How to get the most out of a small credit balance
Tell Vizard Agent what you have to spend and ask for fewer outputs from footage you already own. Most of the cost in a session is rounds of correction rather than the first render, so the real saving comes from checking the pick and the framing before anything is built, and from re-rendering only the clip that changed.
What is the short version?
A small balance does not mean a worse video, it means fewer of them and a tighter brief. The waste is almost never in the render — it is in building three things, finding they are wrong, and building them again. Vizard Agent can do the checking first.
- Tell Vizard Agent the balance and how many outputs you want.
- Work from your own footage and skip anything generated.
- Ask it to verify the pick and the framing before it renders.
What do you need before you start?
Your footage and a decision about quantity. Three clips from one video is a normal ask; ten is the same work several times over, and if the balance is small the honest version of the brief is fewer outputs done properly rather than more done once each.
Also know which parts you do not need. Generated footage, extra voices and stock material all add cost that an edit of your own recording does not, and often the video does not want them anyway.
What do you type into Vizard Agent?
Put the constraint in the brief rather than hoping it is inferred. A balance is a planning input: it changes how many candidates get built, whether anything is generated, and how much verification happens before the first render rather than after it.
I have a small balance, so plan for it. From this one video, give me three vertical shorts of 30 to 40 seconds using only my own footage — nothing generated. Show me the candidate moments and the crop before you render anything, and if something needs fixing, re-render only that clip.
The last clause is the one people leave out. Without it a correction to one clip can come back as a fresh pass over all of them.
What does Vizard Agent actually do?
Vizard Agent front-loads the cheap work. Reading, measuring and looking all cost very little next to rendering, so the order of operations is what decides whether a small balance produces three finished clips or one finished clip and two abandoned attempts.
In the session behind this article it:
- Built a board of frames sampled across the video to see the content without touching the source file.
- Used the transcript, the shot map and word-level timings to find moments that stand alone — the strongest claims rather than arbitrary windows.
- Previewed the three candidate windows before reframing any of them vertically.
- Tested the crop on each candidate to confirm the subject stayed in the portrait frame, before committing to a render.
- Corrected the clip boundaries so each opening hook was not clipped out of its own subtitles.
- Reviewed each finished short from a frame sheet and an audio waveform rather than by watching all three.
- Re-rendered only the one short whose headline was too long, leaving the other two alone.
That last line is the habit. A fix to one output is not a reason to rebuild the set, and saying so keeps a correction cheap.
What does the result look like?
The same finished clips you would get otherwise, arrived at with less building and more checking along the way. The difference is invisible in the output itself and very visible in how much was spent getting there, which is the only part a small balance cares about.
You also get the verification artefacts along the way — the frame board, the crop tests, the review sheets. Those are what let you say "not that moment, the one before it" before anything expensive has happened.
When does this not work well?
When the job genuinely needs generated material. A video that requires shots you never filmed cannot be made cheaply out of footage that does not contain them, and the honest answer is a different video rather than a worse version of this one.
It also fails if you change your mind late. Every round of correction costs, and the brief that arrives half-formed costs more than the brief that took five extra minutes to write — which is the uncomfortable part of working to a small balance.
And a very long source adds unavoidable analysis. Vizard Agent has to read an hour of material before it can find the best forty seconds in it, and that part does not compress.
How do you fix a result that came back wrong?
Name the single output and the single problem. "Clip two's headline is too long" gets Vizard Agent to do one re-render; "the batch is not right" gets a pass over everything, and the difference is entirely in how you phrased it rather than in the work needed.
If the pick is wrong rather than the build, say so before asking for a change. Vizard Agent can offer other candidate moments from the material it has already read, which costs far less than building a replacement and reviewing it.
If you are running low mid-session, say that too. It will tell you what is left to do and what it would drop, rather than continuing until the balance decides for you.
How does Vizard Agent compare to doing it yourself?
An editor working to a tight quote does exactly this: agrees the selects before cutting, and the cut before finishing. The order exists because changing your mind gets more expensive at every stage, and that is as true here as on a timeline.
The habit worth copying is reviewing from a frame sheet. Checking three clips as contact sheets and waveforms takes a fraction of the time of watching them, and it catches the framing and audio problems that watching tends to excuse.
Common questions
Should I tell it my balance? Yes. Vizard Agent plans differently when it knows the constraint.
What costs the most? Rounds of correction, not the first render Vizard Agent produces.
How many clips from one video? Three is a reasonable ask. Tell Vizard Agent the number.
Is generated footage more expensive than my own? Yes. Tell Vizard Agent "only my own footage" if you do not need it.
Can it check the pick before building? Yes. Vizard Agent shows candidate moments first if you ask.
Can it test the crop first? Yes. Vizard Agent tests it on a few frames before any render.
Will it re-render everything for one fix? Not if you ask for one clip. Say which output changed.
Does a long source cost more to analyse? Yes, and Vizard Agent cannot skip that part.
Can I work at low resolution first? Yes. Vizard Agent previews cheaply and masters once.
What if I run out mid-job? Tell Vizard Agent early and it will say what it would drop.
Is reviewing from frames really enough? For framing and audio, usually. Vizard Agent builds the sheets.
Does a tighter brief actually save anything? Yes, more than any other single thing.
Can it tell me what a plan will involve? Ask Vizard Agent for the steps before it starts.
Should I ask for fewer or shorter clips? Fewer. Clip length is not where the cost sits.