Data Augmentations functions prebuilt or manually?

Hi,
i want to implement Data Augmentation in my SDG workflow for Domain Randomization. I was wondering if there are any “pre built” functions available for that. In the tutorial for Data Augmentation (Data Augmentations — Omniverse Extensions latest documentation) it suggests that the functions already exist and that one can get them from the Annotator registery. Or do the functions first have to be defined manually?
Kind Regards!

Hi @zollfnn I believe I replied to you on discord, but I’ll ask here as well. I’m not sure I understand completely what you’re looking for. Are you looking for a concrete example script that uses these? (outside of the snippets provided on the webpage).

Hi, thanks for the quick reply. A little script example would be great. I also saw one in the Isaac Sim tutorial but there the augmentation function got defined manually beforehand.(10.11. Data Augmentation — Omniverse IsaacSim latest documentation)

Example regarding the link in my first post, it’s the last augmentation on the tutorial page: …rep.Augmentation.from_function(rep.aug_shot_noise… does rep.aug_shot_noise refer to a function already defined by Nvidia ? I tried to use the augementation in a custom writer but it throw an error. I’ll later attach a screenshot

Circling back here. Docs will be updated for this, but a thought here is to check out the automated tests for augmentations available with the replicator extension.

Check out omni\replicator\core\tests\test_augmentations.py for a few examples of using them.

On windows this is: C:\Users\YOURUSERNAME\AppData\Local\ov\pkg\prod-isaac_sim-2023.1.1\extscache\omni.replicator.core-1.10.20+105.1.wx64.r.cp310\omni\replicator\core\tests or something equivalent.

As a side note this is broadly applicable for checking out most functionality in replicator, though the tests are not ideal from a learning perspective, which is why I need to update the augmentation docs. :)