Vizard Agent

How to get a usable logo out of a file that has other things in it

Last updated 2026-09-10 · 8 min read

Tell Vizard Agent which part of the file you actually need. A logo supplied for print is usually a lockup with several marks and generous padding baked into one image, so it locates each element, separates the one you want, crops tight to it, keys it onto transparency, and checks it against the background it will be composited over.

What is the short version?

The file called "logo" is rarely just a logo. It is a wordmark with a symbol, often a certification mark beside it, sometimes a tagline, and always more empty space than you want — and dropped onto a video it lands small, off-centre and next to something you did not ask for.

  1. Go to Vizard Agent with the file.
  2. Say which element you need.
  3. Ask it to check the result on your background.

What do you need before you start?

The best version of the file you have and a decision about which mark you are using. Vizard Agent works from what you supply, and a print-quality file separates far more cleanly than a logo pulled off a website at screen resolution.

What do you type into Vizard Agent?

Name the element for Vizard Agent rather than handing over the file. "Use this logo" passes the whole lockup along unchanged; naming the specific part you want turns the request into a separation job, which is what it genuinely is underneath.

Prompt

This file has the wordmark and a certification mark in one image with a lot of padding around them. Pull out just the wordmark, crop tight to it, put it on transparency, and show it to me on the dark background it will sit on so I can check the edges.

Variants worth knowing:

What does Vizard Agent actually do?

Here is the pattern Vizard Agent follows across real jobs that began with a supplied brand file. Almost all of it turns out to be inspection: looking closely at what is actually in the image before deciding what should be done with any of it.

  1. Opens the file and looks at what it actually contains.
  2. Locates each mark within the image.
  3. Separates the element you asked for from the others.
  4. Crops tightly to that element.
  5. Keys it onto transparency.
  6. Checks the keyed result rather than assuming it worked.
  7. Views it on the background it will be used against.
  8. Rebuilds the edge if it fringes or breaks up.
  9. Prepares any extra layers — a shadow or a keyline — separately.

Step seven is the step that catches the failures. A logo checked on white looks perfect and may have a white halo, soft grey fringing or a missing thin stroke, none of which is visible until it goes onto the dark background it was always destined for.

Step two matters because lockups are not always obvious. A certification mark, a registered symbol or a second brand can read as part of the design at a glance and become a problem when somebody notices it in the finished video.

Step nine is worth knowing about. A shadow or an outline is a separate layer rather than part of the mark, and keeping them apart means you can drop the shadow on a dark background without rebuilding the logo.

What does the result look like?

The mark you asked for, cropped to its own edges, on transparency, verified against the background it will sit on. It scales properly, it centres properly, and there is nothing else in the frame that nobody agreed to.

Vizard Agent also keeps the separated pieces, so the next video can use a different element without going back to the original file.

When does this not work well?

Separation needs resolution, and Vizard Agent will tell you when there is not enough. A logo lifted from a website header is a few hundred pixels wide with its background already flattened into it, and no amount of care will recover a clean edge from a file like that.

How do you fix a result that came back wrong?

Say what you see and on what background. Because Vizard Agent checked the key on a specific background, "there is a halo on black" is a precise report, whereas "the logo looks bad" could be the crop, the key or the scale.

How does Vizard Agent compare to doing it yourself?

By hand the supplied file goes straight onto the timeline, gets scaled until it looks about right, and the padding around it quietly becomes part of your layout. The certification mark comes along for the ride until somebody asks about it.

By hand Vizard Agent
The file Used as supplied Inspected, then separated
Padding Becomes part of the layout Cropped away
Extra marks Included by accident Separated deliberately
The key Checked on white Checked on the real background
Next time Same job again The separated pieces are kept

Common questions

Why is my logo small on screen? Because the file's padding counts as part of the image. Vizard Agent crops it away.

Can it separate a certification mark? Yes. Vizard Agent locates each mark and splits them.

What if I need both? Say so and Vizard Agent keeps the lockup intact.

Will it work from a website logo? Sometimes. Vizard Agent will tell you when the resolution is too low.

Does it check the transparency? Yes. Vizard Agent checks it on a real background rather than trusting the file.

What about a white halo? That is a key problem; Vizard Agent tightens it and re-checks.

Can it give me a light and a dark version? Yes, and it will verify each on its own background.

What about the shadow? Vizard Agent keeps it as a separate layer you can drop.

Will it distort if I scale it? Not if the aspect is kept, which Vizard Agent holds.

Can it vectorise? Vizard Agent can rebuild edges, but a real vector file is always better.

Does it keep the brand colours? Yes, and Vizard Agent can sample them exactly from the file.

What if the marks overlap? Vizard Agent will say when a clean separation is not possible.

Can I reuse the result? Yes. Vizard Agent keeps the pieces, which is the point of separating them once.

Why does it look fine in the file? Because the file is shown on white, and your video is not.