What is the fastest and most efficient way to learn 3D Modeling?

Marie_Allen

I wanted to ask those with more experience than me what they think would be the fastest path to professional modeling for game development.

In my experience, art tends to be more about whether a piece meets the goals the artist set for it, and less about objective measurements of good or bad; knowing this, when I've given critique or advice, I've always found it incredibly difficult to offer any useful responses to vague questions like "How do I get better?" without at least some context for what better means to that person. With that in mind, I wanted to provide some context for what better means to me. I'd also like to thank in advance anyone who takes the time to read this and respond; it takes no small amount of effort to try and condense years of learning into a form that a novice can understand.

I've been doing 2D art (mainly pixel art) for 8 years. I'm a solo developer, so I won't be applying for any jobs, but it's important to me that I reach a level where I could. As of now I've been modeling for 5 days, and I've included a picture below of all my progress over those 5 days. To teach myself modeling I've been using this method:
 1. Attempt to meet my goal
 2. Stop when I feel I've missed the mark by too much, or feel that I've finished the current attempt
 3. Analyze the work to find where I failed to meet my goals
 4. Find solutions to the areas where I went wrong (usually tutorials on Youtube)
 5. Use new knowledge to either start from scratch with the same goals or heavily modify the previous version

My goals for this character, and for all models in my current project, are as follows, in descending priority:
 1. The model should animate as smoothly as possible with as few polygons as possible
 2. The model should easily and appealingly cut apart for modular use
 3. The model should capture a feeling between PS1/PS2-esque low poly models and the gritty textures of Demon Souls/DS1
 4. The model should maintain good edge flow and general topology



Below is version 9. I've completely started over with this one, as while I was beginning to feel the model was meeting my goals with the textures (though they were too bright) and look at V7, I had begun to find that there were issues with my topology when it came to rigging and animating (particularly in the hips and arm joints), something I actually hadn't thought about mattering until the point that I rigged it and tried to pose it; even with PS1 models, there was still a flow to the topology that made for graphically appealing distortions, despite the limitations. After examining some more professionally made models I admit I'm somewhat horrified looking back at the old versions! Even now I feel that I am still too careless when it comes to the use of excess polygons, but my inexperience makes it difficult for me to determine where those cases are. Processors now can of course handle many thousands times more polygons than the amount that I am using, but I have the intuition that my ability to manage poly count in low poly has a direct correlation to my ability to maintain good flow on higher poly counts, so I am taking it very seriously.


My current project and progress are very much in the realm of low poly, but I'm interested in learning all aspects of modeling concerning game development, regardless of poly count. I feel that I've made good progress so far but success now is not a guarantee of success in the future, and whether or not I can keep it up depends on my ability to be able to observe where I'm failing -- something I know from experience will get increasingly difficult the more progress I make. I'd be very interested in what advice more experienced modelers might have to offer, whether it's resources that you've found helpful or your own process for improving. Hopefully I've provided some idea of my goals and current ability so that giving a useful response isn't such a nebulous concept.

Thank you again for your time!

Alex_J

I usually try to crash course tutorials as fast as I can to get a general sense of everything other people already know (or at least what they are making available online). 

Then I'll start doing my own work when I've exhausted all of that. 

This way I dont waste time figuring something out that somebody already figured out. 

Beyond that, just trial and error. Take my time, make sure pacing in a way that its fun every day and not religion. 

Most important advice anybody ever told me, "Just don't quit."

carvuliero

Learn how to model -> learn the tools and how to combine them to solve problems rest is fun

I guess fastest way to learn what you want is to copy someone else models , polycount has this great topic[https://polycount.com/discussion/41232/lowpoly-or-the-optimisation-appreciation-organisation/p1] with bunch of low poly models most of which with wireframs pick one copy it few times to get use to topology and logic behind optimization , look for @braingraft models he is really good

Marie_Allen

Alex_J said:

I usually try to crash course tutorials as fast as I can to get a general sense of everything other people already know (or at least what they are making available online). 

Then I'll start doing my own work when I've exhausted all of that. 

This way I dont waste time figuring something out that somebody already figured out. 

Beyond that, just trial and error. Take my time, make sure pacing in a way that its fun every day and not religion. 

Most important advice anybody ever told me, "Just don't quit."

Advice like this is always so encouraging. I'll keep this in mind and take my time!

carvuliero said:

Learn how to model -> learn the tools and how to combine them to solve problems rest is fun

I guess fastest way to learn what you want is to copy someone else models , polycount has this great topic[https://polycount.com/discussion/41232/lowpoly-or-the-optimisation-appreciation-organisation/p1] with bunch of low poly models most of which with wireframs pick one copy it few times to get use to topology and logic behind optimization , look for @braingraft models he is really good

I'll try this, it sounds fun! I was actually already following Braingraft so it's funny that you mention it. I saw their Vampire Lady model a few days ago on Sketchfab and was shocked at how low poly it was.

chien

i don't think this going to help at all, i follow the common process to get anything done because just want the results, using the blockouts and not focus on any small detail and i recycle alot of assets that are modular

Brandon.LaFrance

I think you're on the right path, and definitely thinking about the right kinds of things. The way you learn this stuff is by doing, so you're already well on your way.

One piece of advice that I wish I was told when I started out:

"Done is better than perfect." 

Especially as you're learning, you're going to constantly find areas that you wish you had done better. This will never change. Its part of the learning process, and how we improve. If you constantly restart from scratch every time you notice these "sub-optimal solutions", you will never finish a project. Its far better to get something "done", reflect on its shortcomings, and then apply what you've learned to the next project.

Now "done" can mean whatever you want it to mean. Just make sure the scope is reasonable. Based on your original post, maybe something like: One character, 1k triangles, UV mapped, diffuse only 256x256 texture, rigged. Give yourself a reasonable deadline, knock it out, evaluate, then put it on the shelf and move on to your next one. You will progress so much faster this way.

Its okay to take extra time to get things right, just don't get caught up chasing perfection. Perfection is just an illusion, and that way lies madness.

"Done is better than perfect."