Theia3D + GeForce RTX 5090: Next-Level Speed for Markerless Motion Capture

June 11, 2025
Theia
BIO
Recent Posts
Aligning the Frame: How Reference Frame Differences Impact Markerless vs. Marker-Based Comparisons
When comparing markerless and marker-based motion capture, researchers often find differences in the kinematic outputs. But how much of these discrepancies are true differences in how the movement is measured, and how much comes down to how each system defines the anatomical reference frames? A new study out of Rush University Medical Center and Aesculap AG explores this question using Theia3D - and finds that aligning coordinate systems dramatically improves agreement between systems.
How Tonal Uses Markerless Motion Capture from Theia3D to Scale Biomechanics Research and Personalize Strength Training
Tonal’s biomechanics team uses Theia3D markerless motion capture to accelerate research, validate motion models, and power features like Smart View. Learn how they’ve streamlined data collection, scaled biomechanics across participants, and turned lab insights into real-time, personalized strength training.
Improving Change of Direction Performance in Adolescent Soccer Players Using Markerless Motion Capture
A recent study published in Frontiers in Sports and Active Living highlights how brief technique modification training can significantly improve change of direction (COD) performance in adolescent soccer players. Using a markerless motion capture system, the study demonstrated measurable improvements in COD mechanics. This research underscores the value of markerless technology like Theia3D for coaches and sports scientists aiming to optimize performance and reduce injury in adolescent athletes.

Summary

In case you missed it, we recently updated Theia3D Axiom to include support for the newest generation of NVIDIA graphics cards - the GeForce RTX 50 Series. With this exciting update, these new graphics cards can be used to unlock even greater processing performance from Theia3D, allowing you to push the limits of your markerless motion capture workflow. After our initial announcement which included benchmarks for the new RTX 5080, we’re following up to provide even more exciting results - RTX 5090 benchmarks.

Theia3D just got faster. A lot faster.

We’ve expanded support for NVIDIA’s newest generation of GPUs — the GeForce RTX 50 Series — and the performance gains are generating serious torque. Following our initial RTX 5080 benchmarks, we now have results for the flagship GeForce RTX 5090, and the numbers speak for themselves.

Whether you’re analyzing athletic performance or clinical gait data, Theia3D now delivers insights even faster, with the RTX 5090 unlocking up to 3x faster person detection and nearly 50% shorter total processing times compared to the RTX 4090.


Why Hardware Still Matters


At Theia, we’re always improving our markerless tracking algorithms, but software is only half of the equation. For truly next-level performance, the right hardware makes all the difference.

That’s why we were eager to get our hands on NVIDIA’s newly released RTX 50 Series. These GPUs promise major performance gains, and after some effort (and a bit of a wait), we managed to secure the RTX 5090 for testing.

To see how it stacks up, we ran Theia3D on three different GPUs:

  • RTX 4090 – previous-gen flagship
  • RTX 5080 – next-gen mid-tier
  • RTX 5090 – next-gen flagship

Same dataset, same settings — just different hardware. And the results speak for themselves.
‍

Test Setup


For these tests, we used a dataset with the following attributes and configurations in Theia3D:

  • Theia3D Settings:
    • Standard GPU configuration for each card in Theia3D
      • Default Person Detection Threads
      • Default Joint Detection Threads
    • Person Detector Skip Frames: 3
    • (Results for Person Detector Skip Frames: 1 are also reported)
  • Dataset:
    • 7 cameras
    • 1750 frames per camera
    • 960 x 540 resolution
    • AVI file type with MJPEG encoding
    • Standard Theia3D configuration for each

Performance Highlights at a Glance


Upgrading to the GeForce RTX 5090 delivers serious speed gains in every step of the Theia3D pipeline:

  • 3x faster person detection than RTX 4090
  • 83% faster joint detection rates
  • Total processing time cut nearly in half

Even compared to the already powerful RTX 5080, the 5090 delivers another big jump in performance, making it the clear choice for users who want the fastest possible motion capture workflows.
‍

Detailed Performance Results
‍

  GeForce RTX 4090 GeForce RTX 5080 GeForce RTX 5090
Person Detection
Time [s]		 37.6 18.2 (52% ↓) 14 (63% ↓)
Rate [fps] 325 673 (107% ↑) 817 (151% ↑)
Skip Frames: 1
Rate [fps]		 108 224 (107% ↑) 437.5 (305% ↑)
Joint Detection
Time [s]		 42.1 34.1 (19% ↓) 23 (45% ↓)
Rate [people/s] 657 811 (23% ↑) 1202 (83% ↑)
Total Time 1 min 30 sec 1 min 4 sec (29% ↓) 46 sec (49% ↓)

‍

Once again, throughout our testing the RTX 5090 performed reliably under sustained loads. For those monitoring their own GPU performance, we recommend GPU-Z - a simple, effective tool for real-time GPU diagnostics.

‍

What This Means For You 


Theia3D now delivers faster turnaround without changing a single step in your workflow. Whether you’re running large research studies or working with athletes in the field, the RTX 5090 allows you to:

  • Process more data in less time
  • Maintain high fidelity without sacrificing speed
  • Scale your motion capture studies without bottlenecks

It’s not just about speed — it’s about unlocking new research and performance possibilities.
‍

Thinking About an Upgrade?


If you’re considering a hardware refresh, the GeForce RTX 5090 is a clear leap forward. And because Theia3D Axiom auto-detects and optimizes for supported GPUs, the upgrade is seamless once installed.

These GPUs are in high demand, so if you’re building or refreshing a Theia-ready system, now is the time to act.
‍

Get Started with Theia Axiom


Already a Theia3D user? Axiom is available now via the customer support portal.

Want to see Theia3D in action?  Book a demo or email us at info@theiamarkerless.com

Stay in the loop — follow us on LinkedIn and X (Twitter) for the latest updates.

‍

Recent Posts
Aligning the Frame: How Reference Frame Differences Impact Markerless vs. Marker-Based Comparisons
When comparing markerless and marker-based motion capture, researchers often find differences in the kinematic outputs. But how much of these discrepancies are true differences in how the movement is measured, and how much comes down to how each system defines the anatomical reference frames? A new study out of Rush University Medical Center and Aesculap AG explores this question using Theia3D - and finds that aligning coordinate systems dramatically improves agreement between systems.
How Tonal Uses Markerless Motion Capture from Theia3D to Scale Biomechanics Research and Personalize Strength Training
Tonal’s biomechanics team uses Theia3D markerless motion capture to accelerate research, validate motion models, and power features like Smart View. Learn how they’ve streamlined data collection, scaled biomechanics across participants, and turned lab insights into real-time, personalized strength training.
Improving Change of Direction Performance in Adolescent Soccer Players Using Markerless Motion Capture
A recent study published in Frontiers in Sports and Active Living highlights how brief technique modification training can significantly improve change of direction (COD) performance in adolescent soccer players. Using a markerless motion capture system, the study demonstrated measurable improvements in COD mechanics. This research underscores the value of markerless technology like Theia3D for coaches and sports scientists aiming to optimize performance and reduce injury in adolescent athletes.
In this blog
Summary