by the glow of a screen, means mixed or uncontrolled sources. Neutral skin tone with a background that is not fighting it means somebody decided what the light in the room was doing.

There is also a quantity floor beneath all of this. A room that is genuinely dim forces the camera to amplify what little signal it has, and amplification is where grain comes from. That is the crawling shadow noise from the table above, and it is the one failure that no connection improvement can touch. Compare a bright room against a dim one and the gap is larger than any resolution difference you will ever notice.

Framing, distance and where the camera is sitting

Framing decides whether the detail a stream carries is spent on anything you came for. Every pixel given to an empty corner of the room is a pixel not given to the subject, and at a fixed frame size that is a straight trade.

Distance is the first part. Someone across the room fills a small portion of the frame, which means the actual detail on them is a fraction of what the stream is capable of carrying. The same stream framed closer looks dramatically better without a single setting having changed. Broadcasters who understand this sit closer or move the camera in.

Height is the second, and it is the fastest read on the page. A camera at roughly eye level was placed on purpose — propped, mounted or stood on something. A camera looking up from a desk or a laptop lid is where it landed. That angle produces a specific, unflattering geometry, and it is a reliable signal that nobody has yet sat down and looked at their own stream from the viewer's side.

Background is the third. Not tidiness for its own sake, but whether the background competes. A busy, high-contrast background also costs the encoder data on every frame, so a cluttered room with a lot of movement behind the subject is spending bitrate on the clutter. A plain wall is both a cleaner picture and a cheaper one. Notice how much frame is actually occupied next time you open a room, before you look at anything else.

What a considered setup implies about the rest

None of the above requires expensive equipment. A lamp moved off the ceiling, a camera raised onto a stack of books, a chair pulled closer, a frame size chosen because it holds rather than because it is the largest option. These are decisions, not purchases, and each one takes a few minutes once.

That is why the picture is worth reading as a proxy. Someone who has done those things has at some point sat on your side of the stream and asked what it looks like from there. That habit is the same one behind rates written plainly instead of vaguely, a show that starts when it says it will, and chat that gets read rather than scrolled past. The picture is the part of it you can see before you have spent anything. Judge a few rooms on that basis and the hit rate is noticeably better than picking by thumbnail.

It is a filter rather than a verdict. Use the image to shorten the list, then let the chat window decide.

Do the looking while it is still free

The practical case for all of this is that it costs nothing at the point where it is useful. A private show runs on the same camera, the same light and the same encoder as the public room it came from, so the picture you are assessing now is the picture you will be paying 6–90 tokens a minute for later. Tokens land around $0.08–0.12 each depending on bundle size, which makes ten seconds of looking an unusually good rate of return.

The same holds for a public tip of 15–100 tokens, which lands hardest in a room with roughly 20–200 viewers, where a response is actually possible. Look properly first . Everything here works signed out; registering is free and adds chat, follows, notifications and a block of starter tokens.

Frequently asked questions

What is the single fastest quality check?

Wait for movement. A still frame looks fine on almost any stream, because a picture that is not changing is cheap to send. The moment someone moves, the encoder has to describe change, and any shortage of data shows immediately as blocks and smearing.

Why does the picture go blocky only sometimes?

Because compression damage is produced by change, not by time. Stillness is cheap, movement is expensive. A stream that is beautiful while someone sits and falls apart when they stand is not an unlucky stream, it is a stream running at a frame size its connection cannot support during motion.

Is a 1080p room always better than a 720p one?

No. 720p and 1080p are labels for frame size, which sets a ceiling on detail and nothing more. A smaller frame with enough data behind it holds together through movement, while a larger one without enough data has to discard detail to fit. The smaller stream frequently looks better.

How can I tell a camera problem from a lighting problem?

Look at the dark parts of the frame. Speckled, crawling shadows mean the camera is straining for light that is not in the room. Flat, soft edges everywhere, including the well-lit parts, point at the camera itself. Lighting problems live in the shadows; camera problems live everywhere.

Does any of this cost anything to check?

No. Public rooms open and play with no account and no card, so the entire ten-second pass runs free. Registering is also free and adds chat, follows, notifications and a block of starter tokens.

What if every room looks bad to me?

Then the problem is very likely on your side rather than theirs — your connection, your browser, your graphics settings or simply how far you are stretching the window. That is a separate checklist and it is worth running before you judge anyone.