Postshot
Runs Inside Blender
The add-on trains the scene and shows it in the Blender viewport. You can move the model while it learns.
Two Outputs
One training run targets either a Gaussian splat for visual presentation or a mesh for geometry you can measure and model against.
Video In, Scene Out
Postshot accepts footage directly. It extracts the frames and trains the scene, so a single video is a complete input set.
Introduction​
Postshot, from Jawset, is a free desktop application and Blender add-on for 3D reconstruction. It trains neural radiance fields and 3D Gaussian splats from photos or video, and it can also produce a mesh from the same capture. It runs locally on your GPU and previews the scene while it trains.
Pick Postshot when the capture is video or a mixed set and you want the scene in Blender quickly, or when you want a splat for visual presentation. For a measured, text-based reconstruction pipeline see COLMAP. The two reconstruction methods are documented separately: Gaussian Splatting and photogrammetry.
Installation​
- Download Postshot for your platform (Windows, macOS, Linux) from the Jawset site.
- Install the Blender add-on from the same download: in Blender, go to Edit > Preferences > Add-ons, click Install, and select the add-on file. Enable it.
- Verify the version in the Postshot preferences panel and record it. The 1.1 series is current as of September 2026 (latest build v1.1.71).
Usage​
A run is a training job: input media goes in, a scene file comes out. The training runs on the GPU and shows a live preview.
Prepare the input· photos or one video
Collect the media: a folder of photos or a single video file. What goes in: the capture. Keep the scene static, keep exposure stable, and make sure the camera path covers the subject from the views you will present. For video, a slow orbit beats a fast one.
Start the training· pick the target
In Postshot, add the media and choose the model target: a splat for visual presentation or a mesh for geometry. Start the run. The GPU does the training; the viewport shows the scene forming, including the estimated camera positions.
Inspect the scene· the QC gate
Orbit the finished scene. Check the camera positions against your capture path. Look for holes, ghosting, and floaters. Use the selection tools to remove bad regions: sky, moving people, reflections. Re-train after changes.
Verify against reality· before you trust it
Compare the scene against what the capture shows: known dimensions, the position of known features, the camera views. A splat is a rendering approximation. What it shows is not what a measurement instrument would record. Record the checks.
Export· scene file and mesh
Export the splat file and, if you ran a mesh target, the mesh. What comes out: files for Blender and a record of the training settings for the case file.
In an Investigation​
Use the block below to document how you ran Postshot on the evidence.
Documentation: how to document Postshot
In Your Method Section​
Guidance: Describe the purpose, the tool version, and the input data.
- Investigative Objective: To generate a 3D scene of the capture for [spatial analysis / visual presentation / camera position recovery].
- Software Version: Postshot [Version, build] (e.g., 1.1, build v1.1.71), Blender [version].
- Input Data: [Media name/ID], [N] images or [duration] of footage derived from [source document], Source Hash: [First 6 digits].
In Your Decision Log​
Guidance: Record the specific procedural parameters and integrity checks.
- Specific Settings: Model target (splat or mesh), training settings that differ from defaults, GPU used, regions removed with the selection tools and why.
- Assumptions: The scene was static during the capture; the camera path covered the subject.
- Integrity Check: Analysis ran on a working copy, never the original evidence file.
- Date Performed: [YYYY-MM-DD].
Verification & Mitigation​
Guidance: How did you verify accuracy and mitigate bias?
- Verification: Compared the scene camera positions against the capture path. Checked known dimensions and features in the scene.
- Bias Mitigation: Treated the splat as a visualization, not a measurement. Derived a mesh or used photogrammetry for any measurement that enters the findings.
Example Methodology Statement​
"A 3D scene was trained with Postshot 1.1 (build v1.1.71) in Blender [version] from [N] frames of [source] (Hash: A1B2C3). The target was [splat/mesh]. Selection removed [regions]. Camera positions in the scene match the capture path; known features were verified against the footage. The scene was used for [purpose only]; measurements come from [mesh/photogrammetry source]."
Into Blender​
The add-on keeps the work in one application.
- Splat: the trained scene displays in the viewport through the add-on. To move it to another project, export the splat file and import it with BlendSplat or SkySplat.
- Mesh: export the mesh from Postshot and import it with the matching Blender importer.
Three caveats before you work against the result:
- Scale is arbitrary unless the capture carried GPS data. Anchor the scene to a known dimension before you place other objects in it.
- Splats do not measure. Use the mesh output, or a photogrammetry pipeline, for any dimension that enters a finding.
- For georeferenced scenes: place the reconstruction relative to the site with the QGIS baseline (custom CRS and origin point).
Limitations​
- Gaussian splats are a rendering approximation. They produce floaters, ghosting, and view-dependent artifacts, and they carry no solid geometry.
- Training needs a capable GPU, and large scenes take real time.
- The tool is free but comes from a small company with a fast-moving build line. Pin the build number in your method section.
- The same capture weaknesses apply: reflective and transparent surfaces, and moving objects, degrade both splats and meshes.
Related​
Resources​
- Gaussian Splatting, the reconstruction method this tool runs, with its investigation limits
- Photogrammetry, the measurement-grade alternative
- BlendSplat, the Blender add-on for importing splat files
- Colmap, the command-line reconstruction pipeline
Sources​
- Jawset Postshot, official site with downloads
- Postshot Getting Started guide, official first-run walkthrough
- Postshot release notes, version 1.1 features and history
- Postshot Windows builds, build list with the latest v1.1.71