fixed critical bug where saving a config completely let the inference stop functioning.
Nothing else was changed.
Changelog of TVP Beta 5
- Added new Model. This model performs very well in all tests i have made, therefore it got merged with the other model which got trained. More Models are on the way.
- Removed Batch Size from the UI, as tests found out, Batch Size 1 may be the fastest. there still is the possibility to change it in the config file, once you saved.
- Cleaned up the UI
- now shows Patrons + their proper Patron Tier with a scrollbar!
- cleaned up code, reduced loading times, faster inference
- tweaked x264 config from crf 20 to crf 18
- added twoloop aac - Audio will sound better than before.