Haha thanks a lot! I'll upload the before & after vfx comparison very soon, it's really fun to see what the original animals in the videos were and what I made them into.
Thanks a lot, I'm glad you like it! Later I'll try to make some good step by step of my process, also I'll share the before&after vfx comparison too. Of course my workflow was through lots of trials and errors, but in summary: I take a single frame (or 2) from a short video, then I generate img to img in stable diffusion using controlnet to keep most of the base shapes of the source image (it's quite much the fun part of experimenting how to twist the image of a normal animal into something what doesn't exist), then I separate the image in photoshop into different parts (background, the subject, certain extra elements), then with EbSynth unsing the original footage I render different passes of certain elements of the Stable Diffusion result. Then in After Effects I composite all the png sequences together, and in some of them I use tracking, depth map distortion, puppet warp, liquify, also using the virtual camera and lights to give a little extra life to the images!
@@chromery Thanks! I can see that probably in the near future there will be more and more advanced tools to create visually anything, but I think there will always be space left for some creativity!
I would definitely watch this
Wow! Want to check this out.
Just saw this on 9gag. Excellent video!
Hello fellow 9gagger, thanks a lot for the support!
This was fucking amazing!
Haha thanks a lot! I'll upload the before & after vfx comparison very soon, it's really fun to see what the original animals in the videos were and what I made them into.
Looks awesome!! How did you do this, could you share your workflow?
Thanks a lot, I'm glad you like it! Later I'll try to make some good step by step of my process, also I'll share the before&after vfx comparison too. Of course my workflow was through lots of trials and errors, but in summary: I take a single frame (or 2) from a short video, then I generate img to img in stable diffusion using controlnet to keep most of the base shapes of the source image (it's quite much the fun part of experimenting how to twist the image of a normal animal into something what doesn't exist), then I separate the image in photoshop into different parts (background, the subject, certain extra elements), then with EbSynth unsing the original footage I render different passes of certain elements of the Stable Diffusion result. Then in After Effects I composite all the png sequences together, and in some of them I use tracking, depth map distortion, puppet warp, liquify, also using the virtual camera and lights to give a little extra life to the images!
@@avianbrothers oh wow, that’s definitely complex, no wonder I couldn’t figure it out! 😃 good job man, really well done!!
@@avianbrothersthe continuity you achieved is really amazing, the more I look the more I’m impressed
@@chromery Thanks! I can see that probably in the near future there will be more and more advanced tools to create visually anything, but I think there will always be space left for some creativity!