What we tested in ARMORY /
The tests behind ARMORY: inputs, actions, observed results and the limits of that evidence.
A mineral-green asset workbench pairs a large interactive 3D stage with an object inspector and collection shelf. Three original GLBs can be inspected and downloaded from the public interface.
Internal local testing, using synthetic inputs and owned assets. The results below establish the observed software behavior in those scenarios. They do not establish customer ROI, uptime or external deployment readiness.
Original assets were actually inspected
Browser QA loaded the original three-model Lumia collection and inspected a lamp’s named parts through wireframe and exploded views. These were actual self-contained models rather than generated screenshots of a product interface. The collection provides a lamp, side table and planter for exercising the library’s inspection and delivery workflow.
A two-version delivery test
An original Nexus One device model was uploaded as the first version of a synthetic asset. Its report showed 99 meshes, 12 materials, 120 nodes and 1,249,892 bytes. A delivery was saved, then the asset received a Nexus Pro second version with 153 meshes and 1,614,892 bytes.
The older delivery stayed exact
The ZIP was downloaded after the second upload. Its model still matched the original first-version file byte for byte, and the manifest retained the original rights statement. This checks the important handover invariant directly: updating the library did not replace the bytes selected for an existing delivery collection.
Library work survived reload
The uploaded asset and saved delivery remained after an ordinary browser reload. A synthetic tag search returned the intended device study. Desktop and mobile screenshots were inspected; the 390-pixel mobile page had no horizontal overflow. The fresh authenticated page had no runtime errors or warnings. No sales, licences granted by third parties or real client deliveries are claimed.
What this evidence does not establish
The supported asset input is a validated, self-contained GLB 2.0 model. Arbitrary DCC project files and external dependency trees are outside this delivery contract.
The current aggregate delivery cap is 20 MB; larger handovers need separate collections.
A rights statement is supplied by the owner. ARMORY does not provide automatic legal licence verification.
Viewport inspection and structural counts do not establish animation quality, material compatibility with every renderer or fitness for a client's production.
The original collection demonstrates the software; it is not evidence of asset sales or external client work.
A similar problem in your business?
Bring a real workflow, representative inputs and the result that needs to be reliable.
Shape a project brief