At CGMA, Ali Ghadimi studied Intro to Environment Art and did a breakdown of his small vignette Old Southwestern Gate made within the course.
My name’s Ali Ghadimi and I’m an environment and prop artist at Northrop Grumman. I went to the University of New Mexico with an interest in film and VFX and eventually found my way into a game pipeline when we started doing a lot of VR stuff at work. Non-entertainment industries are very interested in VR for all kinds of things from equipment maintenance training to PTSD resilience training.
I have no game art background so I decided to take the CGMA course Intro to Environment Art to get the best training possible. I think a lot of people are uncertain whether CGMA is worth the cost since you can find so many free resources on YouTube, Gumroad, and ArtStation. But one of CGMA’s values for me is confidence that the process you are learning is professionally tested at studios like Naughty Dog where my instructor, Andres Rodriguez, works.
For people like me who are self-learning something as complicated as environment art, this is invaluable guidance. The problem these days is not too little information, but too much. There are too many tutorials and polycount threads to sift and many of them can teach you bad practices. You can waste a lot of time doing things and actually learning an inefficient or misguided workflow.
It’s easy to get overwhelmed and have trouble even getting motivated to start a project when you don’t know how to approach problems. CGMA lets you take a step back and just focus on one objective for the class. Then at the end, you have a complete portfolio piece that has been critiqued by an industry pro.
Old Southwestern Gate: Blockout
Let’s move on to the environment. It is very important to start with a good blockout when approaching even a small vignette like Old Southwestern Gate. From the very start, it’s necessary to put enough time into blockout because it makes a lot more work for you down the road. You’ll end up putting too much time into pieces that are, in fact, not that visible and forgetting things that you need. It is important to plan ahead, especially when you work on more complex scenes with aspects like modularity and areas the characters will be involved in.
I didn’t do anything crazy in the modeling stage. For organic stuff, it would generally go the following way:
- Mid-poly export
- HP Detail in ZBrush
- Decimate to LP
If you need a cleaner/lower poly mesh you may have to retopo back in your 3D package after. For hard-surface stuff that wasn’t going to get sculpted on, I often have 3 meshes. I first model the mesh with as few bevels as possible, then it gets split into a low-poly and a high-poly. If you do a bevel too early on your LP, it can be hard to make an HP from that mesh later, so try to keep the mesh intact without bevels as long as possible until I can split it into a high and low poly and bevel them appropriately. Then I can add other details to the HP and smooth it. For some complicated hard-surface stuff, it’s just easier to make a high-poly then retopo it for your low poly, but nothing in this scene was very complex.
Since I was using Marmoset for the final presentation and not a real game engine, I focused on visual fidelity rather than efficiency. However, I still tried to keep everything as low-poly as possible. The most expensive asset ended up being the Prickly Pear clusters, the largest of which was around 10k tris. This is quite high but could be LOD’ed down very low or to cards if I was trying to optimize the scene. I skipped that.
For other foliage, I modeled a high-poly in Maya and baked it down to a card in Designer. You can get a lot of variation exporting your maps out of Designer, but for some things like the Tumble Mustard, I actually brought my Designer maps into Painter to do some extra detail work. Here you can get as detailed as you want, with the drawback of it becoming harder to edit them later if you need to rebake. I think Painter is more powerful than people realize for certain foliage work. I actually did the Yucca leaves entirely in Painter. Once you start hand painting you can make pretty much anything just using default materials and fill layers. Everything is situational but it can actually be much quicker than trying to make a custom material in Designer.
Texturing & Substance Designer
Let’s expand on texturing. This was my first project with Substance Designer in the workflow and it is definitely a game changer. At this point, I don’t think you could do environment art without it or Quixel.
If you have used other node-based programs like Nuke or blueprints in Unreal, SD isn’t too much of challenge to learn. I bought some Josh Lynch tutorials on Gumroad to augment my course lectures and those really helped me get a deeper understanding of the software quickly.
Once you get over the initial steep learning curve, you’ll find yourself using a lot of the same nodes over and over again, you don’t have to have a super clever solution to every problem.
You see a lot of crazy complex materials on ArtStation like intricate metal work or sculptures. I think this leads to a kind of false sense of what you need to know to make good materials for environment art. Making complicated materials is good for obtaining mastery of the software and showcasing on ArtStation, but they won’t necessarily be useful in environment art because you don’t want something that is too specific and will obviously tile or relies too much on a height map that needs tesselated Geo. Utility is better, and speed is crucial. Most of Andres’ materials from the lectures were very simple and practical. You can then vertex blend those base materials to get the variation you need, and suddenly a few simple materials look great in the scene.
Joakim Stigsson’s cactus breakdown on ArtStation got me halfway there on this material.
Since we used Marmoset in the class, I assembled everything in Maya because Marmoset’s transform tools are pretty awful with no ability to adjust pivots. All the foliage were made into different sized clumps for layout. In a real game engine you’d combine clumps of clumps to reduce draw calls, but since this was a small vignette I actually combined everything for export together by shader so it was fast to setup the materials in marmoset. Though this meant if I wanted to move a grass clump I had to do it in Maya then re-export the fbx. For example, all the grass was one mesh, all the mustard plants were one mesh. This made Marmoset run faster and let me set up the materials very easily and drag them on to one mesh.
You can’t get away with this in a larger scene though, because you want to be able to cull non-visible assets and that won’t work if they are all combined.
Lighting is one of the most important aspects of your environment. You can make bad assets look good and great assets look awful with lighting. It’s one place you should never cut corners. Anyone interested in environment art should study film lighting. There’s like 120 years experience in dynamic, dramatic lighting for environments there.
Vilmos Zsigmond said, “I like to say that lighting is about taking the light away. I often like to use the shadows more than the light.” That’s where I like to start, by turning off all the lights in the scene then adding them back in one by one. It’s easy to get too many lights battling each other, and a lot of beginning environment art I see is very flat and over lit because of this.
I started by turning off the HDRI since it looks flat and bad by default. I added a moonlit key light, then turned the HDRI back up a bit for ambiance. Clarity of space is important in games so the player knows where to look and what paths to go down. I used the lanterns to draw the focus up the steps into the arch and then used the sky to give the walls contrast. I added a couple spot lights to accentuate some dark areas and give the cacti a bit of rim light. There’s an emissive channel in the lanterns and some orange point lights to make them pop. Then I just added in the skybox in the back and put a LUT on the camera to make it pop some more.
Well, this was my first real environment piece so everything was kind of a lesson learned. There were all the software issues like learning Substance, Marmoset, and ZBrush and trying to figure out a good workflow with them. Most importantly, I think even on a small vignette like this it’s essential to make a good blockout and keep very organized. There is the temptation to just kind of wing it, but you’ll end up with problems for yourself down the road. I probably spent several weeks longer on this project than necessary because of bad planning. Every time you are doing a quick fix or just exporting something to test it and you don’t put it in the proper folder structure or name things right, of course, that’s the thing you end up using and things start getting lost.
Another big lesson I learned from the class was how to call something good enough and move on. I tend to get hyper-focused on each asset and take a long time trying to make it perfect. That may be OK for hero props but environment art you must move quickly. I think you can make a great environment with decent assets. It’s more about laying them out and lighting them well. You can spend a week making the perfect grass clump or spend a week making five different plants that will fill out the scene much better and ultimately look much nicer.
If you are interested in environment art but haven’t done it before, I highly recommend doing a small vignette like this before trying anything more ambitious.
Advice for Learners
If this article was interesting to you, I’d suggest that you out the Intro to Environment Art with Andres Rodriguez video on CG Society (below), and, of course, the CGMA courses.
I hope there was something interesting in this article, and special thanks to Andres Rodriguez for the class, as well as everyone at CGMA and 80lv!