Environment Art Portfolio Critique

b_beauchamp

Hi everyone, I’m looking for some feedback on my portfolio. A few weeks ago I sent out some job applications and received radio silence in response, and one offer of an art test which I failed. At this point I know I need to improve further, but I don’t know “what” I should focus my efforts on. Currently I’m slogging through a (mind numbing) Gnomon Workshop environment modeling tutorial and working on improving my vertex painting skills.

Regarding the art test, I’d like to share my work and get some critique about where exactly I screwed up, but I don’t know what the protocol on sharing art tests online is. I wasn’t required to sign an NDA or anything, but I have not asked permission to share it. Since they did not respond to my request for feedback I don’t want to bother them anymore. Thanks!

You can view my current portfolio here: https://www.artstation.com/breannabeauchamp 

Edit: I’ve decided to share my art test. You can access it here: https://drive.google.com/drive/folders/1BsjcHu2e55AatMe_1UQHaoY60wk64ay6?usp=sharing 

The folder is exactly what I submitted to the studio, minus the test document.

After thinking about it for some time I’ve made a list of what I think were my main errors. In addition to these things, the whole art test was plagued with every technical issue possible, from regular engine crashes to power outages that corrupted my save file. I would have probably been able to submit my project three days sooner if not for those problems. I don’t think this would have changed anything in the long run though.

-Prioritizing visual realism too much over optimization/clean modeling techniques

-Not snapping models to the grid/having random dimensions for meshes

-Misinterpretation of the term  “modularity” 

-Bad mesh optimization for a game environment

-Bad/lazy lightmaps

-Too many meshes with unique UV sets compared to modular pieces

-Overreliance on decals compared to other texturing methods

Ashervisalis

I think the lighting in your latest scene could be improved a bit. A lot of the images are overly dark.

Your portfolio is pretty good! Don't get discouraged, there will be a looooot of radio silence. The fact that you got an art test shows you're somebody worth looking at. Don't view an art test as 'failed', view it as they got ahold of someone who did their test a little better than you. You will do art tests and not get the job, it's a part of this industry. If you did not sign an NDA for the art test, I don't see any reason why you couldn't show it here.

Keep moving forward and making more art! I would personally like to see some complex sub-d modeling in your portfolio. Your environments are built with trim sheets and small props, which you've proven you can do. If an animation studio approached you, could you model a sub-d car? Would be cool to see a more complex asset added to your portfolio.

Is your location OK? Are there a lot of studios near you?

b_beauchamp

@Ashervisalis Thank you for the feedback and encouragement. I found lighting the basement scene to be pretty tricky. Looking at it now, I see that I put too much weight into matching the concept art versus visibility in the engine. I’ll link the Google Drive folder where my art test is stored in my original post along with my personal critique of what I think went wrong. The folder is exactly what I submitted to the studio except for the test document which explains the guidelines I was given.

I’ll admit that I am a bit intimidated by complex sub-d modeling. There is one fairly complex robot model I’ve been working on and off for a while now as a practice project. I’ll link an upload of that project below. The current mesh has some issues, but if you or others say it looks alright I can press forward and finish it. The prop for the Polycount environment art challenge this round is a pretty complex looking steam carbine so I’ll have a go at that and hopefully end up with something decent. 

There are a few studios in my area but the selection is not too good. I’ve only applied to local opportunities and studios that offer remote work so far.

Robot practice project: https://drive.google.com/drive/folders/1Q13r3re7LGzj9ayk64mvTkfI8Ce8dqVw?usp=sharing

Ashervisalis

I do see quite a few issues with your robot mesh, but I also see it's meant to be a low poly mesh? Check out this tutorial;

https://www.pluralsight.com/courses/tips-modeling-complex-shapes-2082

It was the greatest thing to happen to my sub-d skills. You need to create an account and sign up with a credit card, but you can just do a free trial and bang through Part 1, Part 2, etc. It basically shows you the general idea behind sub-d. I'd recommend spending a few days just focusing on that. Basically, you want to make sure that if you press 3 in Maya (or smooth in whatever modeling program you're using), it looks nice and neat. Once you have this technique down pat, you can apply outside of games, to animation and such. If you need a push in the right direction, feel free to drop me a PM anytime here. Sub-D is literally the only thing I do at work hahaha (I miss texturing).

Your art test doesn't look bad. It would be good to join a chat community to get feedback in real time when doing something like an art test. Without seeing the actual test info, I'd say the weakest point in your result is the lighting (no shadows and such).

JamesBrisnehan

I think your art test looks pretty good too. I agree with Ashervisalis that someone else's might have just been a little better, or even same quality + longer resume. The only thing that looks off to me about the scene is that the windows look a bit too opaque. Also, I think the wall textures could use a bit more color break-up to match the ref.
By the way, did you end up using any photogrammetry assets or texture libraries? The wording of the brief almost sounded like they wanted to see that kind of thing in your finished scene. Maybe that studio uses asset libraries in their day-to-day workflow and they wanted to see how you might incorporate pre-made stuff into your work. . . Or maybe I'm reading too much into it.

If you have the time, you could try participating in the newest Artstation challenge starting in a couple of days. It could get you and your portfolio some extra exposure, as well as bragging rights for your resume if you get 'honorable mention' or higher.

b_beauchamp

@Ashervisalis Thank you for sharing that tutorial and offering to help if I get stuck. I look forward to working through it, my sub-d modeling skills could definitely be improved. Is there a space here on Polycount where I can join a chat community for art tests or would I find one elsewhere? 

The “Art_Test_Doc” file in the folder I shared was all the information I was given to work off of for the art test. The scene lighting is pretty washed out because I didn’t see any shadows or directional lighting in the reference I was given. I tried to cast really subtle shadows a couple of different ways, but I couldn’t figure out how to make it look realistic with an overcast sky.

@JamesBrisnehan Thank you for the feedback! I agree the windows/doors definitely look off. I tried making them slightly transparent at first, but the result looked pretty bad. The main thing that bothers me about them is that they are just stuck onto the wall without any recess. I really wish I had learned vertex painting before getting the test. I would have been able to break up the textures on the walls and ground much better instead of just using decals.

I actually used a ton of megascanned assets and textures from Quixel for the scene, and some stuff from Substance Source. Everything that you see in the renders that is not uploaded in my art test folder is a megascan asset or texture. Probably the most prominent examples are the piles of scrap around the edge of the building and all of the decals I used.

I’ll take a look at the Artstation challenge! Are you referring to the “Grand Space Opera: Light Age Game Environment/Level Art” one?

Ashervisalis

The Polycount Slack chat has a smaller community, which I like. We've all gotten to know each other pretty good, and there are some really good artists on there. If you pop in there, just introduce yourself and I'm sure you'll get a friendly greeting.

http://goo.gl/forms/Iw3kDU42la

The Polycount Slack is quite a bit bigger, but I've not really ventured in there;

https://discord.com/invite/0Vu6wFfSmdogS2Lm

JamesBrisnehan

Yep, 'GSO: Light Age' was the one I was talking about. You could choose either Environment Art, or props if you want to get some sub-d modeling practice in (or characters if you want to try something different). If you decide to join, be sure to post regular updates throughout the challenge. That will increase the chances of other contestants dropping by your thread to offer advise and encouragement. You can post W.I.P.s here on Polycount too if you want.

b_beauchamp

@Ashervisalis Thanks for sharing about the Polycount Slack chat. I’ll introduce myself over there sometime soon. I’ve also started going through the Pluralsight tutorial you recommended. It’s been really helpful so far!

@JamesBrisnehan Thanks for explaining the Artstation challenge. I think I have a better chance of succeeding with the Environment Art one but I do need to get better at sub-d modeling. Maybe the challenge brief will help me decide. I will definitely post updates on Artstation and Polycount for whichever one I end up choosing.

birb

@b_beauchamp Are you okay with me accompanying a critique of that art test with quick edition of it? I believe a number of minor adjustments would greatly improve the realism, but trying to explain in writing would take three times the what it takes to show them. Being an art test though, I won't touch it unless you're okay with it.

b_beauchamp

@birb You are more than welcome to do a visual critique of my art test. I look forward to seeing it! 

birb

Nice! I'll preface this with the fact that you did a nice job already! If the alterations below are minor it's because the work you've done here is pretty solid. The textures are yummy, the light and colors are pleasant and in line with the reference, the rubble is beautifully scattered on the ground. Clever choice of ground texture; the striations are perfectly positioned to look great in the path of incoming players due perspective.

The main issues I see are some scale slips—small but enough to break immersion—and the distracting windows. I've included all points made in the gif as text in the spoiler tag too. Sorry about the speed, this would be better as a slide or miro presentation!


 (Note: I adjusted dimensions only enough they read a slightly more correct so you get what I mean. They're still not true to the ref.)

Transcript

1

Rework scale to match real life measurements. The passageways were too wide, making them look uncomfortably low.
Excluding 1st person games, you’ll likely need higher than real life passages to allow space for the camera, so entrances lower than in the reference are concerning.

Also be careful about (the piping) transitions, there were thickness issues in some areas. + Thicker pipes match the reference better and will also look better from a distance in engine.

TIP: You can extract dimensions of elements in the reference with architecture apps (eg Image Meter) or Photoshop. Take note of how these elements relate to each other.

2

The door should have a slimmer and inset frame. The window frame should be thicker.

TIP: Most exterior frames on facades exposed to the elements will be inset to avoid water damage and dirt; otherwise they’d have rotted away a long time ago.

3

The second floor windows overlook a corridor (or a room that reads as one from the ground).

Take care to maintain the original alignment when working from ref. Yes, that section of the wall is wider, but that’s easy to gloss over. The misalignment is harder to ignore.

4

The relationship between the chair and entrances establishes the scale of the scene. The scale of the grass should be adjusted to reflect it. Also, make cracks from which the grass is growing more obvious.

5

Adding an extra layer of dirt to the ground will:

1. Match the ref better

2. Increase realism

3. Look great from a distance due perspective

Add a matching layer of dirt to the foot of walls. It not only grounds them in the scene but makes sense, with wind and rainfall splattering dirt from the ground on them.

6

You can read the windows as either rusted metal or glass.

Rusted metal will follow the same pattern of damage from the walls.

Glass will look more opaque on the corners because dirt loves crevices and when looking through a window you wipe the middle, not the corners. You can overlay leak-type dirt to decrease visibility, but be careful to not go overboard.

Glass windows will be inset while metal windows might be a cover to inner windows. (Note: I didn't inset or adjusted the scale of those windows, but you know the jam.)

7

You missed the opportunity to add those yellow parking(?) markings from ref. They'd make the ground more interesting besides helping with spatial orientation.

8

Be careful with decals leaking onto surrounding surfaces.

That's it! I hope it helps. And bonus tip: The reddish hue of the extra layer of dirt on the ground and walls should be enough to make the scene more grounded (no pun intended!), but if it's insufficient then a slight color tweak in the post process or a super soft reddish static light would help. This happens because the sum of bounced colors on the ground level will read as a warmish tone under the overcast sky, so a dash of a warm color here sells the idea there's air between the viewer and the subject.

b_beauchamp

@birb Thank you for the detailed breakdown! I’ll be keeping these notes for the future to hopefully avoid these mistakes in future projects. I think one of my primary errors when I started working on the test was modeling the modular pieces of the building as complete elements instead of creating simpler pieces to work with. This locked me in to using certain proportions and made it impossible to create important scene elements like door insets and continuing corridors. I also used pretty arbitrary dimensions for all of the models, based on what I thought “looked right” which created its own host of problems as you have pointed out.

valentin_baguirov

Seriously I woudlnt get discouraged. Some of the applications I sent out, took like 3 months or even longer for them to get back to me, maybe they were just waiting until I added new pieces to my portfolio, but idk.

I agree your stuff looks pretty good, however maybe focus more on coming up with something thats not based closely to a concept? Like it cool to take ideas from other peoples work, but I wouldnt fill too much portfolio with something thats close to 1:1 to a concept.

teodar23

Nothing wrong with using concept art created by someone else. In fact most environment art jobs require this exact thing. If you can muster some original concept, great, but no one is going to discard your portfolio because you recreated concept art made by someone else. You will probably not have a concept to work off all the time but from what i can gather, recruiters like to see that you can reproduce something, either from real life refs or concept art.

b_beauchamp

@valentin_baguirov @teodar23 I've entered the Artstation environment art challenge. My hope is to apply all the critique I've received here in that project. Thank you for the feedback!