Wine Carafe

Goals

My goal was to push my sculpting skills harder, to train my eyes to read and understand volumes more properly. This would build upon what I had learned with the first prop, which was a lesson in hard surface modeling. My mentor presented me with many different choices of crazy and complex props, and naturally, I drifted towards what I thought would break me.

The carafe became everything I thought about until it was done. It took a very long time, and I learned so much through this grueling journey.

Initially, I had two images of the carafe, which we had found on Pinterest in low resolution. While discussing the sculpture with my friends on stream, we were jokingly calling it anything but a carafe, to the point that “Teapot” became the title of every stream. Wanting to settle the debate on the name, I jumped on www.karaffensammler.at which was stamped on the two images.

To our surprise, we found several more, higher-definition images, and I realized I had been sculpting some features incorrectly. Wow, doing your research early is so important! That was another unexpected lesson. The entire process was done on-stream, www.twitch.tv/the_mortanious , where I saved my VODs for people to watch if they do happen to be interested. This has been very humbling, as I do get a lot of people asking for them.

Blockout

I started the blockout in Maya so that I could measure the carafe and model it to scale. After getting the measurements, I started to model the glass. I tried many different ideas that didn’t work out or look clean.
I couldn’t Zremesh, Dynamesh, or even Decimate it in Zbrush because all the results weren’t clean enough.

For example, the cube cuts were getting muddled. I was also afraid it wouldn’t bake correctly later as well. It was so tricky for me. I overcame it through poly modeling in Maya. I get asked a lot on my stream about how I modeled the glass, and how I achieved such a clean look for the cuts. The way I did it was to create one section of the cuts, on a flat plane, prepare the edges and faces to extrude, selected the faces, and multiplied it by the measurements of the cuts on the glass bowl, and then welded them together.

After welding, I went into polygon mode where the faces were still selected, then I saved them as a quick selection set.
I set the glass bowl as a live surface and conformed the cut plane to it. I then loaded the quick select set and extruded the cubes out. The rest of the shapes were done in Maya as well, then taken to Zbrush to boolean.

Sculpting

As for the sculpting, it was a long process… I really learned everything I know throughout sculpting the model and attempted every section multiple times, especially the Venus character on the handle. My anatomy knowledge was surface level at best, and I had to expand my knowledge and research on anatomy.

The artist who made this piece back in 1880 definitely knew what he was doing. All the faces and characters, even though somewhat stylized, because of the confines of the media, were still grounded in real anatomy. I divided everything into sections. I would sculpt and move on believing it was done, but by the time the next section was complete, my skills had evolved beyond everything I had done up to that point, and the previous section didn’t match the rest of the model and looked horrible.

I studied some anatomy, as anatomy really is an endless study, and collected a lot of references that involved similar poses to Venus on the handle. In addition, I had to research jewelry smithing. This included the way silver was molded and sculpted in reality, with respect to the time period and the different techniques used like stippling and chasing. I mimicked the way things were done.

For example, the chasing was a manual carving that was done groove by a groove from the side. That was done as an efficient way to achieve rows of peaks and valleys and not waste silver, as it is an expensive metal.

Stippling is done to darken the negative spaces between the filigrees to push the contrast of the carving. Those were straightforward sculpting techniques, with the stippling being an alpha that was created to be stamped between the reliefs. That alpha stamp was made using Dam-Standard on a plane with a swirling motion, that was consistent with size and direction until the plane got properly worked.

For the filigrees, it was more of a challenge to keep things clean, to not warp the surface around the filigrees. They had to stay clean because I had to add the stippling later, and the noise and warbly surface would give a messy result. It was a simple, yet very effective solution to mask the area and take my time to get the shapes correct, then flip the mask, go into Deformation > Inflate, but very gently, and then inflate the balloon about 0.1. Finally, this was followed up by a combination of Alt-DamStandard, Claybuildup, and HPolish.

Low-Poly/UVing/Baking

For the Low-Poly, it’s honestly straightforward and was done in Topogun 3. There are some guidelines one would follow and hold in the back of their mind while solving the puzzle of keeping the polycount low, maintaining the silhouette and adding edges where necessary.

Have a very clear and readable edge flow, keep your edges relatively the same size throughout, and take care of the peaks and valleys. UVs were done in RizomUV. Seriously, nothing matches RizomUV with its unwrapping tools, options, and efficiency. Beyond highly recommended, I would use nothing else.

For baking, I used Marmoset Toolbag 4. The ability to see your cage, paint the variations in the height and paint skew should be standard everywhere.

Materials and Texturing

Texturing the carafe, at first, seemed way more daunting than what it really was. Keeping things simple and tackling materials one at a time keeps my thoughts from bouncing all over the place. First off, Smart

Materials are a big no for me. That’s what I was taught and what I truly believe now that I’ve experienced efficiency and difference first-hand. A lot of people are content with dropping a smart material and lazily masking areas, but it just shows lazy work that will look like every other project out there. It is very important to be capable of understanding the materials yourself. Smart Materials are materials made by other people, and that is like loading someone else’s Saved Game and picking up where they left off.

It is really important to have good-baked maps to start from and having a very high detailed high-poly properly paves the way. Most of the damages, dents, areas eaten away by corrosion and time, imperfections in casting, and variation in the properties of the metal were done through sculpting. The bakes created Normal, Curve, and Ambient Occlusion maps that have a lot of information which helped the generators create a good base to work from.

Something to note, which really used to mess with me, is setting up Substance Painter and Marmoset Toolbag to look the same in materials, roughness and color space. So, I took the same HDRI map from Substance Painter and put it into Marmoset, Studio Tomoco; and set the Tone Mapping to ACES.

As you can see, the base material is composed of roughness and color variations that will be the underlying affecting base throughout the entire group for each particular material, (not including the glass.) While working on the color variations, for example, colors started to make more sense and material breakup became clearer. If smart materials were the go-to, the brain wouldn’t be forced to break down the materials and layers, and get complacent, when the purpose here is to grow and become a better artist.

It is important to build unique-looking materials. Generators made from the baked maps make a huge difference. Having a strong base to build from makes life a lot easier and helps with the unique look of the overall texture. After building the overall look of the material with the generators, I masked out the areas I didn’t want to be affected and then hand-painted the mask for more unique variations.

The glass was worked on in the roughness map for the most part, as crystal glass is clear. The scratches, fingerprints, and smudges are all roughness variations. On the reference, there were some deeper cuts and bubbles that form from manual glassblowing. To make these, they were added and masked in the heightmap.

Presentation: Lighting and Rendering

Marmoset Toolbag is a great place for presentation, very similar to what you’d be getting in Unreal Engine. That’s important for game assets, as this is what the carafe was technically made to be. Mainly, I followed the “Rule of Thirds” for the presentation, to present it as an art piece and at the same time, show off all the hard work I had done.

I did a lot of test renders, which was honestly a lot of trial and error, treating it like photography, using different lights and camera angles, and then selecting the ones I liked the most. I tried several HDRI maps, setting up lights and placing them where the light source would be in the maps, then testing the angles with what had the best contrast while keeping true to the metals of the model.

Conclusion

All that’s left to say is, that if you love the process, you won’t get burned out and let it consume you. Don’t forget to take breaks and take an extra day off if you’re tired. I certainly wish I gave myself a break here and there.

It’s not burnout from the work that gets you; it’s the feeling of not stopping to take a breath that makes you feel burned out. Ego should be checked at the door, and one’s heart and mind should be open for critique to allow for the building of skills and knowledge. Also, remember to have fun. Most of us got into this because of video games, and video games are about having fun!