Once you have uploaded an Input successfully, a Computation will be launched automatically. The Computation is the overall optimization process where your Input 3D is essentially reconstructed into a hierarchical level-of-detail representation and proprietary graphics optimizations are applied to the data.

The Computation Info


This page shows the Computations that have been initiated on your account.

  • Input: Which Input was used to initiate the Computation
  • Computation ID: A unique identifier given by SceneStream to each Computation that helps us debug in case there any kind of errors occur
  • Output: The outcome of a Computation, click on to move to the Output-page
  • State: Status of a Computation as a whole. Can either be Running or Finished
  • Started: When the Computation was launched (timestamp in UTC)
  • Single Computation Task States: A Computation comprises of many invididual tasks.
    • Waiting: The dependencies related to a task have not yet finished
    • Pending: Dependencies have been completed but no computing resources are available yet
    • Queued: An internal metric to measure slack in the Computation
    • Running: Task is being processed
    • Success: Task has finished processing
    • Failed: Task has encountered an error in processing
    • Total: Total number of tasks processed and/or failed
  • Completed: When the Computation as a whole was finished (timestamp in UTC)
  • CPU-hours: How many CPU-hours were used to process the content. Sometime in the future as we intend to move towards usage-based billing, this data point will be one the key metrics. For now, it's visible for informational purposes.
  • Progress: How much of an Input's Computation progress is done. 100% means the Output has been successfully produced

Additional Remarks

  • The number of tasks might keep growing despite the Computation already starting. This is simply due to our system minimizing slack and starting to compute the moment there is something to compute
  • Older Computations on your account might show 0 tasks. This is simply due to us changing the way how data is collected and data after a certain data doesn't exist anymore.
  • Computations can be deduplicated. This means if you launch a computation with identical input, computation parameters and computation version, you’ll only see 1 task processed. This is because all the results already existed, thus the system will not re-compute them.
  • This can also affect a computation only partially - perhaps you changed only a small part of the input while the rest remains the same. This usually manifests itself as a quickly completed computation

Next up, we have a lecture about Outputs or, if you feel like skipping class, go back to Basics