A pretty exciting tool I got to work on didn't take off in the final production in the end but it was nevertheless very exciting to work on anyway. Take an image and make it alive, sound cool doesn't it?
I've seen some pretty cool examples of style transfer and machine learning methods to make stuff to images but there was something particularly compelling when it came to actually putting it through the Houdini combinate and see what can turn out from this. I did run into quite a few issues as I started from bare scratch (knowledge-wise in Houdini as well) and got lots of help from my peers when it came to potential ways of how to actually build the tool but in the end it did get someplace and fusing renders in Photoshop turned out quite compelling.
I took the idea further in 2021 and implemented some cleaner techniques to make it actually functional, if you'd like to see it all in motion and volume/VDB magic. I might actually go back to experimenting and updating this subpage with more stuff in coming weeks as I've got some ideas :)
Oh, yeah, uh, this all is a combo of Houdini + Renderman!