This one is just too close to home (literally) for me to ignore! I’ve seen a few versions of this floating around but I just never thought any of them were any good, and that rubbed me the wrong way because I live about 45 mins from Fairplay Colorado, in the South Park Valley which, although not publicly confirmed, is the inspiration for the show. If you go, you know.
I honestly haven’t seen a new episode in years, but I’ve been a fan of the show for decades, way before moving to its home state. Both Trey Parker and Matt Stone have a deep connection to CO, and for me, living in Colorado (specially so close to the show’s birthplace) makes the whole show click in a much more meaningful way.
Side Note: While working on this other project years ago, I was super lucky to get animation tips and direction from Chris ‘Crispy’ Brion himself—South Park’s OG animator 🙏
FYI: These renders were all done with V6, I found that V6.1 did not respond to my text+image ref. prompts very well in comparison. Also, if you’re asking where the group shot is, well, that one was pretty much near impossible with Midjourney alone so I leave that one to you my friends!
🥳 South Bark Prompts + Ref. Images:
As far as reference images go, everything in the reference image matters, I don’t care if you are using the image as a character ref, style ref or image ref, everything in the image plays either a more prominent or a less prominent role in the final gen, but it plays a role no matter what. I used two reference images per character, one for the South Park character and one for the dog.
Kenny McCormick
https://s.mj.run/Ib-XRt4OrD8 https://s.mj.run/j5O42CMoXFo kenny mccormick as a dachshund dog, in the style of south park --ar 4:5 --v 6
The reference images have a very simple background and you can see that in the final output. Consistency is key, all the ref images that I used are very similar with the character refs’ white background, and the dog refs’ also very simple background but not necessarily white. Keep in mind that my prompt is very simple and does not describe the background at all, I’m letting the reference image backgrounds guide the model, more so than my prompt. Something else to note is that I’m using the ref images in the most basic reference mode—through Discord you would type the two image URLs at the beginning of the prompt divided by a space (as you can see in the prompt above). If you’re using the Midjourney website it would look something like this👇
Eric Cartman
https://s.mj.run/B6U-UU8jUbg https://s.mj.run/oLkKmlox2ew eric cartman as a bulldog in the style of 3d south park --ar 4:5 --v 6
Stan Marsh
https://s.mj.run/DGnq-8nqzCs https://s.mj.run/npxAOylECbY stan marsh as a terrier dog in the style of 3d south park --ar 4:5 --v 6
Kyle Broflovski
https://s.mj.run/ltbkKl7QF7E https://s.mj.run/NuCJdH2bYBU kyle broflovski as a shiba inu dog in the style of 3d south park --ar 4:5 --v 6.1
