GL Projects
One of the biggest complaints in 2.6.3 was the model crashing on batches larger than 10 images. In 2.6.4, Topaz rebuilt the memory handler. You can now process entire folders of portraits without the dreaded "CUDA out of memory" error.
| Version | Export Time | RAM Usage | Success Rate (50 images) | | :--- | :--- | :--- | :--- | | 2.6.3 | 48 seconds | 6.2 GB | 74% (crashed 13x) | | | 41 seconds | 4.8 GB | 100% |
We tested 2.6.4 against 2.6.3 using a standard test: