You are viewing a single comment's thread from:

RE: The Well of Lost Souls

in OnChainArt10 months ago (edited)

That literally blew my mind. I've got curious to learn how it's done.


It is kind of a long process that I have been developing for a while, I generate multiple versions of the original photo by running it through with different style images, then I layer the new versions over the original photo at different opacities and blend modes in PhotoShop, then I save new copies of the original photo with the generated textures/colors over the top - I take these new copies and use them as a new base layer in the deepdream generator and keep iterating until I get to the vision I am looking for (layering the output back over the original photo each time to keep some of the texture and detail of the original as well). For this particular series I have 18 different versions of the photo layered in my PhotoShop document now :)

EDIT and I forgot to mention, this may be obvious but the first, third and fourth images in the post are mirrored. The first image is derived from the second image, I mirrored the second image four ways, rotated the resulting mirror 45 degrees and cropped out a section of the resulting rotated mirror, and mirrored the new crop 4 ways, then cropped out a section of that to get the post cover image.

Interesting, looks like it involves a lot of work and patience. Thank you for the explanation.