Search Unity

  1. Unity 2019.2 is now released.
    Dismiss Notice

anyone knows a good tutorial to learn low poly art for games?

Discussion in 'Game Design' started by brunoenvia, Nov 4, 2019.

  1. brunoenvia

    brunoenvia

    Joined:
    Aug 5, 2019
    Posts:
    40
    i wanted to create my own assets and background assets for my games

    i always liked low poly art. anyone knows a good tutorial to learn?

    any recommendations?
     
  2. Volcanicus

    Volcanicus

    Joined:
    Jan 6, 2018
    Posts:
    163
    Youtube is rich with this and folks like Jayanam show you how to build from scratch.
    If you like low poly, I highly recommend buying a set from Synty and others since it will come out cheaper than doing it all yourself.
     
  3. BIGTIMEMASTER

    BIGTIMEMASTER

    Joined:
    Jun 1, 2017
    Posts:
    2,697
    Google is the main tool you will use. That is the most important thing anyone can tell you.

    Second, to learn the very basics of what buttons do in Maya, max, or blender, pluralsight, Lynda, or 3dmotive all have thousands of neatly organized tutorials to get you going for first year or so.

    The main thing is to learn about art fundamentals though. For that I recommend studying all types of mediums. Marco bucci has great channel.on YouTube. Couple key terms to consider are composition, color, value, lighting, silhouette, shape.

    Remember that almost any question you'll habe for a long time has already been asked of the internet many times.already. you will get fastest and most thorough answer by searching. Never trust one person's opinion. If one hundred people have answered, read them all. Now you.as smart as one hundred people.

    I'll make one counterpoint to some common advice I don't agree with about learning. Some say to ditch tutorials quick and learn by trial and error. I think this is a very stunted and non efficient way to learn. Entire point of being human is generational knowledge accretion. Taking hours, days, or even months to refigure out something people learned a thousand years ago is just a waste of time. We want to be moving forward, so learning then is a game of catch up.

    You will still have to do much trial and error troubleshooting that will beget understanding. It is inevitable. I try to always do a tutorial three times. First I just skim watch to get the big idea. Next I watch and follow along, pausing as I need. This pass is just about familiarity and repetition. Lastly, I repeat what I just did but with my own imagination. The sets what I learned into stone.

    So the fastest way to learn, I believe, is also very thorough. But you will suffer if you try to rush it. Sometimes people misunderstand and think I advocate for doing things the quick sloppy way. What I am saying is to turn a single tutorial into a complete project that gets broken into a planning stage, a training phase, and an execution phase. Most people won't have the patience for that, but I think it pays off in long run.

    Anyway, good luck. Pplycount is a good forum.and wiki for game art, u should bookmark that.
     
    Socrates likes this.
  4. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    4,572
    Low poly (<1000 tri per characters, ie psx, n64, nds) is actually difficult because it's all about economy, it stress understanding of all art aspects.

    You should probably settle for mid poly (3000-8000 per characters, aka ps2, gc, xbox og, dc) with current hardware, where you can apply good practice without having to optimized shape so much, it give you just enough to make distinct shapes. Even with low end machine you are unlikely to run out of polygon, except on low end mobile, but they still are in these range. You can even get away with unoptimized uv that make it easier to pain.

    Then once you have done at least 10 characters and scene, try to get lower polycount and achieve state of the art optimization like skipping bones, squash and stretch implied animation, and UV tetris with pixel alignment.
     
  5. koirat

    koirat

    Joined:
    Jul 7, 2012
    Posts:
    98
    I found low poly tutorial on youtube

    ;)
     
  6. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    4,572
    I'm sorry, cubism is about trying to represent subjects from all angles at the same time, this very obviously NOT low poly, it's UV UNWRAPPING!! :p
     
    BIGTIMEMASTER likes this.
  7. Murgilod

    Murgilod

    Joined:
    Nov 12, 2013
    Posts:
    6,122
    I kinda disagree with this. Low poly is a great starting point because it forces you to quickly start understanding how to budget where you apply detail. As a starting point, it works as a way to teach you fundamentals that you'll need in mid-through-high poly budget models and also make you consider your topology, which will make retopologizing higher poly models into lower-to-mid poly meshes either.

    In my mind, low poly modelling is as important to 3D as gesture drawing is to illustration.
     
    Martin_H likes this.
  8. kdgalla

    kdgalla

    Joined:
    Mar 15, 2013
    Posts:
    1,788
    <1000 for a character model is more like SUPER low poly :p. I'd consider 2000-4000 still pretty low poly, myself.

    I don't think that low poly modeling is necessarily going to be more difficult for everyone. I think it's actually a good way for beginners to start out. I don't necessarily mean "super optimized", though. I just mean "don't shoot for 3000 polys if you're already comfortable with less."

    The phrase "low-poly" is actually what got my foot in the door, as far as modelling goes. At first I felt like I couldn't model anything, especially as far as characters. It was only when I started focusing on "low poly" learning materials (simple shapes and techniques) that I could really start making progress. Pretty soon I could model anything (as long as it was low-poly).

    Less polygons also make your rigging and UV mapping easier too (although having too few vertices is also a problem with rigging).

    I don't really like that modern flat-shaded/faceted look much at all, though.
     
  9. BIGTIMEMASTER

    BIGTIMEMASTER

    Joined:
    Jun 1, 2017
    Posts:
    2,697
    Some years ago there was a boot camp for modelers on pplycount in which trainees made five hundred tris or less model everyday with hand paint texture. Started as simple props and went into critters and characters. I think it was around six months course.

    Seemed like really positive results with most trainees.
     
  10. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    4,572
    Not under 1000 tri, notion like edge flow flies right out the windows at mach speed. You just don't have the intuition of what constitute good topology.

    You can still have budget issue with mid poly BUT you have enough tri to think about edge loop and have quad only topology. Optimizing is great if you actually understand basic topology. And you have enough tri to make a fully articulated character (with expressions and distinct finger around 3000).

    In before making haiku, you need to be able to form full sentences.


    edit:
    I'm bias toward character that is ... where my number comes from ... :oops:

    Also I'm working on a system where you can generate procedurally character with good (enough?) topology from just a single point and a bunch of extrusion instruction. I'll probably make a thread about it soon to gather advice.
     
  11. Murgilod

    Murgilod

    Joined:
    Nov 12, 2013
    Posts:
    6,122
    When I was young, 1000 tri was what was considered reasonably high. I remember 3D Realms job listings for Duke Nukem Forever where they were looking for a high-poly modeller comfortable working at 900 tri poly budgets. To say that edge flow flies out the window there is pretty ridiculous.
     
  12. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    4,572
    I mean for learning it, it makes it hard, you do have to micro manage. Also the standard and practice of modeling have evolved. I do remember ye old time too, and a lot of old time topology were truly terrible compared to what is being done at the same count today (look at some NDS games, where the polycount is hardware capped at 2048 tri, that's less than state of the art psx and n64 games). Also being able to progress at the same time the state of the art is a very good advantage IMHO, also a smoother curve needed. Nowaday we don't have to go through the hoop and loop to be able to learn way "faster".

    But when I say edge flow, i'm not just talking about the general sense of pleasing distribution of geometry details, I mean it in the topological sense, that is quad that follow each other, IMHO learning to design with only quad first, that is understanding the edgeloop and pole distribution impact on shape (that is loop tend to enclose curvature), help form an intuition that feed better as to when decimate, and more importantly there is no pesky triangle placement to mess you up.

    Quad are basically the training wheel. You can get a person up and running on character design in only one day with mid poly, then nudge them toward low poly and deep much more about joints and face distribution.

    But that's my argumented opinion, I'm not saying you are wrong, just laying bare my rational.
     
  13. BIGTIMEMASTER

    BIGTIMEMASTER

    Joined:
    Jun 1, 2017
    Posts:
    2,697
    I think you are using some terms for double duty which leads to confusion. I think you meant silhouette when you said edge flow.

    I think speaking of edge flow or topology refers more to the layout of your.geometry as it pertains to animation, and optimization.
     
  14. Murgilod

    Murgilod

    Joined:
    Nov 12, 2013
    Posts:
    6,122
    Stop explaining modelling to me like I am 5 years old. I am literally doing all the modelling for my project.
     
  15. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    4,572
    I'm not explaining you modeling, :confused: I'm explaining why I argumented about how it relate to the experience of introducing someone to these concepts (that's literally the entire POV!). I didn't explained what a topology or a edge flow is to you, I just assumed you already knew. I actually skeptical that a noob can actually follow what I said because a lot is implied. I actually just put point of reference so you know that I know what you know and so that we can meet in the middle.

    I think that's a bit unfair of a statement to drop upon me :/
     
    Last edited: Nov 5, 2019
  16. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    4,572
    Okay maybe there is too much implied in my post as I realized.

    But basically the context is that it's someone asking to start with modelisation. In my experience, the faster way to jumpstart someone is mid poly with the subdivision/extrusion methodology, then have them transition to low or high poly.

    I don't think topology is just for optimization or animation, the edge flow topology basically help conceptualize how to obtain silhouette, and it make it easy to manipulate adding and removing details for this level. Mid poly is at a level of complexity that is nice with no too much unnecessary economy to do nor too much details to have. Which help transitioning to the extreme of low or high poly. That is, to start low poly fast, learn mid poly first.

    The thing is the more you go to the extreme (low or high poly) the more you have to strategize, low poly mean a lot of cost benefit thinking to cheat around limitation, which mean you have to know how to manipulate a model to begin with. High poly is all about IMHO managing the complexity of details to avoid micro managing single vertex. IMHO mid poly is also way easier to "upgrade" toward both extrem, by adding or removing details strategically. I'm arguing that low poly and high poly are at least intermediate level.

    BUt here is another implied, topology is not necessary, it's a useful tools (hence why I presented them as the training wheel), it's a bit redundant if you have already learn some proficiency in both art and modeling, you will get the result intended by your own mean. And if you are modeling for very long, it doesn't matter as much as you have internalized a lot and the growth was probably at your own pace. I'm arguing for the painless path of least resistance. I learnt modeling with grid paper and reporting back coordinate to the SOF engine my friend was making (never finished) gave me a great insight about modeling, would I recommend people to start there? no, because it's not about me, you have to understand where people come from.

    BUT then one big error of my argument is that I'm character centric, it's a lot less true for most inanimate object. I generally introduce people to modeling with the gingerbread man, and show how you can quickly differentiate silhouette and build details by looking (adding, manipulating) at edge loop.
     
    BIGTIMEMASTER likes this.
  17. BIGTIMEMASTER

    BIGTIMEMASTER

    Joined:
    Jun 1, 2017
    Posts:
    2,697
    I agree all of that. I was just reading your guys talk and thought some words were getting shuffled around which led to misunderstanding. I'm understanding what you are saying. Or at least I think I do.
     
    neoshaman likes this.