Advice for 3d environmental artist?
I've been going through portfolios and decided that I want to really focus on environmental art. I am getting varying details on what is required for that job, though. What do you feel are the requirements for this job? Which software should be essential knowledge?
I appreciate any and all information!
I appreciate any and all information!
Replies
Solid knowledge of how to make props and architecture pieces, how to make convincing textures, how to make assets that won't melt an Xbox.
And if you want to be fancy, learn things that will make the tech artists like you :P
Right on! Thank you!
What do you mean by that second part? What type of things should I look into that would make hte tech artists like, NO!, LOVE ME!?
But above all, make your portfolio the most awesome possible. There's a lot of competition.
Lots of advice here
http://wiki.polycount.com/wiki/Portfolio
When I got my previous game art job, I came into it knowing really only how to do props and modular pieces. So knowing how you'll approach 'level art' as opposed to small/medium props is good.
But yeah, as above, the best thing is to present assets/environments/BOTH that will make the art leads (or whomever) think "I would love our game to look like this"
That being said, you should "master" the art side of things first and foremost. Obviously. These are all strictly secondary skills that will help you grow as an artist and give you brownie points in the review process when applying. I know, for example, at Blizzard they like to have env and prop artists handle asset creation from basically start to finish by themselves. Concept all the way to animated, FX, polished prop ready to be placed in game. But then again, who doesnt want someone who is a master of all? lol. Reading back, a lot of that I guess is common sense.
Master one skill set, build a portfolio, profit. Once you are in, start building secondary skills.
You probably should be familiar with creating tileable textures. Learn how to conserve UV space in your textures and mirror things in your UVs.
3ds max (or) maya
Photoshop
Unity (or) UE4
Zbrush
Bonus:
Substance Designer (Tiling Environment textures)
Substance Painter
World Machine
Here is the gist, you will most likely run into of these two studios or a bit of a mix that is why this is hard at times:
Studio A:
Build modular architecture pieces along with tiling maps needed to fit with in memory restrains, while also creating some hero unique baked props when needed or additional prop support isnt available.
Studio B:
All of the above plus integration and working with a designer to place your assets into engine. This will most likely include lighting the scene and optimizing scene performance.
The gist is if you're good at making props make sure your really good at architecture modeling and using and creating tiling textures as well. Use a game engine to create scene and light them. Once you have epic art (You will be hired as an Artist) for levels dive into learning all the technical's that you can about how shader optimization works, how the level assets get batched, advanced lighting and effects. Basically anything that has to do about optimization or more polish.
Thats the best road I can think about in terms of being the most "hire able" for an environment artist.
I appreciate it, man! Thanks!
Thanks everyone, by the way. Lots of great information. Much appreciated!
Get a solid art foundation, no matter what you do that is the most important.
Learn the tools.
Dont follow the popular genres that others cover in their portfolios. Look outside of popular media like games and films for inspiration.
Stay true to yourself and use that to inspire your art.
It will take years, if youre lucky/talented not too many, but dont get discouraged. Always be on the look out for work in games or portfolios you like and compare your work. Dont be afraid to contact or email them either asking for advice.
Engines you can learn down the road and alot of studios tend to run their own engine or a tweak version of a more popular engine(ue4, unity, cry). So getting too head deep into one or another might not necessarily be the most efficient time spend.
Network, share your work online, get feedback. Constantly compare your work to industry proffesional and ask yourself whether or not your skill level is atleast equal or better.
Self motivation is key. Good luck!
I've been going through portfolios of people work and just spend hours on ArtStation. I'm just gathering information on what is needed.
One thing that I am seeing a lot that I definitely need to look into is seamless tiling textures. Is there a "The one" tutorial that I should watch on this?
Also, a little off-topic, but is Substance Painter any good and worth learning? I ask because I bought it when it was first released and never bothered to play with it. I've seen some talk about it, but would texturing be better served in a different program or is Painter good enough to devote some time?
Thanks guys!
Substance Designer is probably a bit more suited for traditional texture work though. It can be a little tedious and overwhelming at first trying to get used to the node method. But thinking about it like layers in Photoshop seems to have helped me.
Substance Designer has almost entirely replaced Photoshop in my texture creation workflow. Being node based makes it almost an entirely non-destructive workflow too.
I would definitely look into Substance Designer if you can.
There are times when you will want to work with actual geometry meshes for your textures. Zbrush is pretty good for that kind of work. The new Nano-mesh feature is really great and worth looking into.
It's relatively simple to create tiling textures in both Substance Designer and in Zbrush.
Another resource you should definitely check out is the free Vertex e-books: http://artbypapercut.com/ There's 2 of them available so far, a third one is in the process of being made. Both of them are full of useful information and techniques from some of the best in the industry.
Hope this helps!
Nice! Great information. Thanks a lot, man.
Yeah, I'll check out those e-books and give Designer a look. I have a little bit of experience with node based editing when I was playing around with compositing, so hopefully it isn't too much of a shellshock.
Painter allows you to for example paint some scratches or whatever on the model that affect the output of the color, normal, roughness, etc at the same time. You can also isolate the different texture maps and work solely on that within painter if you need to, and some of it's features are sort of non-destructive in that some of the brushes automatically transfer over the work you've done if you decide to up the model from 1024 to 2048, or relatively easy to adjust the color and roughness of a particular layer (same thing as layer masks with color multiplys in photoshop basically, but hey it's there and integrated with a 3D viewport)
I wouldn't say any of these programs are the end of one another, as I still usually leave photoshop for creating certain things and importing them into Painter or for final touch work, but it's good to be able to work with a range of options for what works best with what you're doing. Heck, I love using Corel painter for doing hand painted work (their brushes feel the closest to physical media for me) but always end up bringing it back and forth from photoshop.
Anything normal-map related I'll either model out and bake, or use nDo if it's a small detail that I'm probably going to constantly be adjusting. (Usually large details I'll model out and bake, and save things like buttons and smaller design elements for nDo since I like the option of being able to adjust positioning, depth, etc. at any time I want versus having to constantly remodel and bake. Just seems quicker and more responsive to me as I figure things out on a model I'm not basing off of a 1:1 concept) There's still stuff that I might just pass through something like Crazybump or bit2map with a little bit of photoshop tweaking if it makes sense for what I'm doing, saves time, and looks just as fine.
I basically view photoshop as the glue between all of the different programs for texturing.
But yeah, so many different tricks. Don't go nuts and remember to have fun
I've been reading and I've seen quite a few different software packages talked about, but I don't know if one is better than the other in certain areas. For example, I've read that Marvelous Designer is great for cloth based materials. I wouldn't even have thought about using that, but it makes sense after reading about it. Does Marvelous Designer do anything else well, especially better than other programs?
What about the rest of the programs like:
Quixel, nDo and dDo? I know nDo is great for normals, but is it better than xnormal?
Is dDo better than Painter, Designer or Photoshop for texturing?
Do they all do pretty much the same thing and it just comes down to preference or are these programs really better than others at certain areas, ie, should I learn these specific things when considering environmental and prop art?
Thanks guys!
IMO Painter+Designer > dDo > stock Photoshop.
There is some pretty big differences between the software, there is some overlap, but there's a lot of differences.
oh, ok. Right on! Thanks!
Someone earlier said that it was a smart idea to make textures that aren't destructive. When it comes to creating a texture, is it considered destructive if I model something out in zbrush first, then bring it into a texture program? Or is it only non-destructive to do everything in a program like Substance Designer?
Also, Designer just released a cool walkthrough on Youtube:
[ame]
He talks about how he made that texture. Maybe I missed it (only watched the first two parts) but does he model that piece first or is that created completely in Designer?
Thanks!
It's a term applied in anything art-related, such as how vector-based artwork is basically non-destructive in that you can easily scale it up so it's the size of a billboard or shrink it down so it fits on a business card. (That's why almost all corporate logos are done as illustrator files/vector-based) If something is done as just pixels/any regular image you see as a jpeg for example, you can only maintain quality at that size or by making it smaller. If you enlarge it it'll become a blurry mess, which is basically the type of destruction you'd might worry about when making a texture.
IE say I made everything at 512x512 pixels as a texture map and flattened all of my layers/artwork so it's just one flattened image (which is what a jpg or bmp does). Now my art director tells me that we can up the resolution of the textures for everything to 2048 (this is an exaggeration of an example) as we have some more memory left in the specs. Now I'm screwed/either have to redo it at the new resolution, or do a bunch of photoshop hacks to try and get it to be less blurry. If it was done non-destructively such as in substance painter or designer, it becomes almost a matter of changing a box from 512 to 2048 and hitting export.
Obviously it's not too big of a worry if you know what your target resolution is ahead of the time. Some people work at exactly the resolution it will be (technical artist says two 2048x2048 materials, one for the body, one for the head on every lead character. One 2048 for background characters, etc.)
I personally always like to work one step up above of what I think the final texture map will be. IE I'll work at 2048x2048 and decide if I should scale down to 1024 or 512 after the fact. It's kind of similar to the way painters like Norman Rockwell worked. If you looked at the paintings he did for book illustrations you'd see they're freaking huge comparatively and you can actually see his brush strokes, but scaled down and it becomes very tight and detailed. IE for some people it's easier/more comfortable to work big-to-small.
Non-destructive workflows when it comes to modeling basically can come down to things like using modifiers in Max or whatever they're called in Maya. A popular method being to create models that are actually alot more simpler/boxy looking in form, and then apply smoothing on top to create the curves. You can always go back to the unsmoothed form and adjust the geometry, which is easier as you have less vertices and polygons to deal with moving and selecting. It becomes destructive when you actually export the model into a game engine, or flatten the modifiers in Max/delete history in Maya. Flattening/deleting history tends to speed up the program you're working in and clean up the history, but will mean it'll be alot more difficult to alter if you wanted to go back and adjust something. That's why making multiple save files as you go is important, and knowing how to work smartly can decrease the time spent having to fix and adjust things.
In summary, non-destructive basically is just a term for preserving the ability to go back and change/adjust things in the least time-consuming way possible.
Nice! It's funny because I knew a lot of that, but just didn't realize it wasn't considered non-destructive. Lots of great stuff. I appreciate the information, my man!
Ok, I know I said I wouldn't, but I have another question.
If I am designing an interior scene, what is the best way to do that with intentions of exporting it into UE? Do I just make the props and so on and then import them into UE? Use UE to then build the walls?
or do I block it all out in Maya/Max, zbrush anything I want (wall details, props, ect), bring that back to Maya/Max and bring it back to low-poly THEN export it to UE?
I could see the benefits of just making modular pieces and importing them to UE, then using the engine to block out the building, but I could also see the practicality of doing it all in Maya, then just importing it to UE as an almost finished product.
Does it matter which way or is there a better way to do it?
It may seem silly, but it's building with the end product in mind. Like a painting, artists will go into finer details and focus on one area of the painting. The artist should take a step back to see the bigger picture often, because when it's finished, that's what matters.
Some people use BSP brushes in Unreal for getting a quick blockout working in engine. I personally just prefer to do a block out in the modeling program itself. But the idea is the same. Build for and test in the engine that it's actually going to be seen in.
Art directors do like modular assets for technical and re-usability reasons. This is a pretty good article that goes into a little bit of how Skyrim used modular assets to help build the world of Skyrim: http://blog.joelburgess.com/2013/04/skyrims-modular-level-design-gdc-2013.html
If you're environments are going to have any gameplay whatsoever this is even more important! Pacing is incredibly important in games and if you're working on a platformer you'll want to make sure a ledge can actually be jumped on to and not be a little bit too high, too far away, etc. Most of the time this can be tested with a blockout and doesn't have to be a finished asset.
This also applies if you're wanting to work with specific camera shots/angles. Get your FoV and camera settings how you want it using a blockout as reference before you try and finish the assets.