Substance 3D Collection
Model, texture, and render 3D assets and scenes.
Videogrammetry is a measurement technology that gathers 3D information from video footage to create a 3D mesh. Like photogrammetry, 3D models from video may seem enticing, so we’ll discuss two major differences between using still images verse video.
Let’s look at two major challenges you may face when trying to create 3D objects from video versus still images.
Resolution is crucial to getting the best results while 3D scanning. Resolution is all about how much detail is contained within a space and will influence how well your 3D scan turns out in the end. Lower resolution and any artifacts that can occur in an image is going to be a major hindrance during the capture process. Remember, if your image has more details, your 3D object will, too.
While there are 8k cameras on the market, they are extremely expensive. In the end, it makes little sense to use an expensive, high-end video camera to increase resolution, when still images could be used instead. In every category, photographs are going to produce cleaner, higher resolution images for cheaper, too.
The next major challenge you’ll face using video to create 3D models is motion blur. Blur only occurs in still images that were shot with a low shutter speed from a moving or hand-held camera. When it comes to 3D capture, this shouldn’t be an issue. By using a tripod, clean images without blur can be taken from every angle of the object you need.
For video, on the other hand, the trouble of motion blur isn’t solved as easily. The best way to combat motion blur in video is to shoot with faster shutter speeds. Shooting higher frames per second is doable enough, however, keep in mind that a faster shutter speed also means less light can reach the camera’s sensors. To compensate for this, you’ll want to be in a controlled lighting environment for best results. Adding too much light can introduce shadows and highlights that will also negatively affect your final scanned object.
Image by Andrew Palmer.
We’ve discussed two major reasons why you should use still images over video, so let’s talk about where video fits in this space and why you might consider using it.
It can be time consuming to take the dozens of individual images needed for a successful 3D scan. There are several solutions, usually smartphone apps, which offer video to 3D options because of the ease of use of video. Recording an object is much faster than taking picture after picture.
If you’re looking for a quick answer and have access to a smartphone, then creating 3D models from video with an app might be the simplest solution for you. That said, you’ll always find better results using still images with high resolutions. It’s important to remember that these apps are using individual frames from your video, which are still images anyway. Every artist making models from video should ask themselves if the time saved outweighs the improvement in quality you’ll get by just taking still images yourself with a tripod and camera.
Images by Andrew Palmer.
Images by Andrew Palmer.
If you’re using a phone, for the best results, you’ll want to use a smartphone with LiDAR sensors such as the iPhone 12 and 13 Pro.
As a part of Adobe Substance 3D, we offer Substance 3D Sampler, a powerful and easy to use 3D capture software for creating 3D objects from photos. It takes a lot of processing power to accurately photo or video scan an object. For Sampler in 2023, we recommend an Intel i7 or AMD Ryzen 7 CPU or equivalent. Users will also want a powerful GPU such as one at the 30-level series or higher. For RAM 32GB is good, though for many projects, 64 is optimal.
For a full guide on Sampler’s hardware requirements and recommendations, review our documentation.
Substance 3D Sampler is a great way to get started with photogrammetry. While we don’t support video to 3D object solutions, our AI-powered technology makes the process simple. If you can take clean pictures with good lighting, Sampler will manage the rest.
Using the same masking technology in Photoshop, Sampler will automatically mask your images so only the main subject is recreated. After you’re satisfied with the masking, Sampler will process the images, create a point cloud, and produce a 3D mesh with textures.
Sampler allows you to mix and blend materials to quickly make amazing adjustments to the model’s appearances.
As you can see, creating 3D models from video is interesting, but isn’t the best solution in most cases. To dive deeper into 3D capture and everything you can do with Adobe Substance 3D Sampler, check out this introductory video where one of our talented 3D artists showcases the capture process from start to finish using still images.
Photogrammetry is the process of extracting measurement data from photographs. With enough images of an object from different angles, we can use software to reconstruct that object as a 3D model. Videogrammetry works the same way, only it uses video footage as the source of these measurements.
Sampler does not support videogrammetry directly. To use Sampler’s 3D Capture tool, you will submit a series of still images into the wizard. That said, anybody could use a tool like Photoshop, or a video editing software to extract still images from a video. For best results, we recommend capturing high-resolution still images.
Creating 3D objects from images is a great way to augment your 3D pipeline. While objects that may be prominently featured in a shot may merit the detail and care of hands-on modeling, many shots are made up of countless objects. An artist or team can save lots of time by using photogrammetry to produce realistic 3D assets.