Hey Shawn, thank you for having me as an interview partner! It seems I was a bit surprised about the question about AI creating samplepoints. Well, I can say it does not. Thats for sure. Was fun to answer your questions. Lets keep up the good work.
What a great interview with wonderful insight. I had no idea that we all can contribute to its training based on the fact that we may have a photo where the gradient wasn't removed properly.
Great interview Shawn & Frank. Excellent content and an awesome tool. At the moment I think the best thing about Graxpert is the one click aspect, no having to wonder what and how much to tweak. The tool will only get better with more training. Looking forward to it.
The Pixinsight folks don’t seem to trust AI-based tools, perhaps because they feel the AI could be creating something synthetic that’s not really there. They prefer algorithms based on specific mathematical models. But I think AI is just interpolating data points from the training data, whereas the models are interpolating data points based on statistical assumptions. The AI approach can model more complex distributions of real world data than can be captured in simple statistical models, so I think the AI approach might be more powerful in the long run. For now, it’s nice that we can try multiple methods and pick the best one. Thanks for the interview!
GREAT WORK!! It's also great that you are working on a noise reducing software! I would have loved to hear you say "we are working on a blur terminator AI" 🤔😉
unfortunately it makes no sense for us, to work with older OS, as they sooner or later get obsolete. This would be eaten up the precious time we could spend on further developments. Also it might be, that older OS missing important packages that make the AI work. I am sorry.
Hey Shawn, thank you for having me as an interview partner! It seems I was a bit surprised about the question about AI creating samplepoints. Well, I can say it does not. Thats for sure.
Was fun to answer your questions. Lets keep up the good work.
What a great interview with wonderful insight. I had no idea that we all can contribute to its training based on the fact that we may have a photo where the gradient wasn't removed properly.
Would love to see an AI driven hyperbolic stretching tool that preserves highlights.
Cool interview! You two guys helped me most with your PI tutorials in your YT channels! Big Thumb Up👍
Thanks for creating GraXpert. Great piece of software. Excited to hear you guys are working on a software to denoise! Can’t wait to see it
Great interview Shawn & Frank. Excellent content and an awesome tool. At the moment I think the best thing about Graxpert is the one click aspect, no having to wonder what and how much to tweak. The tool will only get better with more training. Looking forward to it.
Very diplomatic responses.
The Pixinsight folks don’t seem to trust AI-based tools, perhaps because they feel the AI could be creating something synthetic that’s not really there. They prefer algorithms based on specific mathematical models. But I think AI is just interpolating data points from the training data, whereas the models are interpolating data points based on statistical assumptions. The AI approach can model more complex distributions of real world data than can be captured in simple statistical models, so I think the AI approach might be more powerful in the long run. For now, it’s nice that we can try multiple methods and pick the best one. Thanks for the interview!
GREAT WORK!! It's also great that you are working on a noise reducing software! I would have loved to hear you say "we are working on a blur terminator AI" 🤔😉
Be great if they could make it work on older versions of iOS. It works with older versions of Graxpert but not the AI versions
unfortunately it makes no sense for us, to work with older OS, as they sooner or later get obsolete. This would be eaten up the precious time we could spend on further developments. Also it might be, that older OS missing important packages that make the AI work. I am sorry.