How to make an inserted photo look like someone placed it there by hand
Tell Vizard Agent the picture should look as though a hand brought it into shot. It builds an entry from the corner with the hesitation of a real movement rather than a constant slide, sizes the image as a stated fraction of the frame, takes it out again at the moment you name, and checks frames afterwards to confirm it has actually gone.
What do you type into Vizard Agent?
Describe the movement as a person, not as an animation. "Slide the photo in" gets you a slide, while saying somebody is holding it into shot changes the motion, the timing and the angle together. Give two numbers as well: how much of the frame it fills and when it goes. Vizard Agent works to those rather than to a feel.
Prompt
What does Vizard Agent actually do?
An image that slides in at a steady speed and parks in the corner announces itself as an overlay. The same image brought in from the bottom left, held, and taken away again reads as something a person showed the camera. Almost every pass came back to the entry, the size and the exit. In the session behind this article it:
- Reviewed a clip where a document had to appear in shot, to see where the action and the edges were, and opened the image to decide how to bring it in.
- Composited it with a hand-placed entry rather than a slide, then extracted frames to check the entry and the settled position.
- Rebuilt it at the stated size when it landed too small.
- Timed the exit and verified on frames that nothing survived past the named moment, because a leftover of two frames survives a watch-through and shows up on a phone.
- Cropped part of the image when it turned out too tall, rather than shrinking the whole thing, and re-checked the entry and exit timing.
When does this not work well?
This is a look, not a simulation. There is no real hand and no real shadow, so on a shot with strong directional light the missing shadow gives it away, and a static camera hides the trick better than a moving one. Very busy frames leave nowhere for it to sit, and the longer it stays, the more it reads as a layer.
Common questions
How big should it be? Big enough to read. Tell Vizard Agent a fraction of the frame.
What if the photo is the wrong shape? Vizard Agent crops rather than shrinking it.
Does it check the exit? Yes. Vizard Agent checks it on frames, which is where leftovers show.