Showing posts from December, 2019

PROJECT 3: Unwrap & Texture

Back to the fun stage! Finally! Well, not the unwrap per se, that's part maths, and maths is evil. But texturing! I studied the brief a lot for this stage, and I'm glad to say I think I learned the most out of it as well. Long gone are the times where I trial-errored inputting the roughness/metallic values in thousand times because I did not understand what any of them does and I cried myself to sleep not knowing where I went wrong when I finally put my assets into UE4 and they all just looked like dark, shiny unicorn poop. I finally got it! The white means matte, the dark means disco balls!

The source of my epiphany was mainly me trying to follow Matt Wilkins' amazing breakdown and working on the textures in Photoshop. Breaking everything down to Photoshop layers helped immensely. I helped myself out by importing my bakes (from Substance) to 3DCoat which has an awesome bridge with Photoshop, so I was not in blind and I saw every change I made on the texture in the viewport immediately (and could handpaint on the model directly as well). I still don't get why Substance only has an "Export-to-Photoshop" feature, biiiiiig let down. 

As per Matt Wilkins' suggestion in his breakdown, I created a tileable 'dirt-colour-interest-detail' texture which I then used in almost every asset (by just changing colours via PS's Selective Color). Big help in wood texturing, more subtle in fish skeleton bone, added great hints of hues in the golden bird cage and overall made the process much more enjoyable than if I had to add interest individually by hand to every asset :) I still did some overpaints because sometimes it was hard to adhere the texture to the UVs (note for next time: this is why you should assemble UVs as straight as you can, and if possible, all in the sime direction!). ¨

Another thing which blew my mind was the use of Curvature Map. I thought I had lost all the details from Zbrush when baking, but this thing - man, this awesome thing! It just brings it all back! It took me some time to figure out the opacity and blending mode settings and in some cases, I had to clean it up manually, but if there was an MVP award for most useful texturing trick, Curvature Map takes it home, no question. With lacquered wood and some other assets, I boosted the curvature map by lassoing over some wear and tear and the result was even nicer.

In the brief it was advised not to add AO to textures, but with some assets I went and did it anyway, just very subtly.

At the end of the texturing process, I brought all assets back to Substance Painter just because I like the shaders and rendering there more. I exported the final Targas from there as well, with Roughness in Blue channel, nothing in Green channel, metallic in Red channel, as it was asked of us in the brief. Overall, this must have been the most fun I've had when texturing in a while and I would love to work in this style once again after I finish this project!

PROJECT 3: The devil named Retopo

Off to one of my least favourite parts, retopology. Where are the clever robots when we need them? Even Elon doesn't like spending much time on it, judging by the complexity of the cyber truck (which I btw love, you go Elon), so I don't get why we still have to endure this torture in this day and age. 

For some of the assets, mainly the organic ones, I used a combination of ZRemesher and hand retopo-ing. I found ZRemesher provided great results on wider areas, not so great results on pointier or more detailed areas (even when trying to retopo with polygrouping, ZRemesh guides, topology brush and other thousand "shortcuts" I looked into before giving up and, defeated, coming back to my eternal frenemy 3dsMax).

Beginning of my bird retopo -> blue is Zremesh, rest is by hand.
Final retopo

For other assets, I already had a pretty solid basemesh and in those cases, it became mostly about adding loops and conforming to topology. I constantly tried to keep in mind my budget, which was (according to the SoT brief, 2000-4000 verts per object). 

Overall, I had no major issues with retopo as the assets weren't super complex (apart from the fact that it was a boring process). I "watched" a whole season of Brooklyn Nine-Nine while doing it so I can't complain THAT much anyway :D Though in retrospect I am realising, if my sculpting workflow was cleaner with some of the assets, such as the fish skull, I wouldn't have had to spend that much time on its retopo (I left a lot of stuff dynameshed, where I now think I should have Zremeshed, subdivided and sculpted over it again, well, I learned for the next time). 

PROJECT 3: More sculpting

Sculpting the environment wasn't as tedious as I thought. I was afraid of the cloth at first but after studying some Sea of Thieves reference, I found that the cloth was even simpler, flatter than I thought, so there was no need for overdoing it. Still, I feel like I could have done better with some of the cloth shapes, but I did not have time to do studies and learn proper cloth dynamics. I am noting this as a thing to focus on later in my post-uni projects though! One thing that went better than I expected were the floor tiles. I only sculpted on the few pieces of wood under my cloth (around 4 or 5 of them), which I then re-shaped, re-sized and dynameshed to create the floor texture, sculpted over those a little and that was it. If the floor was made for a game environment, not just a small diorama, I would have probably created a tileable texture out of it, but I saw no need to do it in this case.


My sculpts:

I re-used the planks once more when sculpting the box/crate asset. On the left you can see how so (assembled the previous plank sculpts, then sculpted over them to make them look more similar to the concept).

One asset I was looking forward to sooooo sooo much was the little parrot. I found so many amazing ones in the game and I couldn't decide which one to pick. As part of me trying to be more active on my art instagram (, I decided to ask my followers to help me pick the design I should go for. The parakeet won over the classic parrot by a small percentage :)
When sculpting this one, I did not have any major issues, it was mainly just time consuming (but the most fun one as well).

PROJECT 3: Sculpting fish skeleton and additional assets

A small update on my sculpted assets. I feel like it's either taking me way too much time because I haven't sculpted in such a while, or because I'm naturally slow, or because I'm putting too much time and effort into this stage (if there is even such a thing as too much). Now that these main bits of the scene are done (apart from the parrot boi, leaving the best for last), I want to work on the rest of the scene very fast and learn to not be too precious about secondary assets. 

All in all, I do like how the sculpts turned out. I got back into the flow of using my shortcuts and customised UI, I remembered which brush is best for flat areas and how to fill some areas in without disrupting the flatness (hPolish + alt rules) and I genuinely enjoyed the process. I am thinking of creating an engraving for the plaque on the bottom of the fish skull, though I'm not sure what it would read. Maybe a badass name of a pirate who captured the striking specimen? :D I'll think about it and maybe add it in before I start texturing.


Lastly, a small blooper from the basemesh preparation stage. I've been goofed on so much by both 3dsMax and Zbrush during this whole process :D

PROJECT 2: Creature concept brief

PROJECT 3: Basemeshes & ZBrush

For the past week I have been mainly working on the three main assets for my SoT scene and well, it's been a ride. I've forgotten a lot about hardsurface modeling and in retrospect, I probably should have not picked such diverse subjects for my matrix -> picking organic character, 2D concepting and then 3D prop/hardsurface meant that I had to switch into different modes every few weeks and remember different things about all I knew in each discipline numerous times and that took way more time than I had at hand. Good thing is, for my FMP I've chosen a pretty consistent topic so I have hopefully already learned from this mistake! :D 

Anyway, the stage where I had to prepare the basemeshes for Zbrush was by far the most difficult at the start. Meeting the old friends called NGons, rat nests and control loops was a memory I'm not eager to revisit; nevertheless, I pushed through it, frontloaded my meshes as much as I could to save myself some certain tears in Zbrush and I'm happy to say that I think it was worth it, as the sculpting stage was a lot of fun and reminded me why I don't hate 3D as much as I thought I did. 

The following are screenshots from Zbrush where I played with materials and render settings to get an idea of what to shoot for later on in Substance:

Above: some alphas I created in Photoshop to add the engraving from the concept.

Above: Some WIP shots from Zbrush and 3Ds Max. The arch was especially challenging as I didn't know what method to use for it. I first tried to connect the edges and keep it as one mesh but after re-visiting the concept and hunting for some in-game screenshots of the actual modelled cage, I noticed that the bars are overlapping the arches and it would be easier to have them as separate meshes.

The gun took some figuring out as well. The concept wasn't as descriptive as the others and I could not find any indication of size specs, so I tried to guess the proportions and size of SoT characters and kind of wing it according to those. I also found some nice Blunderbuss model screenshots from the in-game shop. Not the exact one I was supposed to model but it was helpful in some minor things like the shape of the inside of the gun's head or the fidelity of detail in materials. Renders from Zbrush: 

High poly stage, turbosmooth in its full beauty.

The fish skeleton is still in a bit more of a WIP stage. I put a lot of time and effort in trying to follow the concept as close as I could to achieve an appealing shape. Now I'll get onto sculpting the teeth and adding wear&tear to the rest of the skeleton. I'm now also deciding whether to take these main assets to finish now (retopo, UV, bake, textures, etc.) to have them finished earlier (but FINISHED), or if I should try to speed sculpt the rest of the scene to keep it all consistent.