How to tell whether a faster render preset actually cost you quality

Last updated 2026-09-15 · 8 min read

What is the short version?

A faster encoder preset saves hours and may cost nothing visible, or may smear every leaf in a field. You cannot tell by watching, because the picture on your screen is scaled and your memory of the source is not reliable.

1. Ask Vizard Agent to compare the two presets on the hardest shot.
2. Ask for a 1:1 crop, source beside render.
3. Ask whether the slowness was the encoder at all.

What do you need before you start?

The job, and a sense of where it is hard. Vizard Agent will pick a dense shot if you do not name one, but you usually know which sequence has the rain, the crowd or the foliage in it, and that is the shot the decision rests on.

• The footage. Especially the difficult shots.
• The deadline. It sets how much speed is worth.
• The delivery target. Platform, size, resolution.
• Where quality matters most. A hero shot, a product.
• What you have tried. Presets already ruled out.

What do you type into Vizard Agent?

Ask Vizard Agent for the comparison rather than for a verdict. "Is the quality acceptable" is a judgement you should be making on evidence of your own, and the evidence here is two crops placed side by side rather than a reassuring sentence about it.

Prompt

Variants worth knowing:

• "A 1:1 crop, side by side." The only useful comparison.
• "On the densest footage." Where a preset fails first.
• "Rather than the disk or the CPU." Often the real cause.

What does Vizard Agent actually do?

Here is the Vizard Agent sequence on a real hour-long render that was running far too slowly. Most of it is diagnosis rather than encoding: the preset change comes very late in the list, after every other explanation for the slowness has been ruled out first.

1. Pauses the render to benchmark encoder settings properly.
2. Benchmarks those settings on the dense field footage specifically.
3. Tests the storage speed and looks for local scratch space.
4. Compares a 1:1 crop of the render against the source.
5. Views the two crops side by side, source left, render right.
6. Makes a tighter crop where the first one was not decisive.
7. Isolates whether storage or the processor is the bottleneck.
8. Restarts with thread limits to stop the encoders competing.
9. Checks the actual processor quota the machine has.
10. Compares the two presets' quality directly before committing.

Step two is why a benchmark is worth pausing for. An encoder preset that looks identical on a talking head can visibly smear a field of grass, so benchmarking the easy shot tells you nothing about the shot that will fail.

Step seven is the step that saves the most time. Several encoders running at once on a machine with fewer processors than they assume will run slowly at any preset, and changing the preset in that situation costs quality and buys nothing.

Step four is the measurement the whole decision rests on. A 1:1 crop shows one render pixel per screen pixel, which is the only way to see what the encoder actually did rather than what your screen did to it afterwards.

What does the result look like?

A decision you can defend. Two crops side by side, the speed each preset achieved, and a clear statement of whether the bottleneck was the encoder at all — and Vizard Agent tells you what it changed rather than just delivering faster.

When does this not work well?

A crop comparison only tells you about the shot it came from. Compression artefacts are content-dependent, so a clean result on one sequence is never a guarantee for another, and grain, rain and confetti each fail in a different way that Vizard Agent has to test for separately.

• Content that varies widely. One crop is not the whole film.
• Very short clips. The encoder never reaches steady state.
• Platform re-encoding. Your master is not what viewers see.
• Grain you want to keep. Encoders treat it as noise.
• A machine under other load. The benchmark is not repeatable.

How do you fix a result that came back wrong?

Say what you can see in the crop. Vizard Agent has both images, so "the grass has gone mushy on the right" points at a specific failure and leads to a different preset rather than a general retry.

• "The detail is gone in the render." A slower preset is tested.
• "They look identical to me." Take the faster one; that is the answer.
• "It is still slow." The bottleneck is re-diagnosed rather than the preset changed.
• "Try a different shot." The crop is taken from that sequence instead.

How does Vizard Agent compare to doing it yourself?

By hand you switch preset, watch the result, decide it looks fine, and ship it. The loss shows up on somebody else's large screen, on the one shot with foliage in it, after the file has gone out.

Common questions

Why is watching it back not enough? Because your screen scales the picture. Vizard Agent compares at actual pixels.

What is a 1:1 crop? One render pixel per screen pixel. Vizard Agent takes it from the same region of both files.

Which shot should it test? The hardest. Vizard Agent picks a dense one if you do not name it.

What if both look the same? Then take the faster preset. Vizard Agent treats that as a real result, not a failure.

Could the slowness be something else? Often. Vizard Agent checks storage and processor load before blaming the encoder.

What are thread limits for? To stop several encoders competing for the same processors and all running slowly.

Does a bigger file always look better? No. Vizard Agent judges on the crop, not on the size.

What about grain? Encoders smear it. Say if you want it preserved and Vizard Agent tests for that.

Will the platform re-encode it anyway? Usually. Vizard Agent can test what survives that too.

Can it test more than two presets? Yes, though two well-chosen ones usually settle it.

Does pausing the render waste time? Less than finishing a twelve-hour render at the wrong setting.

Can I see the timings as well? Yes. Vizard Agent reports the speed each preset achieved.

What if I need a specific file size? That is a different constraint. Vizard Agent can work to a size target instead.

Should I always use the slowest preset? No. If the crops match, the slower one is buying you nothing.