very realistic neural style transfer, based off Titu1994's project on github, just ported to colab and made slight compatability changes in the ipynb itself to remove tf issues. unlike my main lucid tensorflow based neural style transfer, this one is less dynamic, (less randomness) but more overal realisticness. images above 650 pixels tend to crash so keep them small, 100 iterations is good for most results. hardware acceleteration is enabled by default as well as python2 env, you can run online (no installation needed) via the following link below
google colab link: https://colab.research.google.com/github/400lbhacker/realistic-neural-transfer-/blob/main/Titu1994_neural_style.ipynb
van gogh night terrace example
van gogh over the rhine
