Thanks for the tips on Copycat, i was wondering if you have played around with the Modnet tool. There's not a lot of information on it at the moment. In the right circumstances it can produce a mask for a person almost instantly just by dropping it in, but it can be very unpredictable and more times than most can produce flickering artefacts and jagged edges. You can use some nodes above the tool to seed slightly different results, but so far I've found nothing that's dependable. If you see it work well however you can certainly why this would be such a powerful tool, and id love to find out more, maybe the answer here is that the tool is still not quite ready yet, but i would love to hear your opinion or any new findings you may have as there is not much i can find online about it. Thanks for the videos, all the best!
Amazing stuff. One question. When you prepare and use the second data set I understand you resume training in the copycat node, in order to improve performance and add more learning to the .cat files. Is that correct?
Thanks for watching! If you could leave a Like here on UA-cam and/or Follow me on my LinkedIn I'd really appreciate that! ♥
Is roto's job in danger?please reply
Thanks for the tips on Copycat, i was wondering if you have played around with the Modnet tool. There's not a lot of information on it at the moment. In the right circumstances it can produce a mask for a person almost instantly just by dropping it in, but it can be very unpredictable and more times than most can produce flickering artefacts and jagged edges. You can use some nodes above the tool to seed slightly different results, but so far I've found nothing that's dependable. If you see it work well however you can certainly why this would be such a powerful tool, and id love to find out more, maybe the answer here is that the tool is still not quite ready yet, but i would love to hear your opinion or any new findings you may have as there is not much i can find online about it. Thanks for the videos, all the best!
Hi Bullhead, I have not. Sounds promising, I'll check it out!
Amazing stuff. One question. When you prepare and use the second data set I understand you resume training in the copycat node, in order to improve performance and add more learning to the .cat files. Is that correct?
Hi Mac, yes that's right!
id like mix and merge this with mocha and ai to get it done faster haha
Is roto's job in danger?please reply
Don't think so, but roto artists will be using AI as a tool now
Footage🤔🤔🤔