Advertisement

Exploring Impossible Shots and Previsualizations – How 3D Location Scans Help You Test Ideas

February 13th, 2026Jump to Comment Section1
Exploring Impossible Shots and Previsualizations – How 3D Location Scans Help You Test Ideas

Previsualizations… Why should we even bother with them if we don’t shoot blockbusters? Well, several reasons. They do help to save time and money in production, because everyone is on the same page and knows exactly which footage we are after. Also, they allow us to test ideas “on paper” before we invest in executing them. This becomes crucial when we are talking about impossible shots or shots that are particularly difficult to achieve. Below, we discuss one way to test those with 3D location scans and offer a simple, beginner-friendly, step-by-step guide.

I’m a huge fan of previsualizations, even if only for short commercials or music videos, and in any possible form. Simply editing still images with stock music and a temporary voice-over is already a wonderful way to see how your shots work together, and to feel the rhythm and the emotional progression. And if you have an idea for an impossible or difficult shot, just test it beforehand! At least, to ensure it even makes sense for the story!

Previs becomes infinitely more precise when you can do it directly inside a digital version of your real location. (And possibly, even with your actors staged on the correct marks?) Sounds hard? It’s not. In my MZed course “The Efficient Filmmaker: Scan and Plan”, one lesson focuses entirely on this idea: exploring shots that would be difficult, expensive, or even “impossible” to execute in reality – but in a simple, zero-cost way.

First, you need the 3D space

Before we dive into testing one of our impossible shot ideas, we will need a 3D recreation of the location. In the course, I use Luma AI’s Interactive Scenes for this task. Their user-friendly, intuitive tool lets users turn even a smartphone video into a navigable 3D scan of the real world space. Not any smartphone video, of course. There are tricks on how to capture your location to get the best possible results, and we shared them in a previous article.

Let’s imagine you have access to the location where you want to execute an especially difficult shot and need to previsualize it. Your next steps are: Create an account at Luma’s webpage (you don’t need a paid subscription), make a video recording (or scan the location directly in the dedicated smartphone app), upload it, and let their software (that uses Gaussian splats and NeRF technology) figure out the rest. After a bit of processing time, you will get a 3D scan directly in the browser, which you can now move through with your mouse. As simple as that.

Image source: MZed

In my example, we have a 3D recreation of CineD’s studio, with Nino Leitner’s digital double sitting at the desk, and with all the real-life lighting preserved.

Testing a difficult FPV drone shot

Once the scan is done and sitting in our library, the fun part of testing and previsualizing begins. Imagine we want an FPV drone shot in our studio. For instance, it should fly in low, dip under the overhead microphone, circle tightly around Nino’s head, almost brushing his shoulders, and then pull back up.

In real life, this “simple” idea would require a super-experienced FPV operator with a suitable drone, safety checks, insurance, and nerves of steel for Nino (because we don’t want him to be in a helmet for this scene – please, don’t repeat at home!) Before we commit, let’s test whether this shot is actually worth the hassle.

For that, we open our scan in Luma’s library and click “Reshoot” in the top right corner. This will open the browser-based editor. (There is a way to create a previsualization directly on your phone as well, but here we focus on the simplest option, available to everyone.)

The interface that appears after clicking the “Reshoot” button. Image source: MZed

Step-by-step guide for the shot previs

What we need to do now is to animate the virtual camera. As you see in the screenshot above, the Luma app offers some preset trajectories for its movement. However, as we need a particular path, we should select “Custom.” The software also allows you to change the focal length, set the distance to objects, and adjust more precise parameters such as roll, pitch, and yaw. Or we can use the mouse to manually move the camera to the starting position of our drone flight, slightly above eye level, approaching from behind. Once we like the framing, we click “Add Keyframe.” This will lock in Position A.

Starting point of the shot. Image source: MZed

Now we “fly” forward using the mouse controls, lowering the virtual camera just under the mic and orbiting. We adjust the pitch slightly to maintain focus on Nino’s face. Once we’re satisfied with the frame, we hit “Add Keyframe” again. As soon as we have points A and B, Luma figures out the camera movement, and we can already click on the playhead to watch it.

Adding the next keyframe. Image source: MZed

Plus, we can add as many additional keyframes as we like to make the movement even more elaborate. As you see above, the software offers additional controls. If the motion feels too robotic, we can increase its smoothness in the “Motion” menu. If it feels too slow, shorten the shot’s duration. We can also drag particular keyframes closer together or further away from each other to play with the drone’s speed. Or simulate slight handheld energy by subtly adjusting roll values between keyframes.

Once the shot previsualization looks good to you, click “Render Video,” select the required aspect ratio, frame rate, and resolution, and export the clip. Luma currently allows up to 30 renders per month for free accounts, but it is more than enough for pre-production tests, in my opinion. Resolution is limited to FullHD unless you go Pro, but again, for a mere previs, it is totally fine.

And other impossible shots you can test

This is the resulting FPV-style drone shot of Nino, which we got inside Luma’s browser-based app:

It’s not perfect, and no, the detail in the space isn’t top-notch. However, the entire previsualization process only took a few minutes, and now we have something to discuss: Is this idea brilliant or ridiculous? Do we really need it for the scene? If so, which aspects of the shot should we make sure to address during preproduction? Now we can show the previs export to an experienced FPV pilot and decide on how to execute it.

This way, by having a simple 3D scan of your scene, you can also test:

  • A comprehensive dolly zoom, like in Hitchcock’s “Vertigo.”
  • A crane moves in a tight space that would normally require removing a wall
  • A top-down angle, which also requires lots of build-up
  • A camera passing through a doorway into a room with no space for a Steadicam
  • Or even something that wouldn’t be possible without special or visual effects – like hallway rotations in Christopher Nolan’s “Inception.”

And it’s not just difficult or seemingly impossible shots that benefit from this workflow. Subtle elements matter just as much, and you can test those as well to previsualize the scene. You know, sometimes you will discover that a simple push-in can add the emotional depth you need. In my view, that kind of insight is priceless before production begins.

From a virtual shot to a story reel

Once you’ve rendered a few such camera tests for a particular scene, the next step is simple. Import them into the editing software of your choice, lay them out on the timeline, add some temporary music, or maybe even a rough sound design layer, and voila! You have a story reel that demonstrates your vision for a scene in a concrete space.

This reel is by no means polished, but it clearly communicates the shot progression, the scene’s rhythm, the specific framing, and the visual choreography on location. You can share it with your DP, director, producer, and lighting and/or production designers to make further creative decisions with greater clarity and confidence.

Conclusion

So, this was one way to test ideas beforehand that are difficult to execute or even seem impossible. I like it because it is easy and fast, and it doesn’t require any extra technical knowledge of deep learning models, NeRFs, or 3D sculpting. Thus, it is accessible to any filmmaker or content creator out there.

In the MZed course “The Efficient Filmmaker: Scan and Plan”, we go through this process in more detail, talk about further lifehacks, and take a look at additional useful workflows. (For instance, on turning your scans into storyboards, or stylizing resulting shots using AI). Head over here to learn more about it.

Now I’m curious: Have you ever envisioned a shot that felt impossible or too difficult in your head? How did you approach the preparation process? Do you generally do previs? What tools do you use for it? Let’s talk in the comments below!

Feature image source: MZed.

Full disclosure: MZed is owned by CineD.

1 Comment

Filter:
all
Sort by:
latest
Filter:
all
Sort by:
latest