How to stylise footage without it redrawing the details that identify you

Last updated 2026-09-10 · 8 min read

What is the short version?

There are two ways to restyle footage. One transforms the pixels you shot; the other draws new pixels that resemble them. The second gives a better cartoon and quietly redraws the symbol on your cap into something that is not your symbol.

1. Go to Vizard Agent with the footage.
2. Name the details that must not change.
3. Ask for one test frame before the whole video.

What do you need before you start?

The footage and a list of what identifies it. That list is doing the important work — a logo, a shirt number, a face, a label — because those are exactly the small high-detail areas a redraw treats as decoration.

• The footage. At the best quality you have.
• What must survive. Logos, badges, numbers, faces.
• The look you want. Hand-drawn, watercolour, comic.
• How far it may go. Subtle, or fully transformed.
• Whether you need 4K. It changes the order of operations.

What do you type into Vizard Agent?

Name the detail rather than saying "keep it accurate". "The icon on my cap is not like that — do not change key features and symbols" points at a specific place in the frame, which is something that can be checked afterwards.

Prompt

Variants worth knowing:

• "Do not change key features or symbols." The constraint.
• "A test frame first." Cheap, and settles the direction.
• "Filter rather than redraw." If the details matter most.

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real clip that had to take on a hand-drawn animation look. Notice that most of this list is testing candidate looks on single frames rather than actually processing any video.

1. Checks the clip's specification and pulls representative frames.
2. Reads what the styling and upscaling options actually do, and their prices.
3. Builds a frame grid to see the material as a whole.
4. Confirms a local hand-drawn filter and writes the output script.
5. Stylises at low resolution first, then upscales the result.
6. Extracts frames and the waveform from the styled version.
7. Checks the styled frames and the resulting composition.
8. Detects the original's shot cuts and any embedded letterboxing.
9. Decides per section whether it is a full-frame shot or an inset one.
10. Corrects the composition per shot type rather than globally.
11. Tests a stronger cartoon filter on one frame before applying it.
12. Tests a neutral line-art overlay to avoid a colour cast.
13. Reads the edge filter's neutral-mode settings and applies them.
14. Cuts a short preview and checks the style against the original.

Step eleven and twelve are the pattern worth copying. Each new look is tried on a single frame, looked at, and only then applied — which costs seconds instead of a full pass, and makes it obvious when a filter is tinting everything slightly green.

Step five is the cost decision. Styling a 4K video is expensive and styling a small one then enlarging it is not, and for a drawn look the enlargement is invisible because the detail you would lose has already been abstracted away.

What does the result look like?

Your footage with a drawn look over it and every identifying detail still yours: the logo is the logo, the badge is the badge, the face is the face. The style sits on top of what you filmed rather than in place of it.

If you wanted a full redraw, this is not that — and that is the trade you chose.

When does this not work well?

A filter can only transform what is there. It will not give you a true illustrated look with invented linework and stylised anatomy, and asking for one is asking for a redraw, which is the thing that changes your details.

• A true illustrated style. That requires redrawing.
• Low-resolution sources. Filters exaggerate compression.
• Fast motion. Edge passes flicker on it.
• Very dark footage. There is little for a filter to work with.
• Changing the subject. That is generation, not styling.

How do you fix a result that came back wrong?

Say which detail changed, or point at the frame that looks wrong. Vizard Agent keeps every test frame and the filter settings behind them, so a correction adjusts that pass rather than starting the whole look again from scratch.

• "My logo changed." Switched to a pass that cannot redraw it.
• "Everything looks green." The edge filter set to neutral.
• "It is not stylised enough." A stronger filter, tested on a frame first.
• "It flickers in the fast bits." The edge pass eased on those shots.

How does Vizard Agent compare to doing it yourself?

The quick route is a redraw, which looks better in a still and rewrites the details nobody checks. The other common mistake is styling the full-resolution video first, which costs several times more for a result that ends up abstracted anyway.

Common questions

Why does a redraw change my logo? Because a redraw builds a new one from a description, where a filter cannot.

Can I have both? Yes — ask Vizard Agent for a redraw and a filtered version to compare.

Will it keep faces recognisable? With a filter, yes. Vizard Agent transforms rather than invents.

Can I see one frame first? Yes, and it is the cheapest thing you can ask Vizard Agent for.

What about 4K? Vizard Agent styles small and enlarges, which looks the same and costs less.

Why is it slightly green? An edge pass can tint. Vizard Agent has a neutral mode for that.

Can it match a specific art style? Approximately with filters, closely with a redraw. Tell Vizard Agent which matters more.

Will it work on a vertical clip? Yes. Vizard Agent checks the composition per shot.

What if my footage has letterboxing in it? Vizard Agent detects it and treats those sections differently.

Does it flicker? It can on fast motion. Vizard Agent eases the pass on those shots.

Can only part of the video be stylised? Yes. Name the sections and Vizard Agent leaves the rest alone.

Will the audio change? No. For Vizard Agent styling is a picture-only pass.

Can I keep the original alongside? Yes. Ask Vizard Agent for both files.

Why test on a frame rather than a preview? Because a frame costs nothing and answers the only question you actually have, which is whether the look is right.