Appearance
Choose Metashape when you need a fully documented, scriptable, reproducible pipeline you control end to end; choose RealityCapture when raw speed on large datasets and lower upfront cost matter most. Both produce heritage-grade models, so the decision is really about workflow philosophy, licensing, and how defensible your process must be. This guide gives you a concrete checklist rather than a verdict, because the right tool depends on your constraints.
What actually differs between the two engines?
Under the hood both perform structure-from-motion, dense matching, meshing, and texturing — but their defaults and ergonomics diverge sharply.
| Factor | Agisoft Metashape | RealityCapture |
|---|---|---|
| Speed on large sets | Good, tunable | Very fast, GPU-heavy |
| Parameter control | Extensive, granular | Streamlined, fewer knobs |
| Scripting | Mature Python API | CLI and limited scripting |
| Licensing | Perpetual standalone/pro | Free-to-use, pay-per-output/sub |
| Laser scan + photo fusion | Strong | Strong |
| Default mesh density | Moderate, you push it | Very dense out of the box |
| Reproducibility | Excellent (scriptable) | Good, less granular |
Neither is "better" in the abstract; they reward different working styles.
Which one is faster for big jobs?
RealityCapture is generally the speed leader, frequently finishing several times faster on sets of thousands of images because it leans hard on the GPU. If your project involves large monument or excavation captures and turnaround time is the constraint, that speed is decisive. Metashape narrows the gap on smaller artefact sets and lets you trade quality for time deliberately with explicit settings.
Which is better for reproducible, defensible work?
For heritage documentation, reproducibility often outweighs raw speed. Metashape's Python API lets you encode your entire pipeline so the same parameters run identically every time:
python
import Metashape
doc = Metashape.Document()
chunk = doc.addChunk()
chunk.addPhotos(image_paths)
chunk.matchPhotos(downscale=1, generic_preselection=True)
chunk.alignCameras()
chunk.buildDepthMaps(downscale=2)
chunk.buildModel(surface_type=Metashape.Arbitrary)
chunk.buildUV(); chunk.buildTexture(texture_size=8192)
doc.save("artefact.psx")A committed script like this is your method statement — it documents exactly what was done, which is the gold standard for defensible heritage outputs.
How does licensing change the decision?
Licensing shifted the landscape. Metashape sells perpetual licences (Standard and Professional), so institutions can own the software and budget once. RealityCapture moved under Epic to a free-to-use model with pay-per-output credits or subscription for exporting commercially. For a publicly funded archive, the question becomes: do you need predictable ownership (Metashape), or low upfront cost with usage-based export fees (RealityCapture)? Read the current terms before committing a multi-year programme.
Can I avoid lock-in between them?
Yes — design for interchange from the start. Both tools import and export open formats, so treat your project data as portable:
- Export meshes as OBJ or PLY, textures as PNG/JPEG sidecars.
- Export camera positions and intrinsics for re-use.
- Keep your source images plus a capture log as the true archival master — the project file is derivative.
If the master photos and a documented capture record survive, you can re-derive a model in either tool years later.
What's a practical checklist to decide?
Work through these questions in order, and the answer usually becomes obvious:
- Must the pipeline be scripted and reproducible? Lean Metashape.
- Are datasets large and turnaround time critical? Lean RealityCapture.
- Does your institution need perpetual ownership? Lean Metashape.
- Is low upfront cost more important than export fees? Lean RealityCapture.
- Do you fuse laser scans with photos? Both work; test on a sample.
- Whichever you pick, archive source images and document settings.
Key Takeaways
- Both tools make heritage-grade models; the choice is about workflow, not quality ceilings.
- RealityCapture wins on speed for large datasets; Metashape wins on granular control.
- Metashape's mature Python API makes it the stronger choice for reproducible, defensible pipelines.
- Licensing differs fundamentally: Metashape perpetual ownership vs RealityCapture free-to-use with usage fees.
- Avoid lock-in by exporting OBJ/PLY/cameras and archiving source images as the true master.
- Document your settings whichever tool you choose — the parameters are your method statement.
- Run a real sample dataset through both before standardising an institution on one.
Frequently Asked Questions
Which is faster, Metashape or RealityCapture?
RealityCapture is usually faster, often several times faster on large image sets, because it is aggressively GPU-accelerated. Metashape is competitive on smaller sets and offers more predictable, tunable timings.
Which produces better quality models?
Both can produce excellent heritage-grade models; differences come from your capture and settings, not the engine. RealityCapture tends to generate very dense, detailed meshes by default, while Metashape gives finer manual control over the whole pipeline.
Is Metashape or RealityCapture better for reproducible workflows?
Metashape is generally preferred for documented, reproducible pipelines because it exposes every parameter and offers a stable Python API for scripting. This matters when results must be defensible and repeatable across a project or institution.
What about licensing and cost?
Metashape sells perpetual standalone and pro licences, which suits institutions that want to own their software. RealityCapture moved to a free-to-use model under Epic with pay-per-output or subscription options, which changes the cost calculus significantly.
Can I move a project between the two tools?
You cannot move a native project directly, but you can exchange intermediate data: export cameras, point clouds, and meshes in open formats such as OBJ, PLY, and standard camera or XMP exports. Plan exports up front so you are not locked in.
Which should a small heritage team choose?
Choose Metashape if you need scripted, fully documented, reproducible pipelines and own-your-licence stability. Choose RealityCapture if speed on large datasets and lower upfront cost matter more than fine-grained parameter control.