Ornate jewellery box

Well, that's another class biting the dust... The final-final assignments are the worst ones. I don't know about you but I always feel the pressure of having to go beyond whatever I managed to do until that point. Which usually means that I spend way too much time on choosing what to do, then I panic and cram ridiculous amount of work into the remaining few days. Did it go differently this time? No. No, it didn't.

I started on Thursday, not knowing how I was going to approach the tasks at hand. I'm not very good at sculpting details (which is, incidentally, one of the reasons I picked this jewellery box) so I needed to make some sort of plan beforehand. But since it was Thursday already, that plan consisted mostly of "do whatever you can but make it quick".

So I built myself a basic box that matched the real-world dimensions (this is the reference I used). Then I kept messing around with it because the ornate panels that I "glued" to it didn't want to fit in nicely. I only changed the dimensions of everything a few hundred times... In a slight state of panic I jumped right into the sculpting phase. I knew right off the bat that retopologizing was out of question, so I decided to go with the MultiRes sculpting. Haven't had much practice with that, I thought, so that was going to be another area I could improve.

It took me a whole side panel (the long one) to get used to it. The gist of MultiRes is that you are not generating new geometry with brush strokes, you only use the subdivisions you've created. And if you subdivide, you subdivide the whole thing. To get the level of detail I wanted, I had to go as high as 7 subdivisions. (For those of you this is just a number, know that it is a really high number that makes for a really laggy viewport). But I traced the shapes nonetheless. Another side effect of a mesh this dense is that you can't see the background image very well. So, technically, I was tracing the pattern, realistically, I was just winging it.

The funny thing with skills is that you don't usually notice when you get better at something. But as I was working on the top panel, I suddenly realised I was tracing the lines with single, uninterrupted strokes. There was something really enjoyable about being able to guide my hand exactly where I wanted it to go. Especially since I'm not very proficient with a tablet. But for whatever reason, I managed to do the whole top panel in a fraction of time it took me to do the front one. 

Blender 2.8 matcap - side

And then I came to a halt with the side panels. I had no direct reference for them and the two images I could use showed the side from an angle. So I had to manually pick various elements from the views I had and stitch them all together. That involved a *lot* of the background image shuffling. And, since it was in real-world scale, the movements had to be minuscule. A fantastic exercise for long winter evenings. Which I didn't have. So, in a severely heightened state of panic I started to rush things a bit.

At which point I ran into the corner ornaments. Or rather, the corner ornaments ran into me, over me, trampled me some more for good measure and let me to despair. You see, they are not too difficult shapes to make, in theory. They are just cover pieces meant to hide the panel seams. In reality, they are impossible to get right. I tried everything I know. Box modelling, MultiRes, DynTopo... Nothing. Only an abomination after abomination. The shape I eventually managed to squeeze out of this came through all the stages I mentioned. I modelled a simple L-shape and I Multi-Resed the ornament on it. Then I applied the modifier, deleted the extra faces, manually connected the gaping hole that ran through the whole side, and then I smoothed the horribleness with DynTopo sculpting. Which naturally meant that I got a mesh that was too dense for UV unwrapping, so, hello retopology on Saturday night. 

Blender 2.8 matcap - top, detail

I was so flustered from the ordeal that I just had to finish it on Sunday. Which was probably for the best because then I got it done rather quickly. The only thing that remained was UV unwrapping, texturing and rendering. Piece of cake. I textured it in Substance Painter even though I played a bit with the options I had in Blender (using dirty vertex colours to create a map for assigning metal and dirt parts of the texture). The SP did try its best to make me regret it because it crashed on me - twice - and every time after thirty minutes of tweaking, as I was getting to where I wanted. 

In the end, I managed to tame it and I learned a few tricks in the process. I also have a new smart material that I can use in future projects. Speaking of future, though, there is one extra reason why I made this box. And that is: I can show you next week what's inside it. Are you curious? Well, you should be. If everything goes well, it should be one of my passion projects. See you next time!