Great work.. Thanks for this lora
Great work.. Thanks for this lora
Definitively an interesting lora. A very low res test run, seems to work ;-)
Some leaking of identity (gray hair at the kid...), but that might be the prompt not being clear enough.. just a test run ;-)
Will test it a bit more, see how it works
finally we have, thanks for your hard work LiconStudio with VBVR and RuneXX for all your wf
I'am asking me if it works with lets say just a black background and u prompt it like "The black background fades away and the person in image 1 and image 2 doing this and that" since i think the background is required i think?
Definitively an interesting lora. A very low res test run, seems to work ;-)
Some leaking of identity (gray hair at the kid...), but that might be the prompt not being clear enough.. just a test run ;-)
Will test it a bit more, see how it works
Thank you for your feedback! I also noticed similar issues during testing. Using multi-angle character reference images can help mitigate this problem. I will work on addressing this in the next version.
I'am asking me if it works with lets say just a black background and u prompt it like "The black background fades away and the person in image 1 and image 2 doing this and that" since i think the background is required i think?
I haven't tested this, I require a background pic because my training data always include a background at the end.
Thank you for your feedback! I also noticed similar issues during testing. Using multi-angle character reference images can help mitigate this problem. I will work on addressing this in the next version.
With better prompting it all worked fine (explicitly saying the young boy wears a read hoodie with hoodie over his head.. etc. And starting the prompt by describing a bit the two characters)
So a bit my own fault too ;-)
But played around with it a bit more, and definitively a really nice lora .. and try some multi-angle ref too