Vizard Agent

How to zoom into a photograph without the picture going soft

Last updated 2026-09-10 · 8 min read

Give Vizard Agent the photograph at its full size and ask for a focus still for each move. A push that is built by enlarging a screen-sized copy has no detail left to reveal, so it renders each move from a crop of the original at high resolution, builds the shot with room for the camera to travel, and checks a frame from the middle of every clip where the softness would show.

What is the short version?

A slow push looks like it is revealing detail. It only does that if the detail is still in the file — a move built on an image that has already been scaled down to fit a screen is enlarging pixels that are no longer there.

  1. Go to Vizard Agent with the original photographs.
  2. Ask for a focus still per move, at full resolution.
  3. Ask to see a mid-frame from each finished clip.

What do you need before you start?

The photographs at the size they came off the camera or the scanner. Vizard Agent can work with anything, but the amount of movement a photo will take is decided entirely by how many pixels it has, and that is a fact about the file rather than a setting.

What do you type into Vizard Agent?

Ask for the crop, not the zoom. Those sound like the same request and are not: a zoom scales what it is given, whereas a crop of the original takes real pixels and leaves them at full size.

Prompt

Build slow pushes into these photographs. For each one, render a high-resolution focus still cropped from the original rather than scaling the whole picture up, make the shots at a size the move has room to travel through, and show me a frame from the middle of each clip so I can check it is still sharp.

Variants worth knowing:

What does Vizard Agent actually do?

Here is the pattern across real photo-led films — an archive piece, a product sizzle, a family story. The work is in preparing each still rather than in the move itself, which is why the moves usually look identical in a timeline and different on screen.

  1. Downloads the photographs at full resolution.
  2. Checks each one's real dimensions.
  3. Decides where each move should end.
  4. Renders a high-resolution focus still for that move.
  5. Builds the shot large enough for the camera to travel.
  6. Sets a slow move rather than a fast one.
  7. Renders the clips.
  8. Extracts a mid-frame from every clip.
  9. Looks at those frames for softness or stretching.
  10. Re-renders any that went soft from a tighter crop.

Step four is the difference between a push that reveals and a push that blurs. Cropping the original to the region the move ends on keeps those pixels at their native size, so the end of the move is as sharp as the start.

Step eight is the check that costs nothing and catches everything. The start and end frames of a move often look fine while the middle is where an image gets stretched or resampled, so Vizard Agent looks there specifically rather than at the boundaries.

Step six matters more than it sounds. A fast move hides softness because the eye cannot settle, but it also hides the photograph, and the reason to move slowly is that a slow move is the one that has to survive being looked at.

What does the result look like?

Photographs that hold up as shots — sharp at the end of the move as well as the beginning, moving slowly enough to be read, and going somewhere rather than drifting. On a large screen the difference from a scaled push is immediate.

Vizard Agent tells you which photos did not have the pixels for the move you wanted, which is more useful than quietly delivering a soft one.

When does this not work well?

Some photographs simply do not have the resolution. A small scan, a screenshot or a social-media download has a hard limit on how far you can push before there is nothing left, and no amount of care in the rendering changes that.

How do you fix a result that came back wrong?

Say which photo and where it goes soft. Because Vizard Agent built each move from its own focus still, a note about one shot is a re-crop of that one rather than a change to the whole sequence.

How does Vizard Agent compare to doing it yourself?

By hand a photo goes on the timeline and a zoom gets applied to it, working from whatever copy was to hand — often the one that was emailed round. The softness is noticed on a big screen, after everything is cut.

By hand Vizard Agent
The source Whatever copy is handy The full-resolution original
The move A zoom on the whole image A crop rendered at full size
Render size The output size Large enough to travel through
Checking Start and end frames A frame from the middle
When softness is found On the finished film On the individual clip

Common questions

Why does my zoom look soft? Because it enlarges a copy that no longer holds the detail. Vizard Agent crops the original instead.

How big does the photo need to be? Bigger than the output by roughly the amount you intend to push. Vizard Agent works that out.

Can Vizard Agent tell me if it is too small? Yes, before rendering, which is when it is useful.

Does a faster move hide it? It does, but Vizard Agent will point out that it hides the photo too.

What about scans? Scan larger than you think. Vizard Agent will say what it can take.

Can it upscale instead? Sometimes, though Vizard Agent will note that a real crop always beats an invented one.

Where should the move end? Somewhere specific. Tell Vizard Agent what the photo is about.

Will it keep the aspect ratio? Yes. Vizard Agent watches for stretching, which is the other way this goes wrong.

How slow is slow? Slow enough to be read. Vizard Agent sizes it to the shot length.

Can it do a pan rather than a push? Yes, and Vizard Agent will note a pan usually needs even more width.

Does it check every clip? Yes. Vizard Agent pulls one mid-frame from each, which is quick.

What if only some photos are large? Vizard Agent gives the small ones shorter or gentler moves.

Does this apply to generated images? Yes. Tell Vizard Agent to generate at the size the move needs.

Why check the middle and not the end? Because resampling shows up mid-move, where the eye has time to settle on it.