I’ve been working on this for about two weeks. While not my largest project folder (only 8 GB), this scene has been a real pain in the rear. At the highest point, this had roughly 200m tris (that is, 200,000,000). Eventually, I reduced it to a more manageable 12m.
Rendering and compositing, however, takes 50 GB of RAM. I made everything in individual files- linking everything for the final rendering was the worst experience I’ve ever had in Blender. It took two days of work and at least a few dozen crashes. Each linkage took 70 GB of RAM and about 30 minutes, and I ended up having to trash the main file a few times because it kept getting corrupted.
It looks amazing, the final product is even better than I thought it would be, but what I am a little confused about is the fact that this sub was built in 1920? The rear portion with the legs and such looks a little more modern, but the overall render is amazing, it has been a pleasure watching this project evolve from the start.
-Julian
Not really- more just “inspired by the technology of that time”. Not from our world, not from our history. “A long time ago in a galaxy far, far away…” and all that
I cheated a bit there - I started with an image, heavily altered with compositional nodes. Didn’t feel like doing them procedurally… would have taken weeks, probably
great work! even tho i’m not much into NPR it kept me looking for a while.
really looove the water surface/light on top. almost realistic that i feel it’s adding some magic to it. not sure how to describe it maybe adding a contrast making the NPR art stand out more or making NPR fits in a realistic world and giving it more life
This was a fun one to watch evolve; the tech constraints were problematic, but nonetheless I think the overall project time was rather fast from concept-to-final.
I feel like the Pre-comp/post comparison slider really tells a good story, for future readers who are just starting out … and wondering why their initial renders lack a certain “pop factor”.
Post is such a great tool… but, it can sometimes go sideways and make an image into something it perhaps shouldn’t be. But here, everything from the shading, to lights, to final really all work well together, each built well on the previous.
I think it’s like the reply below says- the stuff before compositing is really only 30% or less of the final result. A lot of NPR stuff doesn’t bother with compositing and post, and it just doesn’t work well without it.
This might be an understatement
I already kind of said this, but let me agree with you and reiterate- just throwing a cel-shader on your model and rendering it will yield amateurish, flat-looking results. Cel-shading is a bit of a unique case; with realism, you can get away with skipping post sometimes and you’ll still have something good-looking. Cel-shading? Nah.
I really dig the sub design… super cool. I wish it were more centered, or a bit closer in the image. The set dressing with all the seaweed and clams is excellent. I think your final image lacks readability though, the diver blends so much with the background I didn’t realize what this silhouette was until after a few seconds, and the terrain appears mostly one shade of grey/brown, which doesn’t give justice to the cool erosion your gaea capture shows. Perhaps the fog could be a little denser to help with that (detaching foreground from background). Cheers
Absolutely love it! Only thing would be that it took time for me to find the diver, could use some more contrast when looking at it in b/w. But that’s a simple fix, checking the values, otherwise I love the business, the crabs, fish + of course the submarine
I see where you’re coming from. I wanted the divers to blend in- my goal compositionally is the viewer is immediately drawn to the sub by the stark contrast and the lines of the chain and sun rays, and the sub is the focal point of the image. The viewer’s eye next follows the yellow fish down to the orange crab, the second focal point of the image. Everything else is intended to blend together. The two focal points are framed by the straight lines on either side.
It might not be perfect execution, but my eye does follow the exact path I planned out compositionally, so I’m happy with it. The orange crab should be less centered and further to the left, but making major changes at this point is not feasible without a Threadripper
I agree. The terrain was a pain- my tests of maintaining stylistic consistency and capturing all the terrain details overwhelmed the image and made the terrain the focal point. My original terrain shader was very detailed and complex but it ruined the composition. I eventually decided that the final image was more important than the terrain details. I may have overcorrected a bit, but I could probably spend the rest of my life tweaking this scene and still never get it 100% perfect
Yeah, I wasn’t very good about posting updates in my sketchbook during the second half of the process Documenting a large project like this is fairly intensive, I really get why a lot of artists just drop an image with no accompaniment. I do enjoy documenting progress, though, because it gives me a lot more chances for feedback, so I’ll be better about that
Documenting a large project like this is fairly intensive, I really get why a lot of artists just drop an image with no accompaniment. I do enjoy documenting progress, though, because it gives me a lot more chances for feedback, so I’ll be better about that 
Just - poof! - came, saw, conquered.