Fidelity

This is the page that answers "why does my image look soft, blurry, or wrong". The short version: your scope is probably fine and your file is probably fine. What you are seeing is a speed limit in the audio itself.

Why corners are the hard part

Draw a square. At each corner, the beam has to stop moving in one direction and start moving in another, instantly.

An instantaneous change in a waveform is, locally, a square wave. And a perfect square wave needs an infinite series of harmonics — an endless stack of ever-higher frequencies — to make that vertical edge. Audio cannot carry infinite frequencies. Everything above half the sample rate, the Nyquist limit, simply does not exist in the file.

So the beam does the best it can with the harmonics that survive, and what you get is a rounded corner with a little ringing on either side. It is not a rendering artifact. It is what a bandwidth-limited square wave looks like, and it looks exactly like that on a real CRT too.

Every fidelity problem in this medium is the same problem: you asked the beam to move faster than the audio can describe.

The two controls that matter

1. Sample rate — raise it

Higher sample rate means a higher Nyquist limit, which means more of the harmonics that sharpen a corner actually fit in the file.

A widely repeated piece of advice among people who do this is to set your interface to 96 kHz and leave it there — it is roughly the natural refresh territory of a scope and it solves most detail problems without quadrupling your file sizes. Go to 192 kHz when the artwork actually demands it.

In OSC-1 the working rate is shown in the audio readout, and export offers the higher rates explicitly. See Export.

2. Drawing frequency — lower it

This one is less obvious and often more powerful.

Drawing frequency is how many times per second the beam completes the whole figure. Drop it and every harmonic your shape needs drops with it, pushing more of them safely under Nyquist. The beam also simply has more time to get to each point.

So a figure drawn at 50 Hz is dramatically crisper than the same figure drawn at 1000 Hz, at the same sample rate, for free.

The catch is on the other side:

• Below roughly 30 Hz the image starts to flicker — the beam is not coming back often enough to look continuous.
• Somewhere near that floor the figure stops reading as a whole object and starts reading as a dot travelling.

That gives you a working range, and it is narrower than people expect:

Between about 30 Hz and a few hundred Hz. Lower is crisper, higher is more stable. There is no setting that is good at both.

Many practitioners settle around 50–60 Hz. One well-known choice is 55 Hz, because A1 is 55 Hz — so recorded material lands in a DAW already in tune with the rest of the track.

The budget nobody tells you about

Sample rate and drawing frequency together buy you a fixed amount of points per pass — how many distinct positions the beam can visit before it has to start the figure again.

Everything competes for that budget:

• More detail in the artwork costs points.
• More objects on screen costs points, because each is drawn in its share of the pass. See Multiple objects.
• Hidden lines you have not removed cost points for lines nobody can see. See 3D.

This is why "add another object" and "sharpen the corners" pull against each other, and why the fix for a mushy image is often to remove something rather than to raise a setting.

Diagnosing a soft image

1. Check the drawing rate first. It is the cheaper fix and the bigger effect. If you are above a few hundred Hz, come down.
2. Then raise the sample rate. 48 to 96 kHz is a visible step on detailed work.
3. Then cut geometry. Turn on hidden-line removal, reduce edge count, drop decorative detail that is below the resolution you can actually achieve.
4. Only then suspect the display. Beam width, bloom and persistence soften an image too, but they soften it evenly — bandwidth limits specifically round corners and ring on sharp transitions. If your straight lines are clean and only the corners are wrong, it is bandwidth.

What this is not

• Not your oscilloscope. A real CRT has its own limits, and an old one may genuinely be unable to resolve a dense image no matter what you feed it — that is a known and well-documented frustration. But a soft corner on a clean signal is bandwidth, not the tube.
• Not the DAC. People have chased this with progressively more expensive converters and got identical results, because the limit is in the file, not the conversion.
• Not a bug in the artwork. The same file gets sharper the moment you slow the drawing rate down.

Related

• Why your oscilloscope image is blurry — the illustrated version, with the 44.1 to 192 kHz ladder on real text
• Multiple objects — the other big consumer of the same budget
• Export — choosing a rate for a finished piece
• Troubleshooting — symptoms and fixes