Rescaled Bricks & Greenhouse Dragons

I promise we’re going to get to sculpting the decorative dragons from the tops of the greenhouses, but first I wanted to share a “before and after”. A reader who goes by kronkolweg pointed out that my castle’s bricks were a little big. It had already been starting to bug me a bit, and this was just the kick I needed to get off my butt and fix it.

For the original castle miniature, one or more brick textures were hand-sculpted, molded, mass-produced in FastCast resin, and glued to the miniature’s wooden understructure before being hand-painted. There was a transition to more detailed plaster-based textures starting in Chamber of Secrets – I’m not sure if any of the original resin brick textures ultimately survived through to the final Half-Blood Prince iteration of the miniature. In any case, the overall effect is the same, and my procedural brick material mimics it without requiring me to individually texture each wall. The thing is, when I created this material (as documented here, here, and here), I had much less of the castle built, which meant I had fewer points of comparison for the scale of the bricks. Frankly, I don’t remember how much actual measurement there was, as opposed to simple eyeballing, but it’s become clear that the results are a little oversized.

So, being the Excel nut that I am, I sat down and made a spreadsheet. I took horizontal and vertical brick counts for various areas of the castle, comparing my model to the miniature. There’s a certain margin of error in the scale of my castle, and the bricks of the miniature are not necessarily of completely uniform scale, so I averaged the different measurements and arrived at my brick adjustments: a 34% reduction in width and a 21% reduction in height.

The interactive sliders below compare the old larger bricks (left) to the new smaller bricks (right):

The difference is subtle from any real distance, but it’s more accurate now, which always feels good.

With this tweak completed, I shifted my attention to those greenhouse dragons! Fortunately, they’re all identical, so I only had to sculpt one. I did so in a separate file, and – well, if you’re interested, just watch part 1 of the video.

Frankly, I can’t get enough of these sliders, so let’s compare the base mesh to the full-resolution sculpt…

…and now comparing the full-resolution sculpt to the retopologized version with normal mapped details:

If the differences seem really minor…good! The point of retopo is to vastly simplify the geometry, which makes the asset a lot more manageable. The sculpt on the left has over 1.3 million faces; the retopologized version on the right only has around 1,500. Much easier on the computer! Of course, that eliminates a lot of details; baking and applying a normal map is a way of faking those back in, hopefully with an end result that’s nearly indistinguishable from the original sculpt. (In this particular case, there are a few small artifacts in the shading, but these are not noticeable when color is applied and we move the camera away a little.)

You can watch this whole process in part 2 of the video:

The retopology process is still a lot slower, harder, and more frustrating for me than I’d like, but in the end I was satisfied with the result, so I created an appropriate material, brought the mesh over into the main castle model, and duplicated it to the appropriate spots on the single greenhouse I’d already created. Voilà!

That’s a wrap for the dragon, and for this greenhouse overall! The next steps will be to duplicate this greenhouse and create the other two design variations on it. Then I’ll be able to move on to the domed conservatory, the curtain walls surrounding the greenhouses, and hopefully some ground cover and footpaths. Be sure to subscribe to this blog and the YouTube channel for updates as they come!

The Clock Tower Courtyard

Let’s return to the clock tower courtyard. With the fountain in the center finished, I decided to now enclose it with the courtyard itself. Existing as a full-scale set and as miniatures at 1:24 and 1:10 scales, the courtyard was designed for the third Potter film and is surrounded by aged cloisters and overgrown vegetation. There’s a lot of complex geometry to capture. I compared photogrammetry, production stills, technical drawings, film frames, shots from the Warner Bros. Studio Tour, the similar courtyard in front of the castle in films 4-6, the larger version thereof in the final film, etc., etc. Eventually, I felt like I had a decent understanding of the space.

One thing I’d never noticed before was that there are actually two different designs to the arched tracery of the cloisters:

The one on the right is taken directly from Durham Cathedral, lending stylistic unity with the location shoots from the first two films. I’m not sure if the one on the left is taken from a real-world location, but it’s certainly very similar to a lot of real Gothic tracery out there. That one is used primarily in the sections of these cloisters without a roof.

As with the fountain, my first task was to create the courtyard in a more-or-less pristine state, without any of the damage or erosion to be added later. (Adding the ground is a logical last step so I don’t have to worry about it blocking my view of the blueprints beneath the model.) Notice how the last arch on the right has the Durham tracery design:

The courtyard has elongated gargoyles placed at intervals around its interior, and while they look similar, there are actually two designs that alternate: a hippogriff and a centaur, although they’re so stylized that I had to study them for a while to figure out what they were depicting.

When they repurposed the clock tower courtyard set for the viaduct courtyard after Prisoner of Azkaban, they kept these gargoyles in place, providing a lot more visual reference. (Interestingly, while the gargoyles also made it into the larger version of the viaduct courtyard for Deathly Hallows, their features are much more sharply defined and less eroded in this version. I decided to split the difference for the purposes of my model.)

For the centaur design, I even had the benefit of a pretty accurate orthographic drawing published in The Art of Harry Potter. I based my Blender sculpt on that drawing and then baked a normal map onto a simplified version, much like I did for the fountain statues, although I didn’t bother with manual retopology – I knew this guy would never need to be rigged, posed, or animated, so I just let the Remesh modifier take care of the simplified topology for me. Anyway, you’ve already seen me go through this with the fountain statues, so instead of boring you with similar images, I’ll share this unique WIP view of the centaur sculpt from the inside:

Enjoy your nightmares.

Anyway, here’s the completed centaur design in situ, with some procedural noise layered onto the normal map baked from the sculpt:

I made the mistake of not putting much work into the base mesh before sculpting this guy, so there were some areas of the sculpt that got dicey. The hippogriff fared better because I took the time to create a proper base mesh, like I had with the eagle sculpt. See, what a cutie! (I didn’t have original ortho drawings for this one, so I was wingin’ it. Heh.)

I tried to again use the Remesh modifier for my lower-poly version, but it didn’t work as well on this fella. Sharp parts like the beak and claws did not work well at all. Back to manual retopology! Oh well, it’s good practice, and a good opportunity to improve upon some of the poor topology on the eagle statue. My brain is not wired to easily just throw down perfect edge loops.

By the way, if the term “normal map” doesn’t mean anything to you, here’s what the one for the hippogriff gargoyle looks like:

The different colors essentially tell the software to pretend the surfaces point in slightly different directions than they actually do, thereby simulating details that aren’t actually there. Blender “bakes” these maps by comparing the surface of the simplified mesh to the original sculpt.

Here’s the end result:

The tricky thing with this one is that the front legs seem to have fallen off many of the gargoyles in the clock tower courtyard and the original viaduct courtyard, but the ones in the larger Deathly Hallows viaduct courtyard are intact. I think I’m just gonna keep the legs for the time being.

Here’s where we’ll leave off for today. In a future post, we’ll build the remaining walls of the courtyard and damage the appropriate areas.

Finishing the Ruined Fountain

Okay, time to wrap up the fountain in the clock tower courtyard from Prisoner of Azkaban!

First up – finishing the statues. I believe the original Mexican symbology has the golden eagle perched on a prickly pear cactus, but the base of this statue appears to be a nondescript rock, and that’s what I’ve modeled it after. It was fun to glob that together with Blender’s “clay strips” sculpting brush; I also rigged and posed the eagle’s mouth and claws to hold the rattlesnake, whose body started as a squiggly Bezier curve with a circular cross-section. I enjoyed trying to match the sinuous, energetic gesture of the original.

The snake’s head was sculpted separately with its jaws already fully open…no need to worry about rigging and posing this fella. Good reference is scant for the head; as with the eagle, I relied on photos of real animals to help me fill in the gaps, but it was surprisingly hard to get a good “likeness.” Oh well…at least I planned out the retopology a little better than with the eagle. That enabled me to graft the head onto the body and bake a normal map from the head sculpt. He still looks kinda funky without fangs, but they appear to have been weathered away in the film.

With that, the only things remaining were integration with the main model and damage/erosion! Not too shabby-looking, right?

Integrating this into the main model ended up being very easy. There are four statues: one at each corner of the fountain. As I placed these, I spotted some angles and proportions that were a little off, but I suppose that’s the downside to working on the statue in its own file. Oh well. The complete fountain still looks pretty cool!

I suppose it’s a little misleading to describe the fountain as “complete” when it didn’t actually have…you know…a fountain…so let’s run some fluid simulation!

Not great, but it’ll do, especially because we won’t normally be this close.

The last step with this fountain was to destroy it a little. There are big chunks of it that have crumbled away. After creating some jagged meshes to cut these away with Boolean modifiers, I arrived (after over 2 weeks of on-and-off work) at the complete fountain!

Now it just needs a courtyard to surround it!