Pumpkin PBR Refresher Project - Concept + modelling

As a start to this academic year, the first project will be a quick 2 week project to make a carved pumpkin using the high poly to low poly workflow we did last year for the fireplace project. The Tris limit was 3k tris, and we had a limit of a 2048 texture set and a 1024 texture set to play with.

to begin, I made a quick mood board of what pumpkins looked like, and some other artworks that had made some successful pumpkin models. 

Concepting:

Pumpkin Mood Board

The core idea I had was to make a pumpkin tortoise, since I wanted to push myself in this project to not just create a simple carved pumpkin (not that there is anything wrong with that, I just have a habit of overthinking ideas). I found reference of what tortoises looked like, and the structure of pumpkins. I also found a piece of some stylised, two headed pumpkin turtles, and used this as inspiration for my quick concept sketch in the bottom right of the mood board. this idea I had was that either the tortoise had just taken the pumpkin and grown out of it, or that it was made of the same "wooden" stem material throughout. Either way, I thought the idea was cute and I had an idea how I was going to go about sculpting the tortoise. The concept felt almost like I was making a high resolution Pokémon, like a grass/ghost type version of the Pokémon Torkoal, so I kept that in the back of my head as a fun side angle.

The other tortoise piece by Morgan Hilber can be viewed here:

https://www.artstation.com/artwork/v2YQPA

Sculpting:

We all started the creation of our pumpkins in Zbrush high poly immediately, since the shape of a pumpkin could be very easily made using basic shapes and symmetry. I followed the basic tutorial on making the pumpkin, using the shell workflow to mask out the areas that the legs will be and I also added a pair of concerned eyes on the pumpkin to sell the carved pumpkin idea more. I then began making the other elements  (the head, legs, stem and tail) using symmetry. 

1. I made the head and the neck separately and then combined them through merging and Z-remeshing the shapes to remove the seam, and this meant that the geometry of the sculpt was kept consistent. It also meant that before attachment I could do the same shell workflow, this time cutting out the shapes of the eyes on the head. I added a wood grain on the underside of the tortoise's head and neck to try and add a bit of texture to this area too.
   
2. The legs were made by first making a single leg out of one cylinder, and then duplicating that leg 3 times to make all 4. I could line them up fairly accurately because of the holes made in the pumpkin flesh before, so I wasn't too concerned that the legs did not line up perfectly.
3. The stem was made using the tutorial's method of having a cylinder that I then blended in as the top of the pumpkin, and I used Z-remesh quite a bit to make sure that the geometry was consistent.
4. The tail was a very simple tapered cylinder too, however I had a few issues that needed Z-remeshing to fix at the tip of the tail as the geometry got too dense.

The sculpt at this point looked like this:

Basic Sculpt (apologies for the glare it was a photo not a screenshot)

I didn't add the wood grain texture on the legs at this point as I wanted to add the detail asymmetrically, and if they were already textured it would make it harder to add the individual wood grain. So my next step was adding the wood grain per leg, as well as further detailing the wood knot in the middle of the head.

Final High Poly Sculpt

After making this final sculpt, the model was sat at a few million tris. I decimated the model, and it eventually comfortably got down to about 63k tris ready to be retopped in 3DS Max. In hindsight, I could have kept the model at a higher fidelity here since I encountered a few errors around the legs, but this did not really impact the final model so I am thankful for that. A lot of the tris were concentrated on the inner shells of the main pumpkin and the head, so I was able to massively decimate those assets in particular to make the total a lot smaller since they were only seen in some small pockets.

Decimated final - some errors can be seen on the top inner corners of the legs

Retopping:

I retopped each core piece separately, however in areas like the cut-outs and the eyes I merged the two meshes together so I could retop the spaces as one full object. these cut-outs and the eye would end up being emissive later and I was not retopping the interior of my pumpkin which would massively save on tris later. Areas like around the legs or the emissive cut outs were very useful as even if they had some bake errors on the normal or AO maps, these would be voided when the emissive was applied since emissive does not take in lighting information, it outputs. I just had to make sure that most of the emissive areas were flat with a surrounding edge to make the silhouette of these flat areas not so noticeable.

In all, the retop came to just over 3k tris, which through some trimming in areas like underneath the pumpkin which will be barely noticeable I was able to get the model down to 2,960; comfortably underbudget. This was quite easy to do because I could mirror changes made on one side of the pumpkin to the other as the pumpkin was mostly symmetrically retopped. The legs, although unique in the high poly, were simply one leg that I duplicated multiple times in the stead of the other legs and then I could use the conform tool with the right high poly leg underneath to make sure that the high and low polys aligned.

Final Retop

I managed to also save geometry by no adding loops to the joins of the tortoise. I did consider rigging my model during this process however at most I would simply move the head and the neck at an interesting angle, and translate the legs a little bit in a pose. This was not given too much thought because rigging was something I really hated last year with my gladiator and that had a strict tutorial to follow, so I would do that if I had a lot of time left and I was already pleased with the result.

Conclusion:

That marks the end of week 1 of this new project! I am really excited for where this will go through texturing and presentation, and I am happy that I finally got around my fears with Zbrush as at the end of year 1 I was avoiding Zbrush every time I could, I still see it as best for organic objects, and after hearing that soon we will be learning substance designer I am incredibly excited to have to do less finicky work with meshes and more finicky work with nodes, since I think my brain gets a lot of dopamine from having something that works in front of my eyes and doesn't need to painted.