First of all, great work on Capture V2 — the results are impressive. I’d like to suggest a feature that, in my opinion, could significantly improve both usability and flexibility of the capture workflow:
the ability to process/train Capture V2 locally, in batch, without relying entirely on the cloud.
For example:
-
Local training using the user’s CPU/GPU
-
Batch processing of multiple captures
-
Integration directly inside Cortex Control (or a companion desktop tool)
This would be similar in concept to what TONEX does with local model training.
Why this could be valuable:
-
Less dependency on cloud availability and queues
-
Faster iteration when creating multiple captures
-
Better workflow for studios and power users
-
Offline capability for users with limited or unstable internet
-
More transparency and control over the training process
I completely understand that cloud training has advantages (consistency, hardware control, scalability), but offering local/batch training as an optional alternative would be a huge plus.