Search Results

In high-stakes matchmoving, the conflict between speed and sub-pixel accuracy is constant. Manually inputting camera data is a notorious drain on production time, it requires searching disparate sources for specs, cross-referencing active sensor areas and shooting modes, and then hoping the information found online is reliable. The PFTrack Camera Sensor Database fundamentally changes this workflow. It translates repetitive, error-prone technical setups into a professional standard in just a few clicks, enabling you to dedicate your focus to achieving the highest-quality camera solve, not tedious data entry. This article details why the Database is indispensable for your pipeline, how to leverage its functionality, and the methods for extending it to accommodate proprietary or custom camera assets. Why Accuracy Matters: The Foundation of Your Solve Field of View (FOV) is governed by two core parameters: the camera's sensor size and its focal length. For example, a 50 mm lens on a full-frame (36×24 mm) sensor gives about a 39.6° horizontal FOV; change the sensor to APS-C (24×16 mm), and the same lens yields a narrower FOV of only ≈27° (the “crop factor” effect). These values determine how 2D image data is projected into 3D space, directly influencing perceived scale, perspective, and depth. While PFTrack is highly capable of estimating FOV, and even lens distortion, when explicit sensor or focal length data is unknown, PFTrack performs best when supplied with accurate physical measurements. Providing real-world camera metrics ensures the strongest foundation for precise, repeatable results. This level of accuracy becomes especially critical in multi-shot sequences, where consistent focal length interpretation across shots is essential. Without defined physical parameters, slight variations in estimated FOV can lead to inconsistencies in scene scale, misaligned assets, or solve drift across edits. Even small discrepancies in sensor dimensions, on the order of tenths of a millimeter, can introduce meaningful errors in FOV calculations, which in turn degrade 3D track quality and overall solve fidelity. The bottom line is that supplying accurate, real-world camera metrics removes ambiguity: the solver has one less unknown to estimate, which yields more stable and precise production-grade tracking. While manually entering sensor details can be error prone, especially when relying on ambiguous manufacturer specifications, PFTrack provides the perfect solution: the Camera Sensor Database. The Workflow Advantage: Why Presets Matter When you need to enter the camera sensor size manually switching to the camera sensor database provides four immediate benefits: Eliminating Ambiguity: Manufacturer specifications can be vague, often listing "total" vs. "effective" pixels or using very generalised terminology to approximate size such as ‘Super 35’. Preventing Calculation Errors: Calculating the physical size of a sensor based on pixel pitch or unverified online info is prone to error. A single millimetre of discrepancy can degrade the quality of a 3D solve. Managing Shoot Modes: Modern cameras frequently use specific shoot modes (cropping or windowing the sensor). Manually calculating these deviations is complex; the database provides accurate presets from major manufacturers for these modes in seconds. Team-Wide Consistency: On large features with multiple artists, the database ensures every user is working with identical, verified camera parameters, preventing drift between different shots. For a deeper look on sensor modes and terminology we have a camera sensor size guide available here. Using the Fuji X-H2 as an example: switching to 4K 16:9 significantly reduces the active sensor area compared to the full sensor, causing a noticeable change in field of view (FOV) when using the same lens. Presuming the manufacturer’s full sensor specifications apply to all shoot modes can lead to inaccurate tracking results. The camera sensor database addresses this by providing accurate shoot mode data matched to your clip’s resolution. How to Use the Sensor Database The sensor database is integrated into the areas where camera data is defined. You can access it in two primary locations, the Clip Input node and the Camera Lens Distortion Presets panel. It’s designed to be a seamless part of your workflow and can be launched using the following button . It is broken down into three sections: Database selection and information. Camera make and model selection. Sensor information and shoot mode selection. Camera Sensor Database in action Quickly select verified sensor presets and ensure physically accurate FOV across your shots in PFTrack. Find Your Camera: Type in your manufacturer or camera model in or browse the list (e.g. "RED Komodo" or "Sony Venice"). Select the ‘Shoot Mode’: Pick the sensor shoot mode used in the clip. Modes that don’t match resolution will be greyed out. Click ‘Use Sensor’ to automatically fill in the sensor measurements in both Clip Input Node & camera lens distortion presets. You can filter the list further by checking for proxies 2x & 4x as well as matching resolution and aspect ratio. Pro Tip: Need more flexibility The database allows users to switch between The Pixel Farm’s internal database and the Matchmove Machines database. This provides unparalleled flexibility, if a specific or niche camera body isn't found in one, a quick toggle often finds it in the other. Expanding the Database: Custom XML Data For proprietary camera rigs, prototypes, or bespoke setups, PFTrack allows you to expand the database with your own metrics. By creating a custom sensor xml file, you can define custom sensor sizes for your projects that can be selected directly within the camera sensor database. File structure: To simplify the process, PFTrack ships with an example-sensors.xml file. This file serves as a template for defining your own , , and . You can find the template in the installation folder on your system /media/example-sensors.xml. The custom .xml file you create can be named whatever you want. Where to put your custom file? Once you have created your custom xml file you will need to place it in the directory listed below, once in place relaunch PFTrack and you will have access to the updated file via the camera sensor database dropdown. MacOS: /Users/USERNAME/Documents/The Pixel Farm/PFTrack/presets/sensors Windows: C:\Users\USERNAME\Documents\The Pixel Farm\PFTrack\presets\sensors Linux: /home/USERNAME/Documents/The Pixel Farm/PFTrack/presets/sensors For a deep dive into the XML schema and specific formatting rules, please refer to our User Defined File Formats Documentation. Pro Tip: Centralised Management - for Multi-User Pipelines For studios managing multiple PFTrack seats, maintaining a "single source of truth" for camera data is essential for pipeline integrity. Rather than managing local files on every machine, you can centralise your custom presets: Place your custom.xml file in a centralised network location accessible by all workstations. In the PFTrack software settings, navigate to ‘Additional file locations’ and point the ‘Sensor presets’ path to your network directory. Once configured, any update made to the central XML file is propagated across the entire facility and available in PFTrack on relaunch. This centralised approach allows Pipeline TDs to deploy updated camera information for a specific feature or production once, ensuring every artist, regardless of their workstation, is working with identical, verified technical specifications. You can use multiple xml files as well with each defining the cameras used for a specific production, this can be further customised with a project or company logo. Conclusion The Camera Sensor Database isn't just a convenience; it is a precision tool. By moving away from manual entry and unverified data, you safeguard your solves and streamline your pipeline. Ready to upgrade your tracking? Ensure you are running the latest version of PFTrack to access the most recent camera sensor database updates. Want to add a camera model to our database, or discuss a camera system you're using for tracking with other professional matchmovers? Join our support community here. Try PFTrack Solo Free Try PFTrack Solo free for 7 days, with full export functionality, enough time to try out The Camera Sensor database for yourself.

Professional camera tracking and photogrammetry on hardware that works for you Recommended configurations for macOS, Windows, and Linux The Pixel Farm Ltd — 2026 www.pftrack.com Hardware Guide Table of Contents Introduction macOS: Native Apple Silicon Performance Windows: Flexibility and GPU Choice Linux: The Pipeline and Facility Platform Storage Strategy for PFTrack PFTrack Editions and Licensing Multi-Seat Studio Deployment PFTrack Support & Resources Quick-Start Recommendation Introduction PFTrack is a production-grade spatial intelligence platform used by VFX studios, forensic science organisations, virtual production facilities, and architectural visualisation teams worldwide. For over two decades it has powered Oscar-winning feature films, high-end episodic television, and forensic investigations demanding absolute precision. Unlike many professional creative tools, PFTrack does not demand exotic hardware. It runs natively on macOS (including Apple Silicon), Windows, and Linux, and is designed to deliver professional results on a wide range of configurations, from a solo artist’s laptop to a multi-seat studio deployment. However, PFTrack’s workload is distinctive. Understanding what drives performance in camera tracking, photogrammetry, and scene reconstruction will help you choose hardware that maximises your productivity and avoids wasting budget on components that don’t matter. The key factors are: Single-core CPU speed for tracking solves. Camera and object tracking in PFTrack is fundamentally a mathematical optimisation problem. The solver iterates through tracked features, refining camera parameters until the solution converges. This work is heavily dependent on single-core CPU performance. A CPU with fast individual cores will solve faster and keep the interface responsive during interactive work. Core count matters less than per-core speed for day-to-day tracking. Multi-core throughput for photogrammetry and batch processing. Photogrammetry — reconstructing 3D geometry from multiple images, involves ML-accelerated feature matching, bundle adjustment, and mesh generation across hundreds or thousands of images. These operations benefit from multiple CPU cores working in parallel. Similarly, batch processing multiple shots benefits from core count. For studios running heavy photogrammetry workloads, a CPU that balances strong single-core and high multi-core performance is ideal. GPU acceleration for ML features and display. PFTrack uses GPU acceleration via OpenCL for ML-accelerated feature matching, point cloud visualisation, and real-time viewport display of complex scenes with millions of tracking points. A modern mid-range GPU is sufficient for most workflows. PFTrack supports both NVIDIA and AMD GPUs on Windows and Linux, and uses the integrated GPU on Apple Silicon Macs — which delivers excellent performance thanks to the unified memory architecture. Memory for large image sets and point clouds. Tracking a single shot requires modest memory. But photogrammetry from hundreds of high-resolution images, or working with dense LiDAR point clouds, can consume significant RAM. 16 GB is the practical minimum for professional use; 32 GB or more is recommended for photogrammetry and large-scale scene reconstruction. This PFTrack hardware guide covers recommended configurations for all three supported platforms, storage strategy, and deployment guidance for multi-seat facilities. PFTrack runs on hardware you already own. An M4 MacBook Pro, a mid-range Windows workstation, or a Linux box with any modern GPU can deliver production-quality camera solves and photogrammetry. This guide will help you choose the right configuration for your workflow. macOS: Native Apple Silicon Performance PFTrack runs natively on Apple Silicon and takes full advantage of the unified memory architecture of M-series chips. The CPU, GPU, and Neural Engine share the same high-bandwidth memory pool, which is particularly beneficial for photogrammetry workflows where large image datasets and point clouds are accessed by both CPU and GPU operations without data transfer bottlenecks. For the majority of individual PFTrack users, a Mac with Apple Silicon is an excellent choice. It delivers strong single-core performance for tracking solves, efficient GPU acceleration for ML features and viewport display, and low power consumption in a quiet, compact form factor. Powered by the M5 chip, this MacBook delivers exceptional performance and smooth workflows, making it an excellent choice for running PFTrack efficiently and reliably. The M5 Pro and M5 Max MacBook Pro (March 2026) The latest MacBook Pro models, launched in March 2026, introduce the M5 Pro and M5 Max chips built on Apple’s new Fusion Architecture. This is a significant upgrade for PFTrack users: Up to 30% faster CPU performance over the M4 Pro generation — directly translating to faster camera solves and more responsive interactive tracking. PFTrack’s solver is heavily single-core dependent, and the M5 Pro’s faster cores mean noticeably quicker convergence on complex shots. Up to 50% faster GPU performance with Neural Accelerators in every GPU core. PFTrack’s ML-accelerated feature matching in the Photo Survey node and real-time viewport display of dense point clouds both benefit directly from the GPU uplift. Up to 2x faster SSD speeds improve footage loading, image set access during photogrammetry, and point cloud streaming. Combined with 1 TB standard storage on M5 Pro models (2 TB on M5 Max), the new MacBook Pros have more room for project footage out of the box. Higher unified memory bandwidth benefits workflows where large datasets are shared between CPU and GPU — precisely the case when PFTrack is running ML feature matching on hundreds of high-resolution photogrammetry images while simultaneously displaying a dense 3D point cloud in the viewport. Recommended Configurations Solo Artist / Freelancer Studio Artist / Regular Production Heavy Photogrammetry / Large Scenes Best Mac MacBook Pro 14" (M5) or Mac Mini (M4 Pro) MacBook Pro 14"/16" (M5 Pro) or Mac Mini Pro (M4 Pro) MacBook Pro 16" (M5 Max) or Mac Studio (M4 Max / M3 Ultra) Chip M5 / M4 Pro M5 Pro (15–18 core CPU, 16–20 core GPU) M5 Max (18 core CPU, 32–40 core GPU) / M3 Ultra Unified Memory 16–24 GB 24–48 GB 36–128 GB (M5 Max) / 96–512 GB (M3 Ultra) Storage 1 TB (standard on M5 MacBook Pro) 1–2 TB (M5 Pro starts at 1 TB) 2–8 TB (M5 Max starts at 2 TB) Best For Solo matchmoving, moderate tracking workloads, on-location use Daily production tracking, photogrammetry from tens of images, studio workflows Large-scale photogrammetry (hundreds of images), dense LiDAR, 8K footage, batch processing Approx. UK Price From ~£1,599 (Mac Mini M4 Pro) / ~£1,699 (MacBook Pro 14" M5) From ~£2,199 (MacBook Pro 14" M5 Pro) / ~£2,499 (16" M5 Pro) From ~£3,599 (MacBook Pro 14" M5 Max) / ~£2,099 (Mac Studio M4 Max) Our advice: For most matchmoving and tracking work, the MacBook Pro with M5 Pro and 24 GB of unified memory is now the sweet spot. Its 30% CPU uplift over the M4 Pro translates directly to faster solver convergence, and the 1 TB standard storage means you have room for active project footage without immediately needing external drives. For heavy photogrammetry, the M5 Max with 48 GB or more provides the GPU cores and memory bandwidth to handle large image sets efficiently. The Mac Mini with M4 Pro remains an excellent desktop option if you don’t need portability, and the Mac Studio with M4 Max or M3 Ultra is the top choice for the most demanding 8K and large-scale scene reconstruction work. Portability note: PFTrack is one of the few professional tracking tools that runs natively on Apple Silicon laptops. The MacBook Pro 14" with M5 Pro is a uniquely capable portable tracking workstation, ideal for VFX supervisors doing on-set tracking verification, client presentations, or field photogrammetry work. The M5 Pro’s Thunderbolt 5 and Wi-Fi 7 connectivity also make it well-suited for fast data transfer on location. Windows: Flexibility and GPU Choice PFTrack runs on Windows with GPU acceleration via OpenCL, supporting both NVIDIA and AMD GPUs. Windows gives you the widest hardware choice and the ability to configure a workstation specifically tuned to your workload, whether that’s pure matchmoving, heavy photogrammetry, or a mixed VFX pipeline where PFTrack sits alongside Maya, Nuke, and other DCC tools. CPU: Single-Core Speed Matters Most For camera tracking and solving, PFTrack benefits most from fast single-core CPU performance. The solver’s iterative optimisation runs primarily on a single thread, so a CPU with high clock speed and strong IPC (instructions per cycle) will deliver faster solves and a more responsive interface. For photogrammetry and batch processing, additional cores become valuable. The ideal CPU for a mixed PFTrack workload balances both: high single-core turbo speed with a healthy core count for parallel operations. Budget / Solo Mid-Range / Studio High-End / Photogrammetry CPU Intel Core i7-14700K or AMD Ryzen 7 7800X3D Intel Core i9-14900K or AMD Ryzen 9 7950X Intel Core i9-14900KS or AMD Ryzen 9 9950X RAM 16–32 GB DDR5 32–64 GB DDR5 64–128 GB DDR5 GPU NVIDIA RTX 4060 or AMD RX 7600 NVIDIA RTX 4070 Super or AMD RX 7800 XT NVIDIA RTX 4080 Super or AMD RX 7900 XTX Storage 1 TB NVMe SSD 2 TB NVMe SSD 2–4 TB NVMe SSD + bulk storage Best For Solo matchmoving, freelance tracking Studio production, regular photogrammetry, mixed DCC pipeline Large-scale photogrammetry, dense point clouds, 8K footage, batch processing Approx. Build Cost ~£1,000–£1,500 ~£1,800–£2,500 ~£3,000–£4,500 GPU: Mid-Range Is Enough for Most Workflows PFTrack uses OpenCL for GPU acceleration, which means it works with both NVIDIA and AMD GPUs. You do not need a top-tier graphics card for professional tracking work. The GPU is primarily used for ML-accelerated feature matching in the Photo Survey node, point cloud and mesh display in the viewport, and real-time preview of tracking results. You don’t need a workstation-grade GPU to get elite results. PFTrack is engineered to exploit even modest, off-the-shelf consumer cards to accelerate feature detection and solving. Whether you’re team NVIDIA (OpenCL) or team AMD (OpenCL), PFTrack utilises the parallel processing power of your GPU to turn hours of tracking into minutes. Our recommendation: An NVIDIA RTX 4070 Super or AMD RX 7800 XT hits the sweet spot for most PFTrack users. These mid-range cards provide ample compute for ML features and smooth viewport performance even with dense point clouds, at roughly £400–£550. You do not need to spend £1,000+ on a GPU for PFTrack unless you are running other GPU-intensive applications alongside it. NVIDIA vs AMD: Both are fully supported. If you also run applications that require CUDA (such as certain Nuke plugins, Resolve GPU acceleration, or ML training tools), choose NVIDIA. If PFTrack is your primary GPU-accelerated application, choose whichever offers the best value at your budget. Linux: The Pipeline and Facility Platform PFTrack supports Rocky Linux 8 and 9 (RHEL compatible), making it deployable in professional VFX pipeline environments alongside other Linux-based DCC tools. Linux is typically the choice for facilities running multi-seat deployments, render farms with integrated tracking, and custom pipeline automation. Hardware recommendations for Linux mirror the Windows guidance above — the same CPUs, GPUs, and memory configurations apply. The primary differences are operational: ✓ Rocky Linux 8 or 9 (RHEL 8/9 compatible) is the supported distribution. CentOS Stream 8/9 is also compatible. ✓ NVIDIA GPUs require the proprietary NVIDIA driver and OpenCL runtime. AMD GPUs require the ROCm or AMDGPU-PRO driver. ✓ PFTrack’s command-line interface (CLI) enables headless batch processing for pipeline automation, including tracking and photogrammetry jobs dispatched from production management tools. ✓ Python scripting APIs allow deep integration with studio pipelines, including automated shot setup, solve parameter configuration, and export to downstream tools. ✓ PFTrack Enterprise supports deployment in virtual machine (VM) and containerised environments for cloud or virtualised studio infrastructure. Enterprise deployment note: For multi-seat studio deployments on Linux, PFTrack Enterprise with PFBucket licence server is recommended. PFBucket manages floating licence distribution across your network, supports air-gapped environments, and provides centralised entitlement administration. Contact The Pixel Farm for enterprise configuration guidance. Storage Strategy for PFTrack PFTrack’s storage requirements differ depending on the workflow. Camera tracking of a single shot involves relatively modest I/O, reading a sequence of frames and writing tracking data. Photogrammetry, however, can involve hundreds or thousands of high-resolution images, generating dense point clouds and large mesh files. The right storage strategy ensures fast interactive performance across both workflows. NVMe drives provide the near-instant access required for PFTrack to ingest and solve high-resolution image sequences at peak efficiency. By eliminating I/O bottlenecks, the solver can jump between frames instantly, ensuring your tracking points stay locked without the system waiting for data to load. What Drives Storage Performance in PFTrack Footage loading: When you load a clip into PFTrack, frames are read sequentially from disk. NVMe SSD speeds (3,000–7,000+ MB/s) mean footage loads almost instantly and scrubbing through the timeline is fluid. On a spinning hard drive or slow external storage, loading a 4K DPX sequence can become a bottleneck that interrupts your tracking workflow. Photogrammetry image sets: The Photo Survey node reads potentially hundreds of high-resolution images during feature matching and reconstruction. Fast random-access reads from an NVMe SSD dramatically reduce the time PFTrack spends loading images, particularly during the iterative refinement stages where the same images are accessed multiple times. Point cloud and mesh data: Dense point clouds from photogrammetry or LiDAR can be several gigabytes. Loading and saving this data benefits from fast sequential I/O. The viewport also streams point cloud data from disk when the dataset exceeds available RAM, so NVMe speeds directly affect interactive responsiveness. Project files and exports: PFTrack project files are relatively small (kilobytes to megabytes), as they store node graph configuration, tracking parameters, and solve results — not the source footage itself. Saving and loading projects is effectively instantaneous on any modern SSD. Export operations (writing camera data, meshes, and USD/FBX/Alembic files) are also lightweight. Recommended Storage Configurations Workflow Primary Drive (NVMe) Bulk / Archive Storage Approx. Cost Matchmoving only, HD/2K footage 512 GB–1 TB NVMe SSD External USB-C drive for archived projects ~£60–£120 Regular tracking + moderate photogrammetry 1–2 TB NVMe SSD 4–8 TB external SSD or NAS ~£150–£400 Heavy photogrammetry, 4K/8K footage, LiDAR 2–4 TB NVMe SSD NAS over 10GbE or large external SSD array ~£400–£1,000+ Practical tip: Keep your active project footage and image sets on the fast NVMe drive. Move completed projects to bulk storage (external drive or NAS) when done. PFTrack’s project files are tiny, so the NVMe capacity you need is determined by your source media, not by PFTrack itself. Mac Mini Pro note: The M4 Pro Mac Mini includes an extremely fast internal NVMe SSD (6,000–7,000+ MB/s sequential reads) plus Thunderbolt 5 and USB-C for external storage. It’s an excellent PFTrack workstation with straightforward storage expansion. PFTrack Editions and Licensing PFTrack is available in three editions, each designed for a different scale of deployment. All three editions use the same core tracking engine and run on the same hardware, the differences are in licensing model, pipeline integration features, and support. PFTrack Solo PFTrack Studio PFTrack Enterprise Target User Individual artists and freelancers Teams and small studios Large studios and organisations Toolset Core 3D tracking and matchmoving Full toolset including photogrammetry, image modelling, scene reconstruction Everything in Studio + extended Python APIs, CLI, VM support Licence Type Single-user, perpetual Floating, perpetual or rent-to-buy Floating network via PFBucket (rental or permanent + maintenance) Offline / Air-Gapped No No Fully supported Pipeline Integration Standard export formats Python API, command-line operation Extended Python APIs, macros, custom automation, VM support Support Community + in-app AI assistant Community + optional ticketed support Dedicated liaison + enterprise SLAs Free Trial Free trial — full toolset with exports enabled for a limited period Limited trial mode — full toolset, exports disabled Contact sales for evaluation Hardware choice is independent of edition. PFTrack Solo, Studio, and Enterprise all run the same tracking engine on the same hardware. Choose your edition based on licensing needs and pipeline requirements, not hardware. The configurations in this guide apply equally to all three. Multi-Seat Studio Deployment For studios deploying PFTrack across multiple seats, PFTrack Enterprise with PFBucket licence server provides the flexibility and administrative control required for professional production environments. Reference Deployment: 4–8 Seat VFX Studio A typical mid-size VFX studio running PFTrack for matchmoving and photogrammetry alongside other DCC tools: ✓ Workstations: 4–8 Linux or Windows workstations, each with a high-single-core CPU (Intel i9 or AMD Ryzen 9), 64 GB RAM, and an NVIDIA RTX 4070 Super or equivalent ✓ Shared storage: NAS (Synology, QNAP, or TrueNAS) over 10GbE, providing shared footage access and project file storage across all seats ✓ PFBucket licence server: Deployed on a lightweight server or VM on the same network, managing floating licences across all workstations. Operators can check out a licence when they need PFTrack and release it when moving to other work ✓ Batch processing: PFTrack’s CLI enables headless batch jobs — tracking solves and photogrammetry can be dispatched to idle workstations overnight or to a dedicated processing node ✓ Pipeline integration: Python APIs and CLI tools integrate PFTrack into shot management systems, enabling automated project setup, solve submission, and export to downstream tools Network requirements: 10GbE is recommended for shared footage access. PFBucket licence traffic is minimal (a few kilobytes per licence check-out) and works over any network. For air-gapped environments, PFBucket operates entirely on the local network with no external connectivity required. Manage license distribution securely with PFBucket. Deployable either locally or on cloud-hosted infrastructure, it supports multi-seat pipelines and virtualized environments. Scaling Up: Virtual Production and Forensic Facilities For larger deployments serving virtual production stages, forensic analysis teams, or architectural visualisation departments, the same architecture scales. PFBucket supports multi-site licence distribution, allowing seats across different physical locations to share a common licence pool. Contact The Pixel Farm for guidance on enterprise deployments exceeding 8 seats. PFTrack Support & Resources PFTrack is backed by support and learning resources matched to your licence tier. All users have access to community and self-service resources; Studio and Enterprise customers receive additional direct support from The Pixel Farm. All Users ✓ PFTrack User Group — community forum for peer support and interaction with The Pixel Farm’s product specialists (www.pftrack.com) ✓ Learning Articles — technical articles on sensor data, lens distortion, tracking techniques, and photogrammetry workflows (www.pftrack.com/learning-articles) ✓ Tutorials — step-by-step video tutorials for specific tracking and reconstruction tools (www.pftrack.com/pftrack-tutorials) ✓ Resources — export scripts, workflow guides, and pipeline integration documentation (www.pftrack.com/resources) ✓ In-app AI assistant — context-aware guidance available directly within PFTrack ✓ PFTrack Documentation — comprehensive product and licensing documentation (pftrack.thepixelfarm.co.uk/documentation) Solo (Personal) Accounts ✓ Everything above, plus a perpetual licence with all software updates included for the life of the product — no ongoing subscription, no maintenance fees, no expiry Studio Accounts ✓ Everything above, plus optional direct IM ticketed support for private contact with The Pixel Farm’s support team from within the application ✓ Software upgrades included for the life of the product under perpetual purchase. Rent-to-buy licences include upgrades for the duration of the rental period Enterprise Accounts ✓ Dedicated technical liaison, a named contact who handles onboarding, PFBucket configuration, pipeline integration, and ongoing support ✓ Direct in-app IM support for all operators and licence administrators ✓ Technical support covering installation, PFBucket deployment, multi-site configuration, VM environments, and bug reporting ✓ Custom maintenance contracts with priority issue resolution and accelerated software updates ✓ Onboarding and integration assistance, pipeline setup, batch processing configuration, and integration with Maya, Nuke, Unreal Engine, and other DCC tools For enterprise enquiries, including bespoke support packages, volume licensing, and deployment planning, contact The Pixel Farm directly at sales@thepixelfarm.co.uk or visit www.pftrack.com. Quick-Start Recommendation If you are setting up PFTrack for the first time and want a single best recommendation for each use case: Solo matchmove artist: MacBook Pro 14" with M5 Pro (24 GB, 1 TB SSD). From ~£2,199. A portable, silent, production-capable tracking workstation with 30% faster solves than the previous generation. Download PFTrack Solo’s free trial from www.pftrack.com and start tracking immediately — full export included. Studio workstation: Windows or Linux workstation with AMD Ryzen 9 7950X, 64 GB DDR5, NVIDIA RTX 4070 Super, and 2 TB NVMe SSD. Approximately £2,000–£2,500 built. Handles everything from daily matchmoving to regular photogrammetry. Apple desktop: Mac Mini Pro with M4 Pro (24 GB, 1 TB SSD). From approximately £1,799. Compact, quiet, and powerful enough for professional tracking and moderate photogrammetry. Pair with any colour-accurate monitor. For those seeking an Apple-based setup, the Mac Mini Pro delivers elite PFTrack performance in a minimal footprint. It’s a versatile "plug-and-play" solution that leverages M-series speed for near-instant tracking and fluid interactivity. Try PFTrack Solo Free Try PFTrack Solo free for 7 days, with full export functionality, enough time to take a real plate from track through Hero Cloud and into Postshot, USD, or your DCC of choice.

Overview The new Hero Cloud node in PFTrack Version 26.05.19 generates high-quality dense point clouds from a single tracked camera move, with no LiDAR or additional capture hardware required. Build rich 3D environments directly from your existing camera solves, then export to your 3D package of choice or train Gaussian Splats in Postshot. Get Started Follow along with these tutorials using either your own footage or our example plate. Already using PFTrack? Hero Cloud ships in the latest PFTrack build, download from your account and the new node will appear in the Geometry category. New to PFTrack? Try PFTrack Solo free for 7 days, with full export functionality, enough time to take a real plate from track through Hero Cloud and into Postshot, USD, or your DCC of choice. Start Free Trial Want to follow along with the exact tutorial workflow? Download the example footage used in these tutorials and try the same workflow shown in the videos. Download Example Footage Using Hero Cloud What You’ll Learn Generate dense point clouds from any solved camera Control depth range and frame sampling Optimise cloud density for your workflow Clean up and refine your point cloud data Workflow Summary Start with your solved camera — Hero Cloud works with any camera track and solve in PFTrack. Complete your camera tracking workflow as normal before adding the Hero Cloud node to your tree. Add the Hero Cloud node — Simply add the Hero Cloud node to your node tree. Define your depth range — Use the frustum visualisation in the Cinema view to set your near and far clipping planes. Focus on your area of interest by adjusting the far clipping plane, distant points often lack the detail needed for accurate depth reconstruction. Tightening the frustum to your hero elements ensures computational resources are focused where they matter most. The frustum can be animated throughout your shot for dynamic depth control as your subject or camera moves. Adjust frame density — Control how many frames are used to generate depth maps using the Frame Density slider. The Shot Coverage graph updates in real-time, providing a dynamic estimate of depth map coverage for your chosen frame count. Set cloud density — Balance detail against performance by adjusting the Hero Cloud density. Complex scenes may benefit from lower point counts to maintain viewport responsiveness in your pipeline later on. Build your Hero Cloud — Click Build to automatically generate your dense point cloud. Once processing completes, the actual coverage line appears in the Shot Coverage graph, showing real-world depth data distribution. Refine with point cloud editing tools — Edit your dense point cloud using PFTrack’s comprehensive toolkit: Paint Tool: directly paint to remove unwanted points in the Cinema or 3D Viewer Selection Tools: use lasso or box selection for precise control Select Colours: key specific colour ranges to isolate and delete points by hue Direct Manipulation: work in either Cinema or 3D Viewer for maximum flexibility Exporting to USD for 3D Packages What You’ll Learn Export Hero Cloud data via USD Import point clouds into Blender Assign colours for render-ready assets Export Workflow Add a Scene Export node — Connect a Scene Export node to your Hero Cloud node in the tree. Configure USD export — Select USD as your export format. USD (Universal Scene Description) provides compatibility with all major 3D packages including Maya, Houdini, Blender, and Unreal Engine. Export your scene — Set your export path and click Export. Your solved camera, track points, and dense Hero Cloud will be packaged together in a single USD file. Importing to Blender Import USD file — In Blender, use File > Import > Universal Scene Description (.usd/.usdc) to bring in your PFTrack scene. Locate your point cloud — The Hero Cloud will import as a point cloud object in your scene hierarchy, maintaining world-space positioning relative to your camera. Assign vertex colours — The point cloud retains colour information from your original footage. In Blender’s Shading workspace, connect the Color Attribute node to your material to display photorealistic vertex colours from your capture. Use Cases Pre-light your shot against the actual scene geometry, rather than blocking against approximations Environmental blocking for set extensions and CG integration Spatial reference for matte painting and final composition Collision geometry generation for simulation and animation passes Exporting to Postshot for Gaussian Splats What You’ll Learn Export optimised data for Gaussian Splat training Import into Jawset Postshot Train production-ready Gaussian Splats Postshot Export Workflow Select Postshot export — In your Scene Export node, choose the Jawset Postshot export option. This format is optimised specifically for Gaussian Splat training workflows. Configure export settings — Set your export path and any additional parameters. The export includes your solved camera, Hero Cloud point data, and frame metadata required for training. Export your data — Click Export to generate your Postshot-compatible dataset. Training in Postshot Import to Postshot — Launch Postshot and import your PFTrack export. The software will automatically recognise the camera solve and point cloud data. Configure training parameters — Set your Gaussian Splat training parameters based on scene complexity and desired output quality. The dense Hero Cloud provides strong initialisation data, helping training converge faster than from sparse point sets alone. Train your Gaussian Splat — Begin training. The rich point cloud data from Hero Cloud improves convergence speed compared to training from sparse initialisations alone. Render and export — Once training completes, render your Gaussian Splat for real-time playback, interactive viewing, or export to game engines and real-time platforms. Why Hero Cloud for Gaussian Splats? Most Gaussian Splat workflows today require either dedicated photo sets or LiDAR scans to initialise. Hero Cloud generates that initialisation data from production footage you’ve already shot. The dense point cloud provides: Strong initialisation data, helping training converge faster than from sparse point sets alone Improved spatial accuracy in areas with limited visual features Consistent coverage across your entire frustum No additional capture requirements beyond your existing footage Key Benefits Through-the-lens generation Hero Cloud generates dense 3D data from your camera’s viewpoint — the same perspective you’re already working with. No additional hardware, multi-camera rigs, or LiDAR scanning required. Real-time feedback The Shot Coverage graph updates dynamically as you adjust parameters, giving you instant feedback on depth map distribution and frame sampling efficiency. Flexible export pipeline Whether you’re building CG layouts in traditional 3D packages or training Gaussian Splats, Hero Cloud integrates seamlessly into your existing pipeline. Production-ready results From blocking and layout to final Gaussian Splat renders, Hero Cloud delivers professional-quality point clouds optimised for real-world production workflows. Learn more Read the full Hero Cloud announcement PFTrack version 26.05.19 release notes Hero Cloud documentation Try PFTrack Solo Free Try PFTrack Solo free for 7 days, with full export functionality, enough time to take a real plate from track through Hero Cloud and into Postshot, USD, or your DCC of choice.

A substantial release. New Hero Cloud node, direct export to Postshot for Gaussian splat training, USD export with dense point clouds, Photo Mesh integration with Hero Cloud, 7-day full-export trial mode, and a number of fixes. Available now from your PFTrack account. Overview PFTrack 26.05.19 is one of the more substantial releases we’ve shipped in some time. The headline addition is Hero Cloud, a new node that generates a measured 3D point cloud directly from a single tracked plate. The release also extends PFTrack’s export pipeline with a direct export to Postshot in COLMAP format for Gaussian splat training, USD export with dense point clouds plus per-point normals and colour, and Photo Mesh integration with the new Hero Cloud node (Studio and Enterprise editions). The trial mode has been extended too: new users on the Solo trial now get full export functionality for seven days, which means a trial user can take a real plate from track all the way through Hero Cloud and into Postshot or USD. Full changelog below. All changes are in 26.05.19 unless otherwise noted. New Features Hero Cloud node (all editions) A new node in the Geometry category. Hero Cloud builds a dense 3D point cloud from a solved camera. Run it on any plate that’s been tracked and solved — the node will reconstruct depth from the camera’s movement through the scene and produce a measured point cloud ready for downstream use. Hero Cloud works with any footage PFTrack can track: cinema cameras, drones, body-cams, dash-cams, action cameras, mirrorless and DSLR, CCTV, archival film. The shot needs parallax in the camera move — most production footage has enough. Available in every PFTrack edition (Solo, Studio, Enterprise). Full documentation and a hands-on tutorial are linked at the end of this article. Postshot export node (all editions) A new export option in the Scene Export node. Writes Hero Cloud point clouds and the associated tracked cameras out in COLMAP format — the standard input for Postshot and the wider Gaussian splat training ecosystem. Track the plate, run Hero Cloud, export to Postshot, train your splat. The whole flow stays inside artist-friendly tools. USD export with dense point clouds (all editions) USD export now supports dense point clouds with per-point normal and colour data. Cleaner integration with Unreal Engine, Maya, Houdini, and the rest of the USD-aware ecosystem. The point cloud carries directional and visual information into downstream tools rather than just spatial coordinates. Active Camera viewport (all editions) Viewer windows now have an Active Camera viewport with an option to snap the camera pose to the active clip. Useful for verifying Hero Cloud reconstruction visually against the source plate, but it has wider workflow value too — anywhere you want a quick “see what the tracked camera saw” view. Trial mode now full export for 7 days (Solo trial) New users on the Solo trial can now export at full functionality for seven days, rather than only being able to evaluate the toolset. A trial user can now take a real plate all the way from track through Hero Cloud and out to Postshot, USD, FBX, or any of PFTrack’s supported formats. Updates and Improvements Photo Mesh integration with Hero Cloud (Studio and Enterprise) Photo Mesh has been updated to consume Hero Cloud point clouds as input. For Studio and Enterprise users who need a finished textured mesh from a single shot, the path now exists inside PFTrack — Hero Cloud feeds the cloud, Photo Mesh surfaces it. No external tools required. FBX export for large dense point clouds (all editions) FBX export now handles very large dense point clouds more reliably. If you’ve hit issues exporting big clouds via FBX in earlier builds, this build should resolve them. Fixes Spherical tracking node connection issues (Studio and Enterprise) Several node connection issues in the spherical tracking toolset have been resolved. If you’ve hit unexpected behaviour wiring up Spherical Track or Spherical Orient, this build should clean things up. Multi-sample motion blur on retimed footage (all editions) Some rendering issues when applying multi-sample motion blur through the Retime node have been fixed. Results should now be more predictable on retimed plates. Point Cloud node keyboard shortcuts (Studio and Enterprise) Some missing keyboard shortcuts in the Point Cloud node have been restored. Other UI updates and fixes (all editions) Various smaller fixes and UI updates across the application. If you’ve had a specific niggle that’s been bothering you, there’s a fair chance it’s addressed here. How to Get It Existing customers can download PFTrack 26.05.19 directly from your PFTrack account. Sign in, navigate to your downloads, grab the latest build. New users can try PFTrack Solo free for seven days with the new full-export trial — enough time to take a real plate from track all the way through Hero Cloud and into your pipeline. Studio is available on rent-to-buy for 5 or 30 days, or as a perpetual purchase. Learn More Hero Cloud announcement Hero Cloud hands-on tutorial PFTrack documentation Feedback Hero Cloud is the first piece of a wider through-the-lens reconstruction toolset. We’re curious how it works in real production use — where it succeeds, where it doesn’t, and what edge cases we haven’t yet seen. The reconstruction techniques behind it have considerable headroom, and your feedback shapes what we ship next. Reply via the community forum, or by getting in touch directly. All routes get back to us. Try PFTrack Solo Free Try PFTrack Solo free for 7 days, with full export functionality, enough time to take a real plate from track through Hero Cloud and into Postshot, USD, or your DCC of choice.

A new node in PFTrack generates a measured 3D point cloud directly from a single tracked shot, no reference photography, no LiDAR scan, no second capture required. Export to Gaussian splat pipelines, surface to mesh, or use as a spatial scaffold in any DCC. Available now in Solo, Studio, and Enterprise. What's new Hero Cloud is shipping today in PFTrack. It joins the existing PFTrack node graph as a new reconstruction tool, sitting alongside Camera Solver, Photo Mesh, and the rest of the toolset. Hero Cloud takes a tracked plate, the same plate you’ve already solved with PFTrack’s Camera Solver, and generates a measured 3D point cloud of the scene. It uses through-the-lens reconstruction, a technique that extracts spatial information directly from the moving camera in the shot. No separate photogrammetry set is required. No LiDAR scanner. No additional capture of any kind. It’s available in every PFTrack edition: Solo, Studio, and Enterprise. The node will appear in the next build. When to use it Hero Cloud is built for a situation every matchmove and reconstruction team has encountered: a brief that needs measured 3D data of the scene — for set extension, CG integration, lighting reference, forensic analysis, or architectural measurement — and no reference photography was captured on set. The shoot wrapped before the second-unit crew got to it. The LiDAR scanner couldn’t make it to the location. The brief evolved after the shoot. Or the source material is archival, and there is no on-set crew to send back. In each of those cases, the plate itself is the only spatial record of the scene. Hero Cloud uses what’s already there. Typical use cases: VFX matchmove and set extension where reference photography wasn’t shot or wasn’t usable Forensic reconstruction from CCTV, body-cam, dash-cam, or witness footage Heritage and archaeological documentation from archival film or video Architectural reference from drone footage or walk-through video Source data for Gaussian splat training pipelines Any project where the only available record of a scene is the footage itself How it fits with the rest of the toolset Hero Cloud doesn’t replace PFTrack’s existing reality-capture tools — it extends them. Use it alongside the tools you already know: If you have a dedicated photogrammetry set: Photo Mesh (Studio and Enterprise) remains the highest-fidelity reconstruction path, producing a textured mesh from a planned photoset. If you have LiDAR data: PFTrack’s LiDAR integration (Studio and Enterprise) gives you precise survey-grade point clouds aligned to your tracked cameras. If you have only the tracked plate: Hero Cloud reconstructs spatial data from what’s in the shot itself, the path forward when no dedicated capture was performed. The point of through-the-lens reconstruction is to give you a way forward when dedicated capture wasn’t an option, not to replace dedicated capture when it was. Hero Cloud reconstructs spatial data from what’s in the shot itself, the path forward when no dedicated capture was performed. What you can do with the output Hero Cloud produces a measured 3D point cloud — a set of accurately positioned 3D points representing the surfaces visible in the tracked plate. The release includes new export options that open three concrete downstream paths: 1. Train a Gaussian splat with the new Postshot export This release introduces a new Postshot export node that writes Hero Cloud point clouds and the associated tracked cameras out in COLMAP format — the standard input format for Postshot and the wider Gaussian splat training ecosystem. For artists working with Postshot or other COLMAP-compatible splat trainers, PFTrack now sits naturally at the front of that pipeline. Track your plate, run Hero Cloud, export to COLMAP, train a splat in Postshot. The artist-friendly route into a workflow that has historically required command-line tools. 2. Surface to a textured mesh with Photo Mesh Photo Mesh has been updated to work with Hero Cloud point clouds (Studio and Enterprise editions). For users who need a finished textured mesh from a single shot, Photo Mesh can now take a Hero Cloud output as input and produce a surfaced reconstruction — ready for export to Maya, Nuke, Unreal Engine, USD, and the rest of PFTrack’s supported formats. This gives Studio and Enterprise users a complete single-shot path from tracked plate to textured mesh, all within PFTrack. 3. Use as a spatial scaffold in any DCC This release also extends USD export to support dense point clouds with per-point normal and colour data, and FBX export has been improved to handle very large dense point clouds reliably. Export Hero Cloud output to USD, FBX, OBJ, Alembic, or PLY, and use it directly in Maya, Houdini, Blender, Nuke, or any DCC — as a measurement scaffold for modelling, a volumetric guide for placing CG elements, or a spatial reference for set extension and lighting. Whichever path you take, Hero Cloud is the front of the workflow. The choice of downstream tool depends on what you need to deliver. How to use it Hero Cloud is a single node, designed to drop into your existing tracking workflow without restructuring your graph. The basic process: Track and solve your hero plate as you normally would, using Auto Track or User Track followed by Camera Solver. Add a Hero Cloud node downstream of the solved camera. Connect the tracked footage and solved camera into the Hero Cloud node’s inputs. Run the node. Hero Cloud generates a 3D point cloud of the scene, viewable in the Cinema window and ready for downstream use. Export via Scene Export (USD, FBX, OBJ, PLY, Alembic), the new Postshot export (COLMAP for splat training), or feed the output into Photo Mesh for surfaced mesh generation. New in this release: Viewer windows now have an “Active Camera” viewport with an option to snap the camera pose to the active clip — a useful aid for visually verifying your Hero Cloud output against the tracked plate. The full node reference, including parameter details and quality controls, is in the PFTrack documentation. What it requires Hero Cloud works with any plate that PFTrack can track and solve, including footage from cinema cameras, broadcast cameras, drones, body-cams, dash-cams, action cameras, mirrorless and DSLR cameras, and CCTV. Camera metadata is helpful but not required, the Camera Sensor Database covers most common cameras automatically, and Camera Solver can estimate parameters where metadata is unavailable. Good camera motion will provide rich & accurate results. The shot does need parallax. Hero Cloud reconstructs depth from the camera’s movement through the scene, so locked-off shots and shots with very little camera motion will produce sparse or unreliable results. Most production footage, even handheld stationary shots — has enough small camera movement to produce a usable reconstruction. Shots with deliberate camera motion produce richer results. Performance Hero Cloud typically processes in minutes per shot on modern hardware. The node benefits from GPU acceleration; refer to the PFTrack Hardware Guide for recommended configurations. Processing runs entirely locally, no cloud processing, no data leaves your workstation. Looking ahead Hero Cloud is the first of a wider set of through-the-lens reconstruction tools coming to PFTrack. Future PFTrack releases will extend the workflow with additional reconstruction nodes that build on the same foundation. Watch this space for further announcements. Get started Hero Cloud ships in the current build of PFTrack across all editions. To start using it: Download the latest build Existing PFTrack users: Download the latest build. Hero Cloud will be available in the Geometry node category. New to PFTrack? New users: The trial mode now operates with full export functionality for seven days, enough time to track a plate, run Hero Cloud, and take the output through to Postshot, Photo Mesh, or your DCC of choice. Learn More PFTrack version 26.05.19 release notes Hero Cloud hands-on tutorial PFTrack documentation Try PFTrack Solo Free Try PFTrack Solo free for 7 days, with full export functionality, enough time to take a real plate from track through Hero Cloud and into Postshot, USD, or your DCC of choice.

Build 26.02.11  is now live for Solo, Studio, and Enterprise . This update boosts pipeline reliability, incorporating key refinements driven by our global user community. Key Enhancements: Maya 2026 Export Script: Updated export scripts ensure background image planes are perfectly organized upon import. More information here . Enhanced LiDAR Reliability:  Advanced logging now identifies corrupted or incomplete survey files instantly. Robust Asset Support:  Consistent loading for OBJ mesh textures with numbered filenames within the Survey Solver. Expanded Node Logic:  Optimised connectivity for photogrammetry nodes and secondary Survey Solver inputs (Studio/Enterprise). Refined Navigation: Improved UI responsiveness with resolved shortcut conflicts for the Set-Axis tool. Solo  users download the latest version of PFTrack directly from within the application. Studio  users download the software from within the St udio Account . Enterprise  customers login to the PFAccount Portal   and download tghe software from there. Join the Conversation:  Help us build the next generation of tracking tools by joining our PFTrack Support Community . We’ve also introduced a new PFTrack   Solo Trial Mode . Whether you are on an older build or brand new to the ecosystem, we encourage you to download  the latest version and explore the industry's leading tracking tools firsthand.

What does it do? We have added a new export scripts for Autodesk Maya that improved compatibility by automatically parenting the camera image plane to the moving camera. Just unzip them into: Documents/The Pixel Farm/PFTrack/exports Contents: mayaASCII-2026-zxy.py mayaASCII-2026.py Once installed, choose “Maya 2026” as the export format. These scripts should work with earlier versions of Maya as well. Downloads: Links: Head back to PFTrack Resources. Or check out our Learning Articles for a deeper look at camera tracking and matchmoving concepts. Or visit our PFTrack Tutorials for step-by-step video guides covering the fundamentals of camera tracking and matchmoving in PFTrack.

Building on the internal Camera Library introduced in our last major release, we are excited to introduce a powerful new capability: we are opening up our Sensor Preset Database to third parties. This feature extends our existing library architecture, allowing third-party experts to integrate their own precision sensor data directly into PFTrack. To launch this, we are proud to feature Matchmove Machines CamDB as the first available integration. Why add Matchmove Machines to your library? Enhanced Precision: Sensor sizes are meticulously recalculated from original manufacturer data to eliminate rounding errors. Workflow Efficiency: Instantly access a vast, verified catalog of modern camera bodies and modes without leaving PFTrack. Seamless Extension: Instantly expands your native PFTrack library with specialized third-party data. Upgrade your sensor accuracy today. Learn more about Matchmove Machines CamDB Happy Tracking, The PFTrack Team

#postshot   #exportscripts   What does it do? This is an experimental export script for exporting cameras and points from PFTrack 24.12.19 and later to Jawset Postshot for Gaussian Splat training https://www.jawset.com/ Before using the script, please review the usage guidelines below to get the best results. Download: Shot Setup The script can export movie or photogrammetry cameras, along with tracking points and point clouds. Download and unzip the file into your Documents/The Pixel Farm/PFTrack/exports folder and relaunch PFTrack. This will create a new export format in the Scene Export node called "Jawset Postshot (.json)" When setting up your cameras in PFTrack, it is important to ensure you are correcting for lens distortion. Gaussian Splatting is initialised from your point dataset and trains a radiance field to match your image data, so the results you get will depend strongly on the quality of your input cameras and points. A sparse set of points may not initialise the training as well as a dense point cloud, and datasets with low parallax or coverage may not give the best results when viewed from angles other than your original cameras, so please refer to the the Postshot user guide for capturing guidelines: https://www.jawset.com/docs/d/Postshot+User+Guide Point density You should make sure your shot has enough tracking points to initialise the Postshot training. If you've just used a few User Track points or a small number of Auto Track points, you will probably get better results by adding some more to your shot. You can do this easily by placing an empty Auto Track node upstream from your Camera Solver before solving. Then, after you've solved your camera and are happy with the result, go back to your Auto Track node and generate more tracking points, increasing the Target Number up to 500 or more. In your camera solver, select all your tracking points and click the Solve Trackers button to solve for their 3D positions whilst keeping the camera fixed. Alternatively, you can use the Select Frames node to decimate your movie clip into a set of photos, and then use the Photo Cloud node to create a dense point cloud, and attach both the camera and dense point cloud to the export node. The Select Frames node could also be used before exporting your movie camera to reduce the number of frames being loaded into Postshot if you have a very long image sequence. Exporting from PFTrack Select the "Jawset Postshot (.json)" export format and make sure to enable the "Undistorted clip" Distortion export option, setting the image format to either JPEG, TIFF or OpenEXR with suitable frame number padding. We recommend using TIFF or OpenEXR as this will ensure invalid pixels around the boundary of your undistorted images are written with a zero in the alpha channel and ignored by Postshot. Alternatively, make sure your undistorted images are cropped to the original image size during the solve to reduce empty pixels as much as possible. After exporting, you will find a .json file and a .ply file in the export folder containing your camera and point data respectively, along with your undistorted images in the clips folder. Importing into Postshot You can drag-and-drop the entire export folder directly into Postshot, but it is important to ensure no other files are present in the folder. Macos users in particular should remove all .DS_Store files that are created when opening the folder in Finder, as they will prevent the dataset from loading and give an "invalid string position" error message. In Postshot, make sure to enable the "Treat Zero Alpha as Mask" option to ensure the boundary pixels in the undistorted images are ignored during training. Please refer to the Postshot user guide for all other settings. Links: Head back to PFTrack Resources . Or check out our Learning Articles  for a deeper look at camera tracking and matchmoving concepts. Or visit our PFTrack Tutorials  for step-by-step video guides covering the fundamentals of camera tracking and matchmoving in PFTrack.

What does it do? This script is for PFTrack 24.12.19 and later, and can be used to transfer the motion from a moving object geometry track to a camera, keeping the object locked in position in the first frame. To use the script, download and unzip the file into your Documents/The Pixel Farm/PFTrack/nodes folder and relaunch PFTrack. This will create a new node called Lock Object Motion in the Python node category. Download: Script: # # PFTrack python script lockObjectMotion.py # # Takes object motion from a geometry track and converts to camera # motion, keeping the object locked in position in the first frame # import math import pfpy def pfNodeName(): return 'Lock object motion' def quaternionInverse(q): return (-q[0],-q[1],-q[2],q[3]) def quaternionNormalize(q): n = 1.0/math.sqrt(q[0]*q[0]+q[1]*q[1]+q[2]*q[2]+q[3]*q[3]) return (n*q[0],n*q[1],n*q[2],n*q[3]) def quaternionMult(a, b): return (a[3]*b[0]+a[0]*b[3]+a[1]*b[2]-a[2]*b[1], a[3]*b[1]-a[0]*b[2]+a[1]*b[3]+a[2]*b[0], a[3]*b[2]+a[0]*b[1]-a[1]*b[0]+a[2]*b[3], a[3]*b[3]-a[0]*b[0]-a[1]*b[1]-a[2]*b[2]) def quaternionToMatrix(q): return (1.0-2.0*(q[1]*q[1]+q[2]*q[2]), 2.0*(q[0]*q[1]-q[3]*q[2]), 2.0*(q[0]*q[2]+q[3]*q[1]), 2.0*(q[0]*q[1]+q[3]*q[2]), 1.0-2.0*(q[0]*q[0]+q[2]*q[2]), 2.0*(q[1]*q[2]-q[3]*q[0]), 2.0*(q[0]*q[2]-q[3]*q[1]), 2.0*(q[1]*q[2]+q[3]*q[0]), 1.0-2.0*(q[0]*q[0]+q[1]*q[1])) def matrixMult4x4(a, b): return (a[0]*b[0]+a[1]*b[4]+a[2]*b[8]+a[3]*b[12], a[0]*b[1]+a[1]*b[5]+a[2]*b[9]+a[3]*b[13], a[0]*b[2]+a[1]*b[6]+a[2]*b[10]+a[3]*b[14], a[0]*b[3]+a[1]*b[7]+a[2]*b[11]+a[3]*b[15], a[4]*b[0]+a[5]*b[4]+a[6]*b[8]+a[7]*b[12], a[4]*b[1]+a[5]*b[5]+a[6]*b[9]+a[7]*b[13], a[4]*b[2]+a[5]*b[6]+a[6]*b[10]+a[7]*b[14], a[4]*b[3]+a[5]*b[7]+a[6]*b[11]+a[7]*b[15], a[8]*b[0]+a[9]*b[4]+a[10]*b[8]+a[11]*b[12], a[8]*b[1]+a[9]*b[5]+a[10]*b[9]+a[11]*b[13], a[8]*b[2]+a[9]*b[6]+a[10]*b[10]+a[11]*b[14], a[8]*b[3]+a[9]*b[7]+a[10]*b[11]+a[11]*b[15], a[12]*b[0]+a[13]*b[4]+a[14]*b[8]+a[15]*b[12], a[12]*b[1]+a[13]*b[5]+a[14]*b[9]+a[15]*b[13], a[12]*b[2]+a[13]*b[6]+a[14]*b[10]+a[15]*b[14], a[12]*b[3]+a[13]*b[7]+a[14]*b[11]+a[15]*b[15]) def vectorMult4x4(m, v): n = 1.0/(v[0]*m[12]+v[1]*m[13]+v[2]*m[14]+m[15]) return ((v[0]*m[0]+v[1]*m[1]+v[2]*m[2]+m[3])*n, (v[0]*m[4]+v[1]*m[5]+v[2]*m[6]+m[7])*n, (v[0]*m[8]+v[1]*m[9]+v[2]*m[10]+m[11])*n) def buildTransformationMatrix(t, q): T = (1.0,0.0,0.0,t[0], 0.0,1.0,0.0,t[1], 0.0,0.0,1.0,t[2], 0.0,0.0,0.0,1.0) r = quaternionToMatrix(quaternionInverse(q)) R = (r[0],r[1],r[2],0.0, r[3],r[4],r[5],0.0, r[6],r[7],r[8],0.0, 0.0,0.0,0.0,1.0) return matrixMult4x4(T,R) def buildInverseTransformationMatrix(t, q): T = (1.0,0.0,0.0,-t[0], 0.0,1.0,0.0,-t[1], 0.0,0.0,1.0,-t[2], 0.0,0.0,0.0,1.0) r = quaternionToMatrix(q) R = (r[0],r[1],r[2],0.0, r[3],r[4],r[5],0.0, r[6],r[7],r[8],0.0, 0.0,0.0,0.0,1.0) return matrixMult4x4(R,T) def main(): if pfpy.getNumCameras() > 0 and pfpy.getNumMeshes() > 0 : # fetch the first camera and mesh cam0 = pfpy.getCameraRef(0) mesh0 = pfpy.getMeshRef(0) inp = cam0.getInPoint() outp = cam0.getOutPoint() # take copies to read from safely c = cam0.copy() m = mesh0.copy() # keep the camera in position in the first frame, but transfer the relative object motion in other frames to the camera objT0 = m.getTranslation(inp) objQ0 = m.getQuaternionRotation(inp) objM0 = buildTransformationMatrix(objT0, objQ0) f= inp+1 while (f <= outp) : # object pose in this frame objT = m.getTranslation(f) objQ = m.getQuaternionRotation(f) objiM = buildInverseTransformationMatrix(objT, objQ) # map the relative camera position back to the object in the first frame t = vectorMult4x4(objM0, vectorMult4x4(objiM, c.getTranslation(f))) # and likewise for the camera rotation q = quaternionNormalize(quaternionMult(c.getQuaternionRotation(f), quaternionMult(quaternionInverse(objQ), objQ0))) # position the camera cam0.setTranslation(f, t) cam0.setQuaternionRotation(f, q) # object is no longer moving mesh0.setTranslation(f, objT0) mesh0.setQuaternionRotation(f, objQ0) print('Positioned camera in frame %d'%f) f += 1 # cleanup c.freeCopy() m.freeCopy() Links: Head back to PFTrack Resources . Or check out our Learning Articles  for a deeper look at camera tracking and matchmoving concepts. Or visit our PFTrack Tutorials  for step-by-step video guides covering the fundamentals of camera tracking and matchmoving in PFTrack.

Navigate the learning article What is Sensor Size? How do I find out the size of a camera sensor? Why is ‘Sensor Size’ important for camera tracking? Sensor Size Considerations Loose terminology Full Area Vs Active Area Windowed Sensor Mode Scaled Sensor Mode Metadata and sensor size A multiformat sensor? Full-frame equivalent Intro What exactly is sensor size, and why does it matter for VFX, particularly when calculating the field of view for camera tracking? This post explores the intricacies of sensor size. We’ll demystify key terminology, including "full-frame equivalent," "windowed," and "cropped" sensors. We'll also examine how metadata and other resources can help you out of a tight spot if you don’t have the info. By understanding and applying these concepts, you can gain greater control over your camera tracking projects. What is Sensor Size? So, perhaps a good place to start is defining what we mean by sensor size. Sensor size refers to the physical dimensions of a camera's imaging sensor—the part that turns incoming light into digital images. Usually, the height and width are measured in millimetres (mm); it is where the lens projects an image to be captured and converted into a digital signal. Sensor size affects how much of the lens's image is recorded, influencing the field of view, depth of field, and overall image quality.  How do I find out the size of a camera sensor? While a quick Google search will provide the necessary information for most professional cine and mirrorless camera systems, it is worth delving deeper into the manufacturer's website to find the exact size. However, remember the considerations that will be discussed later in this article when finding out the sensor's 'actual size'. If a simple search doesn't yield results, the following links lead to excellent websites that cover the essential details you may need. Matchmove Machine matchmovemachine.com is run by a knowledgeable team experienced in all aspects of camera tracking and matchmoving. Their standout resource is a comprehensive camera sensor database, covering a wide variety of cinema, consumer, and drone cameras with detailed specs essential for accurate tracking work. The site also offers helpful guides and insights, making it a useful reference point for anyone involved in matchmoving. https://camdb.matchmovemachine.com VFX camera database  The VFX Camera Database is a valuable resource that offers an extensive, mostly up-to-date collection of professional and prosumer cameras. What sets this site apart is its inclusion of detailed measurements, not only for the full active sensor area but also for windowed sizes in different recording modes.  https://vfxcamdb.com/ DXOMARK   This website is a database primarily focused on testing camera sensor performance. However, under the specifications tab, it also provides key details like the actual sensor size and other useful data for matchmoving, such as rolling shutter performance. This resource is especially helpful for tracking phone footage, as it includes information on most of the latest phone cameras, including sensor sizes and field-of-view equivalence.  https://www.dxomark.com/Cameras/ CINED CINED is a production-focused website offering in-depth testing and reviews of the latest camera gear. While it primarily focuses on sensor performance, much like DXOMARK, it also provides valuable details on sensor size and windowing for various popular camera systems, including some phones and mirrorless cameras—making it a useful resource for camera tracking tasks. https://www.cined.com/camera-database/ Why is ‘Sensor Size’ important for camera tracking? Camera tracking applications like PFTrack require the camera's field of view (FoV) to accurately track and solve a shot. The FoV is determined by both the lens's focal length and the sensor size. However, precise knowledge of the sensor size remains crucial for ensuring the virtual camera accurately replicates the real-world camera's perspective and movement. Without this, discrepancies in scale, position, and motion can misalign digital assets, disrupting the realism of the final shot. Sensor Size Considerations Loose terminology for the sizing of an imaging sensor You’ve probably come across terms like Super35, Full Frame, and large format to describe the size of an imaging sensor in cine or still cameras. While it might seem straightforward to search " What is the size of a Super35 sensor? " and rely on the results, the information can often be inconsistent due to manufacturers' generalised and imprecise terminology. To illustrate, suppose we have a camera with a 24mm focal length, and the camera in question is a Sony PMW-F3 , which uses a "Super35" sensor. If we rely solely on Google's definition of Super35 based on the traditional film format, we might calculate the following horizontal and vertical field of view: Super35 format size: 24.89 mm x 18.66 mm Horizontal FoV: 54.82° Vertical FoV: 42.49 Delving deeper into the manufacturer's sensor specifications reveals that its size is not identical to a true Super35mm sensor; instead, it has been rounded up and features a different aspect ratio. This discrepancy impacts calculations, resulting in a field of view that is 2.46° narrower horizontally and 11.52° narrower vertically than anticipated. Actual sensor size: 23.6 mm x 13.3 mm Horizontal FoV:  52.36° Vertical FoV: 30.97° While this may not seem significant, even small sensor size discrepancies can impact your solve's overall accuracy. This is especially true for smaller sensors, such as those used in drones and cameras built into phones, where precise field of view (FoV) calculations are critical.  The term "Large Format" adds further confusion, as it has come to refer to any sensor larger than 36x24mm without clearly defined upper limits, complicating efforts to strictly define sensor size for camera tracking. The situation becomes even more complex when considering sensor crop and various windowed shoot modes.  Windowed, Scaled, and Crop Modes: Effects on FoV If you have looked up the size of the image sensor in the camera that shot your clip and entered the information, and things just don’t seem to be making sense or working, it might be because your camera is shooting in a mode that affects the FoV of your image.  Full Area Vs Active Area The difference between the active imaging area and the full sensor area lies in how much of the sensor's surface is actually used for capturing an image versus the total physical size of the sensor itself. Full Sensor Area:  This refers to the total physical dimensions of the sensor, including all of its pixels and regions, whether they are used for capturing an image or not. The full sensor area accounts for every part of the sensor's surface, including pixels reserved for other functions (such as calibration or stabilisation) or areas that may be masked out during image capture. Active Imaging Area:  This is the portion of the sensor that is actively used to capture an image. It defines the region where incoming light is collected and converted into a digital image. Due to manufacturer-specific design choices, cropping, or masking, the active imaging area can be smaller than the full sensor area. This distinction is important in applications like camera tracking, as it directly affects how the image is projected onto the sensor and impacts field-of-view calculations. When entering information about the sensor in your camera, it is always important to use the ‘Active Imaging Area’ over the ‘Full Sensor Area’ where possible for best accuracy. Windowed Sensor Mode Sensor windowing occurs when only a portion of the imaging sensor is used to capture an image, effectively "cropping" the sensor's active area. This is common in cine cameras when recording RAW and selecting a resolution or format other than the sensor's native resolution. Instead of resampling the full sensor, the camera activates a smaller portion of it, which alters the field of view. Similarly, sensor windowing is often used to achieve very high frame rates, as processing data from a smaller sensor area reduces the hardware's workload.  For instance, the RED MONSTRO 8K sensor, measuring 40.96 x 21.6 mm, utilises its full area when shooting at its maximum resolution of 8K (8192 x 4320). However, the camera applies sensor windowing to achieve lower resolutions, using only a portion of the sensor’s area.  RED MONSTRO Windowed shooting modes:  6K shooting mode (6144 x 3240), area is 30.72 x 16.20 mm 5K shooting mode (5120 x 2700), area is 25.6 x 13.5 mm 4K shooting mode (4096 x 2160), area is 20.48 x 10.80 mm Sensor windowing may also be necessary when using lenses for smaller imaging circles, such as a Super35 (31.1mm) optic on large format sensors). When using optics designed for the Super 35 format, the RED camera can use the 5K Super 35 windowed mode Scaled Sensor Mode Sensor scaling is a simpler concept compared to sensor windowing. It involves resampling the image captured by the entire sensor area to a lower resolution while preserving the full active sensor area on one or more axes.  Full Area Resampling Full-area resampling takes the image captured by the entire sensor and downsamples it to a lower resolution without altering the active area or the FoV. For example, a sensor with a native resolution of 3840x2160 might be resampled to 1920x1080, maintaining the full sensor area while reducing the pixel count.  Despite resampling from UHD to HD, the sensor's full FoV is maintained Aspect Sensor scaling can also account for changes in aspect ratio. For instance, a sensor with a native aspect ratio of 1.78:1 (16:9) may crop or scale the image at the top and bottom to produce a 1.85:1 aspect ratio while maintaining the full sensor width and horizontal FoV. You can enter the sensor's width, and PFTrack will calculate the height automatically. Changing aspect scale can also happen vertically; for example, if you have a native 1.85:1 sensor, the scaling may crop the sides to reach a 1.78:1 aspect ratio (see anamorphic). Camera with a native 1.78:1 aspect sensor with a 1.85:1 ‘scaled’ sensor mode active Anamorphic Anamorphic scaling preserves the full sensor height and vertical FoV, while the sides are scaled/cropped to achieve the desired anamorphic recording ratio, such as 1.33:1. You can enter the sensor's height, and PFTrack will calculate the width automatically.  RED VV Raptor shooting in its “Scaled” 8K 6:5 anamorphic 2X mode It's important to note that resampling a 3840x2160 resolution sensor using the full sensor area to 1920x1080 is not the same as using a 1920x1080 windowed shoot mode, where only a portion of the sensor's full area is utilised. The two methods will result in very different fields of view (FoV). Camera Sensor Database PFTrack 25.11.13 introduces a smart new camera sensor database, adding support for both online and locally-hosted options. No more hunting through specs or scouring the web, finding the right sensor size for your camera is now quick and effortless, letting you focus on the creative side of tracking. The database is built to evolve, with a growing collection of camera presets and the ability to build your own custom lists of frequently used cameras. This means your most-used setups are always at your fingertips, and new cameras can be added as your toolkit expands. By centralising sensor information and putting it directly in the app, PFTrack removes a common source of friction, helping you set up shots faster and keep your workflow running smoothly. Metadata A key advantage of using an application like PFTrack is its ability to read metadata from formats like DPX and EXR and many camera RAW files. But why is this important for determining a camera's sensor size? Metadata often contains critical details, including the camera model and shooting mode, which can help quickly and accurately identify the sensor size from the data or use it to select an appropriate sensor preset.  A multiformat sensor? Don’t worry this sounds more complicated than it is. The term refers to using a slightly larger sensor than standard, allowing the camera to “window” the sensor to achieve various aspect ratios, formats, and frame rates directly in-camera rather than capturing the full sensor area and cropping it later.  ARRI Alexa 35 cinema camera, which has an Open Gate sensor size of 27.99 x 19.22 mm A prime example is the ARRI Alexa 35 , which utilises its full active sensor area of 27.99mm x 19.22mm in Open Gate mode and dynamically windows/scales the sensor to accommodate common standards at the correct measurements. For instance, the 1.78:1 mode uses a 24.88mm x 14.00mm area, while the anamorphic 6:5 mode employs a 20.22mm x 16.95mm area. This adaptability ensures the sensor can handle diverse applications, leveraging its entire surface or specific regions to deliver the desired field of view and resolution for each format using the correct imaging circle. Full-frame equivalent or the actual sensor size? If you've searched everywhere but can’t find information about your sensor size, you might still have a "full-frame equivalent" focal length to work with. The term "full-frame equivalent" refers to the focal length of a lens on a camera with a sensor size other than full-frame (36mm x 24mm) that produces a similar field of view to a lens on a full-frame camera. Essentially, it allows for comparing how a lens on a smaller sensor camera would behave if mounted on a full-frame camera. Manufacturers often use full-frame equivalence to simplify marketing, particularly in systems with integrated optics, such as handheld gimbals and drones. However, this approach can obscure the true sensor size. So, can you rely on full-frame equivalence instead? The answer is both yes and no. For example, the DJI Osmo Pocket 3 does not readily disclose its sensor size, but it states that the combination of its sensor and optics produces a full-frame equivalent field of view to a 20mm lens. Using this information, you could input the horizontal size of a full-frame sensor (36mm) and a 20mm focal length to estimate the field of view. However, this assumes the 20mm equivalence is precise. Manufacturers often round up or down to the nearest common photographic focal length for simplicity. While such minor differences are negligible for everyday filming or photography, they can lead to inaccuracies in precision workflows like camera tracking. Full-frame equivalence can provide a starting point if no other data is available. However, be cautious, as it might not deliver the accuracy required for tasks like camera tracking. Wrap Up In conclusion, we hope this post has clarified some of the challenges in identifying the correct sensor size for your camera while providing a foundation in key concepts and terminology. Whether working with high-end cine cameras or smaller devices like drones, understanding and applying sensor size information is essential for accurate tracking. Leverage resources like the VFX Camera Database and DXOMARK to quickly access precise sensor specifications for your projects. Finally, remember that PFTrack offers powerful tools for calibrating your camera body, and the Auto camera model can be a reliable fallback when all else fails. Armed with this knowledge and the right tools, you'll be well-equipped to tackle your next camera tracking challenge with confidence. Links : Head back to Learning Articles . Alternatively, explore our extensive Resources  for valuable presets, Python scripts, and macros. You can also find step-by-step video guides covering the fundamentals of camera tracking and matchmoving in PFTrack within our PFTrack Tutorials .

What’s the difference between automatic and manual tracking? Which is better? When should I use one instead of the other? And how do the differences affect the camera solver? In this article we’ll take a look at some of the more technical details of how trackers are used in #PFTrack, and suggest some ways of getting the most out of PFTrack’s advanced feature tracking tools. What is a tracker? A tracker defines the location of a single point in 3D space, as viewed by a camera in multiple frames. In PFTrack, trackers are generally created using two nodes: Auto Track and User Track. The Auto Track node is able to generate a large number of estimated trackers automatically, and the User Track node provides manual tracking tools for precise control over exactly where each tracker is placed in each frame. Trackers form the backbone of any camera solve, and they are used to work out how the camera is moving along with its focal length and lens distortion if they are unknown. But how many trackers do you need, and what is the best way of generating them? How are trackers used to solve the camera? When solving for the camera motion in a fixed scene under normal circumstances, PFTrack needs a minimum of 6 trackers to estimate the motion from one frame to the next. This is the bare minimum, however, and we generally recommend using at least 8 or 10, especially if you’re not sure of the focal length, sensor size, or lens distortion of your camera. Using a few more than the minimum can also help smooth out transitions in the camera path from one frame to the next, where one tracker might vanish and another one appears in the next frame. Trackers should be placed at points that are static in the real world ( i. e.  do  not move in 3D space), such as the corner of a window frame or a distinguishable mark in an area of brickwork. This allows the 3D coordinates of the point to be estimated, which in turn helps to locate where the camera is in each frame. To help with estimating camera motion, trackers also need to be placed in both the foreground and background of your shot, especially when trying to estimate focal length, as this provides essential parallax information to help the solve. It’s also important to have trackers placed in as many parts of the frame as possible, rather than just bunching them together in a single area. Think of your camera’s grid display as dividing your frame into a 3x3 grid of boxes - try to have at least one tracker in each box in every frame, and you’ll have good overall coverage. Not every tracker is equal We’ll get into the details of how to generate trackers shortly, but before we do it’s important to understand that not every tracker is considered equally when solving the camera. The most significant distinction is whether a tracker is defined as being a soft  or hard  constraint on the camera motion. Hard constraints mean the placement of the tracker in every frame is assumed to be exact. If you’ve generated trackers manually using a User Track then these will be set as hard constraints by default. The solver will try to adjust the camera position and orientation to make the tracker’s 3D position line up with its 2D position exactly in every frame when viewed through the camera lens. On the other hand, trackers that are generated automatically with the Auto Track node are marked as soft constraints and don’t have to be placed exactly in every frame. The camera solver is able to recognise that some errors in the 2D positions exist and ignore them. These are often referred to as “outliers” and might correspond to a temporary jump in the tracker position for a couple of frames or the subtle motion of a background tree in the wind, resulting in the 3D location of the tracking point changing from frame to frame. So now that we’ve explained some of the details about how the camera solver uses trackers, what is the best way of generating them? Auto-Track? User-Track? Or both? Ultimately, the answer to this comes down to experience with the type of shot you’re tracking, how much time you have to spend on it, and the final level of accuracy you need to complete your composite. To get started, here are some guidelines that should help you quickly get the most out of PFTrack’s tools. Automatic feature tracking If you have all the time in the world to track your shot, then of course, manually placing each tracker in every frame is the way to go, as this ensures each one is placed exactly where it should be. Alternatively, automatic feature tracking is a way of generating a large number of trackers very quickly, but because the tracking algorithm is attempting to quickly analyse the image data and work out the best locations to place them, not every tracker is going to be perfect. The Auto Track node picks out a large number of "interesting" points and corners in each image, and tracks those points bi-directionally between each pair of frames (i.e. from frame 1 to 2 and then from 2 back to 1). It compensates for any differences in exposure or lighting whilst doing this, and also tries to ensure that jumps and inconsistencies in the motion of each point between frames are avoided wherever possible. After all the points are tracked, it filters them down to select around 40 trackers in each frame (using the default settings). The trackers are chosen in a way that tries to distribute them evenly over the image area whilst also ensuring tracks with the longest length are used wherever possible, so each tracker is visible in many frames of the clip to help out the camera solver. However, these trackers may end up being placed on objects that are moving independently from the camera, or at other locations that cannot be resolved to a single point in 3D space. For example, so-called “false corners” that result from the intersection of two lines at different distances from the camera can often be indistinguishable from real corners when looking at a single image. Whilst the camera solver will ignore these outliers to a certain extent, having too many trackers falling into these categories can adversely affect the solve, so how should you deal with them? Identifying errors Whilst PFTrack will attempt to detect when tracking fails, not every glitch can be easily detected, especially when your shot contains motion blur or fast camera movement. It’s always worth reviewing automatic tracking results to check whether there are any obvious errors. For example, the motion graphs in the Auto Track node can be used to quickly identify trackers that are moving differently from the others. Trackers can be selected for adjustment or deletion The “Centre View” tool can also be used to lock the viewport onto a single tracker. Scrubbing forwards and backwards through the shot will often expose motion that is subtly different from the background scene, which may indicate a false corner or other gradual object movement. Adjusting trackers So now you’ve identified some trackers that need some attention. What’s next? If you just need to make a few quick adjustments, such as adjusting tracker visibility or re-positioning it in a couple of frames, the Auto Track node provides some Tracker Adjustment tools directly in the Cinema window you can use to get the job done: Tracker adjustment tools You can use these tools to make any minor adjustments to your tracking points before passing them downstream to the camera solver. If you want finer-grain control over your adjustments you can also use the Fetch tool in the User Track node to convert an automatic tracker into a manual one, and all the tools of the User Track node are available to you to adjust the tracker as needed. To adjust or disable? You can manually correct every single one of your automatic trackers if you wish, but as we mentioned earlier, the Auto Track node generates many more trackers than are actually needed to solve the camera motion. This means you may well be spending a lot of time unnecessarily correcting trackers if you have a particularly tricky shot. It can often be just as effective to quickly disable the bad trackers, especially if time is short. This is certainly the case if you’ve only got a few outliers, and also have other trackers nearby that don't need fixing. You could also use the masking tools in PFTrack to mask out any moving objects before automatic tracking, although it’s important to weigh the time it will take you to draw the mask against the time it takes to identify and disable a handful of trackers afterwards. Remember that trackers should be distributed over as much of the frame as possible, and we recommend a minimum of around 10 in each frame, so keep this in mind when disabling. If you end up having to disable a lot and are approaching single-figures, then maybe a different strategy is going to be necessary: supervised tracking. Supervised feature tracking Ultimately, a lot of shots will need some level of manual, or 'supervised', tracking using the User Track node. This is especially important if you’re tracking an action shot with actors temporarily obscuring the background scene. One limitation of automatic feature tracking is that it can’t connect features from widely different parts of the shot together if something is blocking their view or the feature point moves out of frame for a significant length of time. In these cases, human intervention is often necessary, and this is where the User Track node comes into play, allowing you to create trackers from scratch to perform specific tasks. For example, you may have a shot where the camera pans away from an important area for a few seconds and then pans back. Or an actor may walk in front of an important point before moving out of frame. In these cases, you want to make sure the 3D coordinates of points at the beginning are the same as at the end. Creating a single tracker and manually tracking over frames where it is visible (whilst hiding the tracker in frames where it is not visible) will achieve this goal. The same guidelines apply when creating tracking points manually - try to distribute them over your entire frame, and make sure that you’ve got a good number of trackers in each frame. Also, try not to have many trackers stop or start on the same frame (especially when they are treated as hard constraints), as this can sometimes cause jumps in your camera path during the solve that will require smoothing out. If you do, adding a couple of “bridging” trackers elsewhere in the image that are well tracked before and after the frame in question can often help out. Wrap Up Hopefully, this article has shed some light on things to consider when tracking your points. In the end, this all comes down to experience, and as you track more shots, you’ll get a better feel for when to use specific tools, and whether to start with supervised tracking straight away or give the Auto Track node a go first of all.  If you are using automatic tracking, you can easily place an empty user track node between the Auto Track and Camera Solver to hold any user tracks that you may want to create manually as you solve your camera. Also, don’t worry about getting every tracker perfect before you first attempt a camera solve. It’s often possible to try auto tracking first and see where that gets you, then consider how to address any problems and add a few user tracks to help the solver out. PFTrack lets you adjust and change your trackers however you want. If you’ve almost got a solve but can see a bit of drift in a few frames, try creating a single manual tracker over those frames in a sparsely populated area of the image, then solve for the 3D position of that tracker alone, fix your focal length and refine your solution - you don’t have to solve from scratch every time. Links: Head back to Learning Articles . Alternatively, explore our extensive Resources  for valuable presets, Python scripts, and macros. You can also find step-by-step video guides covering the fundamentals of camera tracking and matchmoving in PFTrack within our PFTrack Tutorials .