After creating the mask how would you go about replacing the bird without changing the background scene? lets say you wanted to replace the bird with a cyborg parrot?
Great video! Thank you for the clear explanation of everything! A small question - I keep getting the rectangular frame from Florence together with my mask. How do I cancel the frame and only get the mask?
you can indeed mask many regions, that's on florence2coordinate, it's number coded, first is 0, second is 1, and so on. so if you want bird and claws, zooms in and check your preview images for the number code. also in newer update of sam2segmentation check the individual_objects to true, and add bboxes also from florence2coordinate, not just mask.
Also Florence 2 doesn't also lock on to that number for that object if it loses the number like 1 or 2 and the 1 become 2 than it'll bounce around. This was made the same day sam2 was released.
@@A.eye_101 I actually came here to say the same thing. I can get Sam2 to mask no problem, but I dont do enough comfyui animatediff workflows to know what nodes i need put in after I have masked the subject. The video combine node only has "filenames" output on that end, or imagecompositemasked node has 'image' output or ImageAndMaskPreview just has "composite" output. Any chance you can share a workflow that has simple conclusion for video workflow noobs :D thank you for the great simple video btw.
Thank you very much for you tutorial. Comfyui UI as a nodal image compositor has made a great step forward
Glad you liked it
GJ, you are fast! New model is really accurate! Tons of possibilities!
Glad you think so!🎉
After creating the mask how would you go about replacing the bird without changing the background scene? lets say you wanted to replace the bird with a cyborg parrot?
Love your username!
🎉😮❤ thanks 🙏🏽
hahaha I just realized after reading your comment
Great tutorial thanks! 🙌
@@josephine.miller glad you enjoyed it 😊
Great video! Thank you for the clear explanation of everything!
A small question - I keep getting the rectangular frame from Florence together with my mask. How do I cancel the frame and only get the mask?
did you found a solution ?
@@philippeheritier9364 Yes, I got some other workflow from someone that just didn't product this
It just released today. thanks for speedy try. but what are the potential use cases?
If you're using segmentation nodes you'll notice a cleaner and better tracking with your mask.
How you do it so fast , this is amazing
Have to ask Kijai, the nodes are his magic. I'm just telling people how to use them a certain way
you can indeed mask many regions, that's on florence2coordinate, it's number coded, first is 0, second is 1, and so on. so if you want bird and claws, zooms in and check your preview images for the number code.
also in newer update of sam2segmentation check the individual_objects to true, and add bboxes also from florence2coordinate, not just mask.
@marhensa yeah, I posted that in the discord for the tutorial. Thanks to an update. 😇
Also Florence 2 doesn't also lock on to that number for that object if it loses the number like 1 or 2 and the 1 become 2 than it'll bounce around. This was made the same day sam2 was released.
great tutorial.
when i output, it keeps the red on the subject. is there a way to output the subject only and without the red?
@lionhearto6238 this is just to show an example, you'll have to change it for your needs.
@@A.eye_101 I actually came here to say the same thing. I can get Sam2 to mask no problem, but I dont do enough comfyui animatediff workflows to know what nodes i need put in after I have masked the subject. The video combine node only has "filenames" output on that end, or imagecompositemasked node has 'image' output or ImageAndMaskPreview just has "composite" output. Any chance you can share a workflow that has simple conclusion for video workflow noobs :D thank you for the great simple video btw.