Uploaded July 2025 | Updated September 2026, 2 weeks ago
Struggling to get a clean camera solve on complex shots in SynthEyes? When the standard auto-tracker fails due to large moving subjects, reflective surfaces, or excessive grain, there's a powerful tool you should use before resorting to manual supervised tracking.
This tutorial by Elizabeth Postol explores the ML Motion Estimation tracking mode. Leverage a deep learning model to significantly improve your results. We explain how this mode uses dense optical flow and motion vectors to guide the auto-tracker, resulting in more accurate placements and a more stable solve, even on challenging footage.
We tackle two difficult scenes, comparing the standard auto-tracker with the ML Motion Estimation mode. See how to overcome common tracking failures and achieve a production-ready solve with a low error margin (Hpix) in just a few clicks.
Learn more about SynthEyes here : borisfx.com/products/syntheyes
// K E Y P O I N T S //
⦁ Learn why standard auto-tracking can fail on complex footage with large moving objects, reflections, or grain.
⦁ Enable the ML Motion Estimation mode for vastly superior automated results.
⦁ Understand how this mode uses a deep learning model to compute motion vectors.
⦁ See a direct comparison of solve techniques.
⦁ A simple workflow to clean up and refine your AI-assisted track for a final, stable camera solve.
// D I S C O R D C H A N N E L //
borisfxdiscord.com
#matchmove #vfx #syntheyes
/ / C H A P T E R L I S T / /
00:00 Intro
00:20 First Auto solve
00:36 Why Auto solve fails
01:00 Creating garbage matte with Mask ML
01:30 Auto solve with ML Motion Estimation
02:18 Tracker clean up
02:36 QC
02:40 Solving dolly shot. Analyzing why Auto solve fails
04:18 Why low HPIX not always mean a good track
05:10 Refining the scene
05:50 QC
06:25 Outro
Struggling to get a clean camera solve on complex shots in SynthEyes? When the standard auto-tracker fails due to large moving subjects, reflective surfaces, or excessive grain, there's a powerful tool you should use before resorting to manual supervised tracking.
This tutorial by Elizabeth Postol explores the ML Motion Estimation tracking mode. Leverage a deep learning model to significantly improve your results. We explain how this mode uses dense optical flow and motion vectors to guide the auto-tracker, resulting in more accurate placements and a more stable solve, even on challenging footage.
We tackle two difficult scenes, comparing the standard auto-tracker with the ML Motion Estimation mode. See how to overcome common tracking failures and achieve a production-ready solve with a low error margin (Hpix) in just a few clicks.
Learn more about SynthEyes here : borisfx.com/products/syntheyes
// K E Y P O I N T S //
⦁ Learn why standard auto-tracking can fail on complex footage with large moving objects, reflections, or grain.
⦁ Enable the ML Motion Estimation mode for vastly superior automated results.
⦁ Understand how this mode uses a deep learning model to compute motion vectors.
⦁ See a direct comparison of solve techniques.
⦁ A simple workflow to clean up and refine your AI-assisted track for a final, stable camera solve.
// D I S C O R D C H A N N E L //
borisfxdiscord.com
#matchmove #vfx #syntheyes
/ / C H A P T E R L I S T / /
00:00 Intro
00:20 First Auto solve
00:36 Why Auto solve fails
01:00 Creating garbage matte with Mask ML
01:30 Auto solve with ML Motion Estimation
02:18 Tracker clean up
02:36 QC
02:40 Solving dolly shot. Analyzing why Auto solve fails
04:18 Why low HPIX not always mean a good track
05:10 Refining the scene
05:50 QC
06:25 Outro
![SynthEyes 2026 - The Important Stuff in Under 3 Minutes [Boris FX]](https://i.ytimg.com/vi/pISFgzMkpRY/mqdefault.jpg "SynthEyes 2026 - The Important Stuff in Under 3 Minutes [Boris FX]
Boris FX SynthEyes 2026 is an important upgrade for all 3D matchmove artists. Whether you are a veteran tracker or just starting your journey, youll find some big improvements aimed at you.
This release marks the beginning of a UI modernization effort, improving discoverability and readability, while respecting the muscle memory artists have built over years. We’ve integrated award-winning Mocha technology to power a new Supervised Tracking option. This holds onto features through challenging motion blur and 3D rotations, without needing to babysit.
For those working with complex rigs, the new Live GeoH preview is a game-changer. It transforms GeoH into an interactive, intuitive experience, closer to posing or animation, providing instant feedback as you adjust constraints or trackers. Combined with expanded USD/USDZ support and a dedicated ST Map exporter, SynthEyes 2026 is the most approachable and capable version yet.
Download a free trial or upgrade your existing license: https://borisfx.com/products/syntheyes/
// K E Y P O I N T S //
⦁ Mocha Point Tracker: A robust new tracker type that handles rotation, scale, and skew for superior stability on difficult shots.
⦁ Live GeoH Interaction: Real-time previews for object hierarchies, making it easier to refine complex setups like wheels, arms, and suspensions.
⦁ UI Modernization Phase 1: A refreshed interface designed for better readability and discovery without disrupting established workflows.
⦁ Expanded USD Support: Comprehensive import capabilities for USD meshes, assets, and full scenes to fit any modern pipeline.
⦁ ST Map Exporter: A dedicated tool for generating distortion maps via the exporter or multi-exporter in a single pass.
Learn more about SynthEyes here: https://borisfx.com/products/syntheyes/
/ / C H A P T E R L I S T / /
00:00 Overview
00:39 UI Modernization
00:59 Mocha Point Tracker
01:34 Live GeoH Preview
02:13 More USD & USDz Support
02:27 ST Map Exporter")
![3D Camera Tracking for Easy Paint & Roto in Silhouette [Boris FX]](https://i.ytimg.com/vi/r5HiV2aaDR4/mqdefault.jpg "3D Camera Tracking for Easy Paint & Roto in Silhouette [Boris FX]
Explore the powerful new camera tracking capabilities integrated into Silhouette 2025.5 and powered by industry-leading SynthEyes technology. Solve 3D scenes directly within Silhouette without needing any other software.
Learn how to perform an automatic solve, generate a point cloud, and create 3D cards for precise paint and compositing tasks. Join Ben Brownlee and walk through several practical examples. Utilize these 3D planes for stabilized painting with the unproject and matchmove functions. Learn professional tips for the best solves by adjusting settings and using occlusion mattes to exclude problematic areas like water or sky.
This workflow is essential for tackling difficult shots with parallax, where traditional planar tracking falls short. Follow along to see how this integrated tool makes challenging paint and roto work more manageable and accurate than ever before.
// K E Y P O I N T S //
⦁ Intro to the 3D camera solver, powered by SynthEyes technology.
⦁ Perform an automatic solve and create 3D cards from the resulting point cloud data.
⦁ Use solved 3D planes for stabilized painting with the unproject and matchmove functions.
⦁ Best practices for improving your solve by adjusting feature settings and using occlusion mattes to guide the tracker.
Learn more about Silhouette here: https://borisfx.com/products/silhouette/
// D I S C O R D C H A N N E L //
https://www.borisfxdiscord.com/
#digitalpainting #VFX #videoeffects
/ / C H A P T E R L I S T / /
00:00 Intro
00:28 Solving the Scene in 3D
01:29 Creating the 3D cards
02:24 Unprojecting the viewer
05:16 Dive deeper into the camera solve controls
08:36 Why use 3D cards instead of planar tracking
10:31 Things to consider with Autopaint")
