Can the model be made open so that users can train it by themselves? For example, I’m interested in enhancing the main characters in movies upscaled to 4K. Then I can specifically train for the restoration of characters in 4K. Officially, perhaps due to cost reasons, the training might stop when it is considered more or less sufficient. But we users may have idle graphics cards, such as 3080 or 4090. It would be a waste not to use them to improve the picture quality. I think it’s great to be able to train by oneself specifically for one’s own special preferences. Can an update be made open?
The answer is probably no. And probably doubly so when it comes to having compatible training data.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.
