LearnViz CompSci
Home / Learn / Graphics, Vision & Media / Structure from Motion

Advanced Topic 44 of 49 Graphics, Vision & Media

Structure from Motion

Reconstructing 3D scenes from image sequences.

How it works

Structure from motion reconstructs 3D structure of a scene and the camera poses from a set of overlapping 2D images taken from different viewpoints. The pipeline detects and matches features across images, estimates geometry with the essential or fundamental matrix, triangulates 3D points, incrementally registers new cameras with perspective-n-point and refines everything with bundle adjustment, a large nonlinear least-squares problem minimizing reprojection error. Dense multi-view stereo then builds detailed surfaces. COLMAP is a popular implementation.

Where it is used

Photogrammetry for mapping, archaeology and film, 3D scanning from photos, drone surveying, visual effects and creating datasets for neural radiance fields use structure from motion.

Common pitfall

It needs sufficient texture and parallax; pure rotation or planar scenes cause degeneracy. The result has unknown scale unless a reference is given. Drift and wrong matches can break the reconstruction, so robust estimation with RANSAC and good image overlap are essential. Processing large collections is computationally heavy. Capture with plenty of overlap and varied viewpoints.

Where it fits in the course

Structure from Motion is topic 44 of 49 in Graphics, Vision & Media, rated Advanced. It follows Stereo Vision. The next topic is SLAM.

Practice and explore

How to study this topic

  1. Read the explanation above and restate the definition in your own words.
  2. Open the interactive lesson and answer its mini quiz.
  3. Read the common pitfall, then write a short example that avoids it.
  4. Check your understanding in the Graphics, Vision & Media quiz, and revisit the topic if you miss a question.

Frequently asked questions

What is Structure from Motion?
Reconstructing 3D scenes from image sequences. It belongs to Graphics, Vision & Media and is an advanced-level topic.
Where do people use Structure from Motion?
Photogrammetry for mapping, archaeology and film, 3D scanning from photos, drone surveying, visual effects and creating datasets for neural radiance fields use structure from motion.
What is the most common mistake with Structure from Motion?
It needs sufficient texture and parallax; pure rotation or planar scenes cause degeneracy.
What should I learn before this topic?
Start with Stereo Vision, the topic just before it in Graphics, Vision & Media.
How can I practice it?
Answer the mini quiz in the interactive lesson, then take the Graphics, Vision & Media quiz.
Is Structure from Motion hard to learn?
It is an Advanced topic. Expect to read the explanation more than once and to connect it with earlier topics. Working through a small example by hand helps most.
How long does it take to learn Structure from Motion?
Most learners can read the explanation and finish the interactive lesson in under an hour. Fluency comes from applying the idea, so plan a few short practice sessions with the quiz.
Which topics are related to Structure from Motion?
Stereo Vision, SLAM, Optical Flow are closely connected. The full list is in the related topics section of this page.
What should I study after Structure from Motion?
Continue with SLAM, the next topic in Graphics, Vision & Media.

Related topics