Why Does God of War's 3D Environment Art Looks Like This?

[@EMC3D]

We were able to have a long and awesome chat regarding topology and environment art with the chap who made this level in one of our recent videos, check it out! kzbin.info/www/bejne/np2pgWduZdh6sLc

[@PedroVencore]

Every hard edge needs to have a seam but not every seam needs a hard edge , thats how it goes. Excellent content thanks, finger crossed in the future wireframes become more common , i don't see these anymore

[@EMC3D]

Ahh thanks so much, I always mixed up the saying :D!

[@shmuelisrl]

to answer your question. blender has a "mark sharp" that splits normals, and the default is what you call soft normals (as long as you set it to smooth normals). the overlay is in edit mode and is by default blue (but could vary based on theme and preferences). the overlay is only in edit mode afaik. the object mode wireframe doesn't show sharp edges, neither do i think it should.

[@FishMan1nsk]

Regarding why you need to have soft edges on flat surfaces Id say everyone need to understand rule #1: "You can store only a single piece of information in a single memory cell". Because it gives you much deeper understanding of what is going on. And if we talk about vertex information you can store normal, 3D coordinates, UV coordinates and vertex colors. But only for a single vertex in a single memory cell. So in order to get different smoothing groups you need to store two different normals for a single vertex which is impossible because of the rule #1. Which is why every time when you do a hard edge - the amount of vertices will be duplicated in order to store two different normals and DCC software doing this automatically, so you dont have to think about it. The downside is that artists are not even aware that this is happening. Artists have this distorted perception like thinking that its a single vertex because you can see and select only 1 vertex on the screen. This is because DCC software like Maya and 3Ds Max are grouping this vertices into a single vertex which you can select and manipulate (but this is just a data representation, not what is actaully happening). There can be actually more than one vertex under the hood. Game engine is not a DCC software - there are no edges for example or welded geometry. Everything "not welded" is actually a duplication. Every duplication is "not welded". Actually the same thing happen in DCC as well but only under the hood. DCC has this visual interface and different concepts for you to easilty manipulate the data which distorting your perception. Sometimes (most of the time actually) you can get the illusion of connectivity while having some difference in data (lets say two different vertex color in a single vertex) this because you have the same data in a single point in all other channels (in our case - same 3D coordinates, same normal, same UV's but different vertex colors). But there actually two vertices there which are not "welded" in the traditional way of understanding this concept. Because welded and unwelded, edges, quads etc - is a DCC concept not 3D in general. In engines everything is triangulated and there's no such thing as "edge". I could even tell you that there's no such thing as polygons. Because polygon is just a vertex enumeration. All what exists in memory are only vertices and lists of vertices. If you understand the concept of storing data in computer memory - you can easlity understand this "hard/soft edge optimization" if you can call it like that. Basically any difference in data in a single vertex will produce a duplicate of this vertex because there's no other way to store it. Because you cant store two different data's in a single memory cell.

[@EMC3D]

Damn! Thanks for the awesome knowledge drop chap, really good info here.

[@jaze_ph]

Danggg... Thank you for explaining this in depth. I'm a 3D Artist for almost 10 years now, and understanding this technical aspect of the process fascinates me. Hope you can share some other resources to read/watch that focuses on technical stuff like what you explain. Thanks

[@DrTheRich]

What flows from this, that you didn't mention, is that because of this you can get bevels for 'free' in the sense that they don't cost much more memory and calculation than an already hard edge. Ever since I learned that I never use hard edges anymore, all my hard edges are beveled edges with face aligned normals. I first learned this from a devblog on star citizen.

[@franksalot114]

@@DrTheRichcould you link the devlog? Would be much appreciated if possible.

[@FishMan1nsk]

@@jaze_ph Not really. I don't know any specific source. However you can try to learn how to program this stuff. For example you can try to do some plugin or a script for your software and your will learn how the data looks under the hood. Because this is exactly how I learned that by trying to make something and reading the SDK\Scripting language documentation. By the way another fun fact which not many people aware about - in DCC UVs and Vertex Colors are the same thing. It just the index is negative (if I remember correctly) for vertex colors and for UVs it is positive. Its not the same in Unreal for example beacause in Unreal UVs are vector2 and color is vector3. But in 3Ds Max for example (and probably in Maya as well) VC is the same as UV. At the end of the day it is just 3 float numbers. You can interpretate them as 3D coordinate as well. Whis is how people stroring pivots for example for different parts to animate trees using vertex shaders for example by storing pivot location and forward vector in vertex colors and additional UV channel or a texture. And because normal vectors are always of length of one you can always derive z component from the equation of vector length (default unreal normal maps are stored in this manner for example). Anyway just trying to encourage people to learn this stuff ))

[@natestephens13]

Hey, I made that level! That was for PS3, and the topology isn’t really relevant anymore. But the reason for all the cuts was also for lighting, in addition to vertex painting. We used to bake the lighting to the vertices, so we would have an average spacing of a vertice every meter for lighting resolution. It did all have to tie into texel density also, so there were lots of considerations to combine and manage. Things are quite different now, so let me know if there is anything I can elaborate on.

[@EMC3D]

Hi Nate! Thanks so much for the knowledge pass, and the awesome wireframes. If you're up for it we could form a chat around the wireframe images as a video in the future, just let me know, always happy to share/learn knowledge :).

[@natestephens13]

For sure, we could totally do that, just let me know when you would like to do it. Lots has changed since then… but also some things are still good best practices.

[@EMC3D]

Oh wow, awesome! If you use Discord, my tag is EMC3D. I believe I follow you on Twitter too, so if my username doesn't work feel free to DM me yours on x.com/EMC3D and can sort something out chap.

[@bbcvscj]

Hey, lovely work, can you please explain elaborate a bit for us plebs what has changed and why, like broad general guidelines. Coming from character art, I do get what is being done here, but have zero guesses on what could be different now.

[@eudymaverickmentor]

In Maya, you can have the display show hard edges as solid lines and soft edges as dotted lines. Display / Polygon / Soft/Hard Edges

[@5ld734]

When I make an environment for a game I use a small amount of 1024 texture atlases (usually I do one for basics like walls and ground, one for small details like vending machines or awnings, and one for alpha objects like vines or background trees) where I make each piece based on a square in the texture, (I have different sized squares, two 4 meter squares, six 2 meter squares, and eight 1 meter squares, although the sizes on the atlas depend on what is being depicted). Then I make parts based on the different sections of the atlas, build them like legos, then “naturalize” the whole scene once it’s placed together (vertex painting, curving some sections, and offsetting vertices here and there and sometimes adding more topology) and in the end it comes out to a very grid-like and easy to deal with wireframe similar to the images here. As a solo indie developer it’s an awesome workflow for fast iteration and coherent aesthetic

[@5ld734]

This method also makes texel density easy to manage in my experience and the outputs are very consistent

[@Spoopwoopy]

Hey Elliot! Its Katie! Glad to see you're doing well in the industry since you left. Frankly ive sort of recently moved away from what was originally wanted to do(being modern environments) and now heading down the path of Hardsurface modelling (both Props/Mechs and will be attempting some environments too) But i still find your videos interesting! Hope you're well ^^

[@EMC3D]

Hi Katie, time sure flies huh?! Been a year and 1+ month already. Glad to see you're doing well and diving into the realm of hard surface, that stuff is voodoo to me, all the booleans! :P

[@ShinGidora]

Really appreciate the free industry expertise! Always being inspired, I'm tempted to make a mid-poly map like this now!

[@FishMan1nsk]

This horizontal slices on the pillar is a tiling points. Basically it is a baked asset which is tiled at top and bottom where the straight line is. So you can just copy and paste this piece to get a pillar with any height. There's probably a straight line at the back somewhere where the vertical seam is as well. So those pillars are more like a prop rather than level-art piece.

[@Achelon]

I teach 3D for 16-19 year old's at a vocational school here in Finland and the part where you talked about trimsheets that it's hard to make people think with trimsheets and "outside the uv box" is absolutetly correct. I have some pretty simple premade environments and premade trimsheets with only a normal map which we use to create a dungeon type environment. I think it's the easiest way since the student doesn't have to make a 3d model first and only then start practising the trimsheet use.

[@EMC3D]

I love the sound of that method , normal map only is genius for easing folks in, nice one.

[@petyavodolaz]

Good stuff! You should make video on GTA 5. I recently used some tools to get some of their environment stuff imported into Blender to check out, and was pretty impressed how much effort they put into it (tiled texture masks, vertex colors for blending, trim sheets, hundreds of tiled photo textures, etc) Also crazy how they made all that open-world environment in just 3 years of production.

[@EMC3D]

Oh damn, that's awesome. Was it really just 3 years? Those Rockstar chaps usually spend like 5-6 years making the games from what I have heard.

[@dayveeman]

The dotted lines are indeed a way to display soft edges in Maya, although not everyone uses that display option. Things like slicing up the columns into chunks are for making them work on a trim sheet. The extra cuts and relatively equally spaced quads are probably because they most likely used a combination of vertex baked color and shadow maps. Hard detailed shadows on the shadow maps, but Ambient Occlusion and color bounce GI would most likely be vertex color. They might've been very stingy on what areas got full shadow maps and what areas just got vertex color baked shadows. 16:46 shows vertex baked lighting, and you need more vertex density to capture vertex baked lighting. That's also why a lot of these are assembled in Maya not in engine. Now a days most engines bake shadows well enough or even use realtime lighting. So just bring in the reusable pieces into the engine, assemble them there and then bake lighting or go realtime. But this mix of per vertex and shadow maps would've required a large chunk of a level to be assembled in Maya to bake the vertex lighting, and also to allow artist manual tweaking to fix issues because light bakes were not perfect. It was also no uncommon to use vertex colors to blend textures on some materials, or just to add variation and break up noticeable patterns.

[@rogerteles]

All that polygon is also for vertex paint, right? Great content!

[@EMC3D]

Nice one!

[@BradfordSmith3D]

A lot of these principles and practices are still very relevant. Particularly for low spec platforms or broad multi platform. Not only are these workflow practices highly efficient, and allow for iteration, but they are very performant. While Nanite is powerful, I've found that it can actually cause great workflow inefficiency and doesn't provide the level of per vertex blending that many game workflows employ. Virtual Textures, while powerful, can actually result in lower overall texel density / blurry textures up close. Naughty Dog and Santa Monica workflows are still world class for the right product.

[@EMC3D]

Thanks so much for the info Bradford! Really awesome to interact with you, your sculpted tilable videos were on loop many times for me when doing late night assignment work at University, good to see you're doing well and still dropping knowledge in 2024.

[@bradley1048]

I'm a Blender user, I've never even heard of this normal softening technique before. Blender has autosmoothing, which does most of this for you. You can also select individual edges and mark them as 'sharp', which might be the same thing. Because of this, I tend to think the other way round-- the edges are soft and you can make them sharp manually. Or you can use autosmooth to automatically do it by angle. 4.2 just switched the autosmoothing setting to a modifier, which makes it easy to adjust the smoothing angle. But you can't visualise them like this, as far as I know. It does set a blue line around edges you've intentionally marked as sharp, but it doesn't visualise the edges sharpend by autosmoothing. At least, as far as I know. Maybe there's a hidden setting that I'm just not aware of.

[@EMC3D]

Oooh nice one Bradley, cheers!

[@troll_kin9456]

Yeah, this confused me for a moment too, because I just use auto-smooth. I didn't even realize that you could smooth those interior flat faces without creating shading issues, but blender was probably doing it automatically.

[@EMC3D]

@@troll_kin9456 Yep most likely. Very good fundamentals to know though to control it outside of autosmooth, so it is worth having a look into how it works for the future as it'll make you more flexible. Glad the vid helps !

[@lhmsc]

In Blender what I usually use is Weighted Normals modifier. At least for beveled models it works out of the box every single time. Won't look right for last gen models that have hard edges though.

[@SianaGearz]

In PS3 era per-pixel shading was largely too expensive, so most of it was per-vertex with Gouraud interpolation on the polygons, especially if aiming for a 60fps game. For this reason you would need a much higher geometric density than today, and you avoid thin long triangles that take up a ton of screen space like the plague. Because as a highlight travels along the surface, it highlights the polygon edges or jumps from edge to edge. So a lot of tessellated flat surfaces with resulting triangles right angled. If you can lean them closer to equilateral that shades even nicer but that's a lot of extra trouble. 1:30 from the hardware perspective it's not necessary, the texture just repeats (if you set repeat mode) when the texture coordinates are outside the range of 0 to 1. Tool limitation? I was an engine/renderer dev working on Wii and Xbox360 engine that shipped several dozen games.

[@EMC3D]

Ace! Thanks a lot for the juicy info.

[@Genexi2]

The original God of War series used an automated camera which makes it an interesting case study to me. I'd be curious to learn how much input the environment artists had with regards to the camera framing and tracked motion throughout their environments. I figure knowing what the camera will show at any given time would allow an artist to cram each frame with as much detail as possible with the memory budget they were assigned, courtesy of being able to optimize meshes for occluding just outside of the viewing frustum, as well as knowing what distant geometry can remain lower detail window dressing, and what may need LODs/imposters if it's a playable section you can also see from afar.

[@sergiosuarez3120]

As someone who has worked at naughty dog, the amount of loops and geometry could be due to a (somewhat older) way of lighting things where it would be stored in verts, so it would need to be evenly distributed especially in a large area like this. I haven’t seen your other video but this would make sense to me.

[@EMC3D]

Thanks for the comment chap!

[@sergiosuarez3120]

@@EMC3D It could have also been for vertex painting to be fair but I think vertex lighting isn't thought about very much nowadays. By the time i left that technique was only used by stuff that was bumpable and had physics enabled and all that, so it could not be baked. Great video! :)

[@totidoki05]

for the horizontal loop in the pillars, it indeed simplifies texturing a lot and allow for trim use but also has the nice benefit of making all pillar of this design modular. you cna just array the pieces and make almost any height you want, double benefit on that.

[@EMC3D]

Nice one.

[@eborge9711]

the extra topology is either for baking lighting to it or vertex painting. OR it is an old engine that does lighting per vertex, therefore you need extra quads. 2010 seems kinda odd to do this, I'm pretty sure we had per pixel lighting at that point but im not sure

[@M_Lopez_3D_Artist]

I never learned about Trim Sheets looks like i need to understand it I want to be a environment artist, but never found a good tutorial about how to be one in the industry so watching this was very helpful at least to research more. Thanks

[@EMC3D]

Here's a great starting point from Tim Simpson for trimsheets: kzbin.info/www/bejne/f6vMeoxsbJmBd9k

[@M_Lopez_3D_Artist]

@@EMC3D oh wow thank u so much that means so much to me to get this insight so I have a better chance to get a job thank u

[@Shiva4D]

Im 90% sure its not about shading but for Vertex Coloring for materials blend.

[@EMC3D]

Nice one :D.

[@kverkagambo]

Most likely they use modular design for levels, so those horizontal cuts in columns are borders of individual art pieces.

[@bs_art3625]

Wireframes are so nice to look at for some reason lol

[@EMC3D]

Mood. Especially when it's sub-d wires.

[@jkasso7]

The reason I know for additional "redundant" edge loops on the meshes is to add more resolution for vertex painting, is there another reason that I'm not aware of?

[@TheJohn0363]

also slices for texturing splits that for example a small atlas section for a column detail strip, if the column is cut 4 times vertically, it can be mapped 4 times onto that one section creating essentially 4x the texel density resolution.

[@JTW3D]

Will those models in Maya be imported into the game enginge as one big mesh? Or will they export smaller pieces from this scene as seperate objects? Like take a pillar and export it as a seperate object!

[@EMC3D]

We were actually able to talk with the chap who made this level and it will indeed be stitched together! Check it out, it's the 'Nate Stephens' talk video.

[@JTW3D]

@@EMC3D Awesome, thank you! I'll make sure to check that video out!

[@KitchenGuy]

Awesome breakdown! Love manual retopology, it's zen like. I started out with Maya 5.0(?) PLE, which shipped on the disk with Unreal Tournament 2004. I still have a soft spot for it, such well thought through software (by Alias Wavefront). Autodesk on the other hand just wants their shareholders happy and spends too little on innovation. Bifrost is cool, though. I'm a drafting technician now, but still tinker with game dev when I've got time. Great Channel, subscribed!

[@chrismetodieff178]

Great breakdown! Thank you! As for your question about blender softening and hardening normals. In the recent versions is exactly like in Maya. You can soften/harden them by angle, add weighted normals or have simply hard or soft mesh. Previously it was the Maya way plus another option similar to max with something like grouping. Cheers!

[@venusdevendra1788]

3:48 Vertex Paint, really enjoying these videos please keep them coming!

[@EMC3D]

Nice one!

[@Savigo.]

The extra loops are probably for vertex painting - you need pretty even polygons for that otherwise the blending between textures is not going to look good.

[@amanthedev]

While being in edit mode, You can find the split normal toggle button in the mesh edit mode overlays

[@stysner4580]

You say "the artist has gone in and make the edges soft", but it's the other way around. You make the whole mesh soft-edged, then you go in and make edges hard where they need to be.

[@EMC3D]

Depends on the artist, but yes, you can do it that way too. Whatever works best.

[@stysner4580]

​@@EMC3D It's is exponentially more work to select all the soft edges than the hard edges for modeling styles like this, I've never seen anyone do it the other way around. Either people do it procedurally or make all edges soft, then harden the UV borders, then harden everything that needs it for visual reasons.

[@proximity88]

The reason they have the dotted lines is because Video game engines do not like long thin Triangles, by slicing the mesh into smaller chunk you stop a triangle stretching half the screen which has rendering costs. also for Vertext painting you need vertexs so you get variation. Another thing to note between the Dotted lines and Hard lines is that the Dotted lines show where Vertexs are joined and the Hard lines show where they are split. They split models to create smoothing groups which the engine considers as one mesh shading it accordingly

[@StigDesign]

i think its 3 reasons for the extra loops and rings, 1: it soften the shadding/Smothing group, 2: even uv mapped repeated texture/or sudden change in texture, 3: Vertex coloring for smothing from 1 to another texture perhaps even in diffuse and normalmap+other maps :) this tecnique was importanton ps1- ps3 because they had limit on texture size :) its often easy to make much of a level in 3dsmax/maya then wield different level parts together that can be attached or wielded and everything has even same 0,0,0 world point and then export it() of course ther is exeptions with models that needs to have it own positioned pivot like hingses bridge door,etc, it can even be seen if you extract all gta vice city or gta3 level and import it in all parts fist together and uses the same worls 0,0,0 as refrence :) while other games can use the grouping method that also varries , Great and Awesome video where you mentioned lots of what i said, and also guessed, like the pillars with uv :D

[@StigDesign]

another note, what i nteresed me specific is in the racing game making, if you watch the making of Need for speed hp2 on ps2/ or on ps2 emulator, you can see the tracks actually are made in Maya perhaps with spline too(tried it in 3ds max and it is time consuming to get good road with good repetition and uv mapping with splines but its funn), but when you see making car its is 3dsmax and it even seems they dont hard surface modeling but using Nurbs instead hehe :D

[@arefghadimkhani6621]

I really find these videos helpful please do portfolio reviews more often ❤

[@petesidtech4adventures415]

16:38 I wonder what's with the shading of the otherwise flat inset of the column on the right, can you explain that?

[@EMC3D]

May be the ambient occlusion baked to the vertex points, we'll be having a cool guest on soon who worked on some of these levels so can dive deeper in the future.

[@petesidtech4adventures415]

@@EMC3D Ah, makes sense. Thanks for the reply! :)

[@Arthoire]

The extra loops are added to use vertex colors.

[@FriendsorFiends]

I would guess that the loop on the column is there to add a variation textures that match at the edges with each other, as the loops break the columns into equal sizes. You could then mix and match textures on the columns to create variety down the corridor and avoid it looking like it's obviously the same column. Also maybe for adding a vertex alpha painting layer. These old vertex lighting painting engines allowed for many layers of vertex alpha painting layers for dirt, variation and shadows.

[@EMC3D]

That's a good point, the quest for this pillar trickery continues :D.

[@depedi5373]

Great video!

[@danialsoozani]

I think evenly distribution of loops are for displacement maps

[@RadostinIliev-nk4cv]

As for the amount of squares on the meshes, I guess even though they don't add anything to the shape itself, but they are crucial for controlling the texel density, when it comes to texturing.

[@LamontGilkey]

I would also add vertex coloring, and material blending.

[@sleepyreapy1222]

Nothing to do with texel density as it's controlled via the scale of the UV, it's subdivided for vertex paint

[@GrandHighGamer]

@@sleepyreapy1222 If they use a single texture it could allow for repeating. But yeah, vertex painting makes more sense.

[@GeometricPidgeon]

texel density isnt determined by the amount of faces, its how large those faces are scaled in 0-1 space. It is most definitely for vertex painting.

[@GeometricPidgeon]

@@GrandHighGamer you can repeat a texture over a mesh without needing additional loop cuts.

[@molewizard]

I suspect the divisions between the meshes are so the game could ship with smaller assets - you can instantiate the smaller pillar multiple times, instead of needing larger pillars which take up more GPU memory, and are less re-usable across multiple pillar heights.

[@EMC3D]

Cheers!

[@Tritoon710]

Thanks, great video. What is vertex painting?

[@MrFabutan]

assigning color values to vertices, usually to mask or blend features in a material, like texture masking or in olden days painting lighting and shading value directly on your geometry

[@juliuseinstein6093]

I thought vertex painting was relatively new, great to see the OG creator's or artist's works.

[@juliuseinstein6093]

I have a question though, do they use LODs or is it only for open world games?

[@EMC3D]

Haha, been around for a very long time :), which is awesome. LODs? Most likely yeah, probably more aggressively back then and done by hand if I had to guess.

[@knairda8282]

vertex painting has been used since at least the N64/PS1 to often create fake lighting and ambient occlusion!

[@natestephens13]

God of War 3 rarely used LODs. Since the camera was fixed we could tune the geometry for the situation the camera was in.

[@rlb1968able]

There's a few optimization things that they should have done. Probably put less loops on the rounded objects. But I'm not an avid 3D modeler

[@MrFabutan]

This game ran perfectly on PS3, why would they further optimize beyond what you see here?

[@rlb1968able]

@@MrFabutan no I'm just bringing up something they could have done I mean they I didn't need to do it

[@EMC3D]

Cheers for the comment. Yeah the poly usage on arches/corners is something I usually advocate for, we've got a bit of leniency on optimization with geometry, say versus textures and lighting, which is waaaay more strict on budgets.

[@natestephens13]

The GOW 3 engine was an in house creation, and it could chew through polygons easily. We never really even counted polys or had a budget since it wasn’t the bottleneck. Texture size was the thing that had to be optimized in the end.

[@SAGERUNE]

When you say harden would that also mean sharpen?

[@EMC3D]

I believe that when you mark for sharp it acts as a 'harden' and splits the normals.

[@rossknowles5608]

blender gives you shades of blue to represent the different edge values

[@indianastilts]

Question for all you awesome experienced people that I hope you can answer for me.... In either blender, maya etc creators make their levels in that 3d modelling programme then module it and re create it in the level editor (unreal, godot unity etc etc). Simple question...why? Why do they do that when if created fully in blender, maya etc, as already seperate models, why not simply export it into the selected level editor?

[@MrFabutan]

I have done this because it can be easier to layout an environment quickly using the tools in modelling package, but then more optimal to import into game engine in pieces.

[@scgstudio]

Game editor is for game editing, modeling soft is to make models. Use the most performing tool for the job. The more complex level is, the more modular it should be. Another reason, you do not want to go to 3d package to move some wall or pillar. Going back to 3d editor limits the iteration speed and shrinks the freedom of placement. For tiny game it might be a decent fit, but this is not a good habit in gegeral. Make modules, lay them in game engine. Later updating module in 3d package will propagate changes where module is placed. There are other reasons to keep level in peaces like occlusion culling, navmesh generation, collision setup and so on. Good luck !

[@indianastilts]

@scgstudio amazing thanks alot

[@ПавелЩетько-ы4й]

Большое спасибо, крайне полезное видео

[@Strelokos666]

Very excessive in its density imo. I assume you only need evenly distributed quad topology for trims and vertex color. Any details like displaced bricks can be done separately without sewing into topology. Texel density is... yeah, just like in case with trim sheets. Greater architecture needs to be divided into quads to keep up to the size, that is if you are not planning to use single tiled texture of course. Barely understood the soft shading and dotted stuff questioning why even doing that being Blender user. I can guess you need to resort to managing sharp / soft shading since you can't bake bevels and surface info via normals due to the fact environment art is not about making unique assets.. Idk, maybe try using 8 x 16 gradient bevel trims to fake that on the edges.

[@matteofalduto766]

Forget the topology, the use of the architectural orders is all wrong 😨

[@EMC3D]

Rules of orders don't apply to the man himself Kratos! All the destruction.

[@matteofalduto766]

@@EMC3D fair enough

[@flow1194]

why should i triangulate whenever i can? i thought game engines do it for me?

[@EMC3D]

Ah yeah not for the whole model, but if you're fanning out a lot of topology from an arch, you'd usually target weld into triangles where possible. Check the Naughty Dog video out if you want to see an example of that on their houses :D.

[@flow1194]

@@EMC3D thanks for the recommendation

[@scgstudio]

Each quad can be triangulated in two different ways. Because of shading, it is good to make triangulation where it is crucial rather than everywhere. Game engine will triangulate topology for you, but will not ask how to trangulate.

[@omgnowairly]

This isn't the in game asset. When you see quads, you know its part of the art pipeline.

[@EMC3D]

Yep, non-engine wireframes are nice to look at, otherwise it's all triangulated, so a bit tougher to look at, but in the end everything gets triangulated in-game, so it isss what it iss!