Playing with this today, thanks very much for it. I made a couple of changes for myself. I realised that if I cloned the ksampler, I could make one path produce good quality output images, while the original one can do a few steps just to make the new poses. Then, I made the accuracy 4 decimal places (in the settings). Then I used the random input node, so the pose Ksampler can have an ice low noise value so it can change a bit, while the high quality image output can output with a value randomised about 0.85 to 0.95 - this way you get a variety of poses, but don't change the apperarance, at least not much,
Thank you for a very easy to follow guide to converting a photographic image to a new image with the same pose. Well done
Playing with this today, thanks very much for it. I made a couple of changes for myself. I realised that if I cloned the ksampler, I could make one path produce good quality output images, while the original one can do a few steps just to make the new poses. Then, I made the accuracy 4 decimal places (in the settings). Then I used the random input node, so the pose Ksampler can have an ice low noise value so it can change a bit, while the high quality image output can output with a value randomised about 0.85 to 0.95 - this way you get a variety of poses, but don't change the apperarance, at least not much,
"Technology is just a tool" - that is it! 👏👏
how can i do two characters interacting ? for example light a fight scene
(00:39) adjust the "THEN_NO_ICE" parameter ? 😋
thanks
Brilliant
Workflow file is gone, only a jpg on github, what BS is this?
It is still here, contained inside the image Infinite-pose-generatorv02.png, you just have to save it and drag-n-drop it on ComfyUI.