How to ask for a trend you cannot describe
Give Vizard Agent the trend's name and, if you have one, a link to an example. It goes and looks at the post, samples it to work out what the format actually does, and measures the thing worth copying — the beat lengths and the points where it cuts — then rebuilds that structure using your photos rather than anyone else's footage.
What is the short version?
You have seen a format go past a hundred times and you cannot describe it. You do not need to. Vizard Agent studies an example and extracts the pattern, and the pattern is what you were asking for.
- Tell Vizard Agent the trend's name, with a link if you have one.
- Let it study an example before it makes anything.
- Give it your own photos or footage to put through the structure.
What do you need before you start?
An example, or a name specific enough that Vizard Agent can find one. A link is considerably better than a name, because a name will get you somebody's approximation of the format, while a link gets you the format itself with its real timings intact.
Your own material is the other half. These formats are structures — a rhythm, a number of beats, a swap at a particular moment — and the structure is only worth having if there is something of yours to run through it.
What do you type into Vizard Agent?
Ask Vizard Agent whether it knows the format, and say what material you have. Treating it as a question rather than as a brief is the right instinct here, because the first thing that has to happen is identification rather than production.
Do you know the hotel lobby trend? Here are two photos of me.
That is close to the original request, and it worked because the name was findable. If yours is not, paste a link to one post and the identification step is solved outright.
What does Vizard Agent actually do?
It watches an example first and measures it rather than describing it. Sampling the reference at several points shows what the edit actually does — how long each beat runs, where it cuts, which character is on screen when — and those numbers are what get carried across.
In that session the sequence ran:
- Opened the referenced post and pulled its public preview to confirm the format.
- Sampled the video at several moments and looked at them together to establish the edit pattern.
- Extracted the exact points where the performance hands over from one character to the next.
- Prepared the beats as separate timed segments rather than one run.
- Converted both of the user's photos onto a single matching studio background so two separate images read as one set.
- Animated each photo to perform its own beat, then assembled them to the measured timings.
The handover points are the part that makes it recognisable. A format like this is identified by when it switches, and getting the switch a third of a second late makes it read as an imitation rather than as the thing.
What does the result look like?
Your material in that format, at the format's own rhythm. The total length matches, the cuts land where they land in the original, and the two photos look like they were shot in the same place because the backgrounds were unified rather than left as they came.
Expect some unglamorous fixes. In that session a microphone kept crossing the close-up face, so the shot was re-framed and then the fill colour matched so a padded medium shot blended into the set; and the segment durations were rebuilt to exact lengths because the first assembly came out long.
When does this not work well?
When the trend depends on material you do not have rights to. The reference's own footage and audio belong to whoever made them — the structure is the transferable part, and a version that ships someone else's clip is a different thing entirely. Vizard Agent works from your material for that reason.
It also struggles when the name is vague or the trend has drifted. Formats mutate weekly, and "the one with the zoom" will get you somebody's variant. A link removes the ambiguity.
And a format built on one specific piece of music has a licensing question attached to it that no amount of editing can answer. That is worth settling before the video exists rather than after it is already posted, and Vizard Agent will raise it rather than quietly using the track.
How do you fix a result that came back wrong?
If the result does not feel like the trend, say which beat is off rather than that it feels wrong. It is usually a swap point rather than the look, and Vizard Agent re-measures that one moment in the reference instead of restyling your whole version.
If a generated face does not look like you, ask for the original photo back in that beat. Late in that session the animated face was replaced with the exact original photograph while the audio timing was preserved — the real face was better than the animated one.
If the total length drifted, say so. Vizard Agent rebuilds the segments to exact durations rather than trimming the end, which is what keeps the internal timings intact.
How does Vizard Agent compare to doing it yourself?
Copying a format by eye works and it is imprecise in exactly the way that matters. You will get the look and miss the timing, and the timing is what people recognise — which is why so many trend attempts are identifiably attempts.
Measuring the reference is the difference. It is an unromantic step, and it turns "I want it to feel like that" into a list of durations and cut points you can build against and check afterwards.
Common questions
Do I need a link or is the name enough? A link is better. Vizard Agent will search for one if you only have the name.
Will it use the original's footage? No. Vizard Agent takes the structure and uses your material.
What about the original's music? That is a licensing question. Vizard Agent will flag it rather than assuming.
How many photos do I need? As many as the format has characters. Two were enough for Vizard Agent here.
Can it make a photo speak? Yes. Vizard Agent animates each photo to its own beat.
Why unify the backgrounds? So separate photos read as one set. Vizard Agent matched both to a single studio look.
What if the trend has changed? Send a recent example and Vizard Agent measures whatever you give it.
Can it tell me what the format does? Yes. Ask and it will describe the structure it measured.
How exact are the timings? To a fraction of a second. Vizard Agent rebuilds segments to exact durations.
What if a prop crosses my face? Say so. Vizard Agent re-frames and matches the fill so the join is invisible.
Can I use video instead of photos? Yes, and Vizard Agent finds it easier. The photo route exists for when you have no footage.
Will it look generated? Less so if you ask Vizard Agent for your real photo in the close-ups.
Can it do several trends from the same photos? Yes. Vizard Agent can reuse the same photos across formats.
How long does identification take? A couple of minutes with a link, longer from a name alone.