Home Technical Talk

Fusion 360 in a Game Art Pipeline?

polycounter lvl 5
Offline / Send Message
Zalek4 polycounter lvl 5
Hi all!

I had never really given Fusion 360 any thought given that it's a CAD program. I first heard about the program in my earlier days of learning the basic high-poly/low-poly workflow when I stumbled on this video:

https://www.youtube.com/watch?v=fS8SBxkXF7c

The 3D artist who worked on the 'Razor' said he used Modo and Fusion to make the ship. Modo I got, but Fusion???
Since this program is used by one artist at Cloud Imperium, a huge company, am I correct in assuming that other artist there would use it in their work as well?

I also read this article on 80.lv by Gregor Kopka:

https://80.lv/articles/building-guns-for-doom/

He worked on the weapons for Doom, and in his article he says he leaned on Fusion quite a bit during his production process. 

I only just got to sit down and really try my hand at the program today, and I must say it feels sort of like it's too good to be true...
It's incredibly simple to use, and it's also easy to get really complex shapes very quickly. And I mean STUPID quickly. 3DS Max will even optimize the exported .iges file when I import it for retopology (a CAD file format to my understanding).

Does anyone else use this in a modern game production environment? What are some drawbacks/benefits of Fusion rather than sticking with standard poly-modelling tools? Does anyone see this tool/similar tools being integrated into the game industry in the near future? Is it worth learning the program and maybe doing some portfolio projects with it?

Thanks for any answers!

Replies

  • pior
    Offline / Send Message
    pior grand marshal polycounter
    Last time I used it (which was a few months ago) it still didn't allow for intuitive placement of points and curves in 3d space. Meaning that the program gets extremely frustrating when modelling anything based on an approved piece of concept art typical of a game art pipeline.

    It's great to create CAD-like parts VERY carefully and tediously - like the things one would need daily in a CNC machine shop, or individual components of an assembly (model kit masters, and so on). It's also a fun tool for folks allowed to semi-improvise designs with loose constraints, basically letting the tool drive the design process.

    But when it comes to the typical requirements of game art hardsurface models  (accurate execution, high amount of surface details, tight delivery deadlines, need to easily edit models after the fact, and so on), the Fusion honeymoon phase ends very quickly. People love fantasizing about this software, but just like any CAD environment it is nowhere near as fluid and responsive as it first seems.

    Oh and, their online licensing system sucks. Last time I checked they were still forcing down updates on startup with no way to opt out, which is a no-no in any serious IT environment.

    In short : it's a very cool piece of software but it is not as stable and powerful as it seems, especially for game art. But if you get into it with lowered expectations you might find some interesting uses for it.
  • oglu
    Offline / Send Message
    oglu polycount lvl 666
    it depends on the asset... if there are complex shapes its often the case you are faster doing that in Fusion... and its cleaner... in the end its a good mix of maya and fusion...

    try to model such stuff without Fusion... 
    https://www.artstation.com/artwork/E5dZq
  • pior
    Offline / Send Message
    pior grand marshal polycounter
    Yeah, but such "freestyle 3d concepts" are not exactly part of a standard pipeline. It works for some rare studios and projects (Wolfenstein mechs and weapons for instance), but it is far from the norm since concept and execution are in most cases split between two specialized artists, and for good reasons. One could try asking the same artist to model a typical Feng Zhu spaceship in Fusion : the results may come close ; but even if they do, the turnover time will likely be way too long ...

    That's one of the issues with popular youtube 3d content and some portfolio pieces : they reflect a distorted image of the realities of production imho.
  • oglu
    Offline / Send Message
    oglu polycount lvl 666
    i dont think so... we get really good concept art we have to nail... 
    and wolfenstein is done with modo (fusion) and not fusion 360...


  • pior
    Offline / Send Message
    pior grand marshal polycounter
    Oh, I was just bringing up the Wolfenstein mechs and weapons as a rare exemple of a single artist doing both concepts and freestyle 3d execution on the fly (regardless of the app being used). It certainly looks great, but it is the exception and not the norm.
  • cromadbomber
    Offline / Send Message
    cromadbomber polycounter lvl 10
    I find Fusion really good and fast for making real life weapons.
  • beefaroni
    Offline / Send Message
    beefaroni sublime tool
    With any software it tends to boil down to..

    Can you do something much faster with a new piece of software and meet the requirements of the pipeline? (And how much $$ it costs too)

    Does anyone else use this in a modern game production environment? What are some drawbacks/benefits of Fusion rather than sticking with standard poly-modelling tools? Does anyone see this tool/similar tools being integrated into the game industry in the near future? Is it worth learning the program and maybe doing some portfolio projects with it?
    You should try an answer these questions for yourself. Just because someone says that you can use "X" piece of software to make "X" asset doesn't mean it will work in your / your teams pipeline. I kept hearing about it so I decided to do a few personal projects in order to prove out whether or not I thought it would be viable to introduce at work.

    I have found that there are areas where Fusion is really great and areas where I tend to avoid it. It really depends on your pipeline and where the time savings are. 

    Good luck!
  • Zalek4
    Offline / Send Message
    Zalek4 polycounter lvl 5
    Thanks for the info and quick responses, guys!

    So I'm gathering that there is some conflict on this issue...

    @pior you say that Fusion doesn't really support,
     intuitive placement of points and curves in 3d space. 
    I could just be misunderstanding, but isn't the Fusion workflow based on doing just that? From what I've experienced, the program can be very, very non-destructive if used correctly, and it's easy to go back and change/adjust shapes.

    You also say that,
    the program gets extremely frustrating when modelling anything based on an approved piece of concept art typical of a game art pipeline.
    I haven't seen a Fusion tutorial that didn't rely heavily on drawings/photos, be it a CAD or a game art tutorial. For example:

    Modeled from real world reference

    Modeled from a concept sketch

    I do get the forced updates thing though, and that's a bummer :/
    I have a relative who works in IT and it's a total pain for his team to keep up with counter-patches for software updates developers push out.

    @beefaroni
     I actually just started a new project today with Fusion, and I'm quite excited about it! I just wanted to see how widely used this software actually is in the industry given the lack of that info anywhere else I've looked. I have to manage my free time as a student wisely when learning new software. It's often easier to get warned early on than to try something new and hit a brick wall that you've all known about for a while :smile:
  • pior
    Offline / Send Message
    pior grand marshal polycounter
    Zalek - That's precisely why the comment above by Beefaroni is 100% on point. It's near impossible to guess if something that works well for a specific artist and/or task will work for another, especially when users have different backgrounds and expectations.

    I could attempt to write in full detail what I mean by "intuitive placement of points and curves in 3d space" (which is unrelated to the modeling paradigm being destructive or non-destructive), but that wouldn't get you any closer to knowing if this tool will work for for you and for the specific models you need to create :)

    It's also worth noting that different users have different levels of tolerance to certain workflows and UXs. Something that one might find unacceptable (because of a process being too slow or too convoluted compared to another program of choice) could be totally fine for someone coming from a different background.

    In short, there are just too many factors at play. Simply use the program for its strengths, and try to not get too distracted by "what you hear people say" and by flashy promotional videos.

    You can actually easily tell that you will have to jump through quite a few hoops just by reading between the lines of that one video you linked : the artist clearly mentions that he had to use Fusion ... but also Modo ... and also Sketchup, just for that one piece. Such a combo may be fine in some cases ("concept" pieces without tight time constraints and low quality control requirements), but not so appropriate in others. That's all I am trying to get at really.

    Anyways, you'll see all that soon enough with practice :)
  • pior
    Offline / Send Message
    pior grand marshal polycounter
    Actually ... for the sake of clarity I should probably provide a practical example. I personally don't do much modeling anymore, but here's a hardsurface facemask / helmet I did a few years ago. It is fanart, meaning that it has to stick very closely to the source material ; but said source material mostly consists of rough 2d sketches and a few painted illustrations, without any blueprints or orthos since it is a made up fantasy/scifi object originally intended for a sprite-based character.



    https://www.artstation.com/artwork/D31yA

    Hard surface ... sharp edges ... metal ... all these terms sound like something perfectly appropriate for Fusion360. But in actual practice there is also : very subtle surface transitions ... facial features .... and overall a certain organic feel.

    It may be possible to reproduce such a design in Fusion *if* very accurate blueprints were provided. And it may be possible to reach something close, *if* some improvisation and wiggle room were allowed. But the reality of game art production is such that such source materials are rarely provided, especially when the setting is stylized scifi/fantasy. In other words :

    "I haven't seen a Fusion tutorial that didn't rely heavily on drawings/photos [...]"

    And that's precisely my point :) A game art modeler will often *not* be given such tight source material, while still having to 99% stick to a provided piece of concept art which can vary from loose to tight. The best game modelers out there are the one who can work from the least amount of input, while still "making it work" on the fly, cleanly and fast.

    Overall that's really what I am trying to get at. Yes, Fusion and CAD programs in general are fantastic tools. But since your question is so wide reaching ("is it appropriate for game art as a whole ?") then I simply have to chime in to mention that it is indeed great at certain things ... but not all. Don't let the feature list trick you into thinking that it may be a be-all end-all solution, because it isn't.

    I hope this makes sense !
  • Zalek4
    Offline / Send Message
    Zalek4 polycounter lvl 5
    Pior - thanks a bunch for the clarification and example! It's definitely not as black and white as I thought it was. I think ultimately Fusion is a good program to learn, but sticking with Maya, Max or Modo as a main program seems like the best idea.
  • JacqueChoi
    Offline / Send Message
    JacqueChoi polycounter
    I used to be a NURBS modeller back when it was Alias PowerAnimator on an SGI Octane2 back in the 90's.

    But the truth is, it's basically the same as drawing with Vectors in illustrator. I see the value when you're doing  surgically precise objects that vectors might be good at (fonts?), but 99.99999% of the time, I find polygons (pixels) to be easier achieve the results you want.

    I don't want to be so dismissive of NURBs, but can someone provide me examples of when its more efficient/effective than using a Boolean method (that allows you to keep your low-poly mostly intact)?


  • poopipe
    Offline / Send Message
    poopipe grand marshal polycounter

    Where CAD packages shine is in creating objects that were (or would be) designed using CAD packages. 

    This holds true for any modelling paradigm and there are trends that stem from the software available to industrial designers at the time things were designed. 

  • JacqueChoi
    Offline / Send Message
    JacqueChoi polycounter
    poopipe said:

    Where CAD packages shine is in creating objects that were (or would be) designed using CAD packages. 

    This holds true for any modelling paradigm and there are trends that stem from the software available to industrial designers at the time things were designed. 

    I see how that's an interesting theory, but I don't see how it's relevant or true.
    Industrial designers need to model things that Game Artist do not. We don't often have to model inner workings of functional mechanics.

    Even if we did, I believe it would be easier with polygons anyways.

    Its literally the exact same things as asking concept artists if they prefer drawing with Adobe Photoshop, or drawing with vectors like Illustrator.

    I'm sure there's very few instances where using Vectors (or NURBS) might be better, but considering the nature of rasterizing the graphics, (or engines needing to render everything in polygons) I can't name any actual practical uses.

    Sorry for sounding obtuse, but can you show me an actual example of something that would be better produced in CAD over polygons??


  • Amsterdam Hilton Hotel
    Offline / Send Message
    Amsterdam Hilton Hotel insane polycounter
    Sorry for sounding obtuse, but can you show me an actual example of something that would be better produced in CAD over polygons??
    Consider that a boolean-heavy workflow, so popular today for hardsurface work, and popularized in no small part by myself, is largely imitating a CAD environment with polygons. Defining matter additively and removing it subtractively is at the heart of all modeling in any environment, but only in polygons do we have to worry about resolving the base forms together, connecting everything, managing transitions. Their advantage was always in computational efficiency rather than work speed, and this erodes by the year. The advantages we see in Fusion, or in boolean-centric workflows, is precisely in abstracting the minutia of polygonal resolution out of the working process, and focusing on a higher plane, that of Design. If CAD affords you greater precision, reversibility, and speed, what do polygons afford you? Well?

    I call it unlikely that any of us will be working at such a low level as individual polygons, edges, and vertices 10 years from now.
  • JacqueChoi
    Offline / Send Message
    JacqueChoi polycounter
    I believe the exact advantage of NURBS over polygons, was actually that it was projected, and therefore had infinite levels of resolution. Just like Vector graphics has over Pixels. You can scale up an Illustrator document a few hundreds times and it never loses its resolution. Just like NURBS you could zoom in as close as you wanted.. and theoretically it would be a perfectly lathed curved surface (which is why it's still used in industrial design).

    This was the primary reason NURBS was used in film back when I started learning 3D (mid-90s). Computers simply weren't powerful enough to handle working with more than a few hundred polygons on screen at a time. 

    Projecting surfaces with NURBS stopped being a more elegant solution in film/VFX in the mid 2000's.

    With Fusion, its not "tangible". It's working with vector projections and calculated curves. It's a step removed from 3D shapes, and requires a whole other logic and foresight that isn't necessarily faster, and is probably way more limiting. (And has much different challenges resolving base forms).


    The boolean workflow (thx for that) is now heavily incorperated into Zbrush' booleans tools, and has its philosophy based in cad modelling, but in practice is a Bajillion times more efficient.


    I don't want to sound like a jackass, but I'm confident in saying I see no point in NURBs modelling in games. I really want someone to show me an example of something that NURBS can do more efficiently in a game pipeline.
  • Aabel
    Offline / Send Message
    Aabel polycounter lvl 6
    Polygons are still vector graphics, just like NURBS. Voxels are a much closer analogy to working in Photoshop.
  • JacqueChoi
    Offline / Send Message
    JacqueChoi polycounter
    Aahh right you are.

    I guess I meant a Polygon is a tangible representation of a 3D form on a screen. Similar to how a pixel is an accurate representation of a pixel screen.

    A NURBS curve, and a 2D vector are not really properly represented on a screen, and need to be converted to Polygons/Pixels to be represented.


  • poopipe
    Offline / Send Message
    poopipe grand marshal polycounter
    poopipe said:

    Where CAD packages shine is in creating objects that were (or would be) designed using CAD packages. 

    This holds true for any modelling paradigm and there are trends that stem from the software available to industrial designers at the time things were designed. 

    I see how that's an interesting theory, but I don't see how it's relevant or true.
    Industrial designers need to model things that Game Artist do not. We don't often have to model inner workings of functional mechanics.

    Even if we did, I believe it would be easier with polygons anyways.

    Its literally the exact same things as asking concept artists if they prefer drawing with Adobe Photoshop, or drawing with vectors like Illustrator.

    I'm sure there's very few instances where using Vectors (or NURBS) might be better, but considering the nature of rasterizing the graphics, (or engines needing to render everything in polygons) I can't name any actual practical uses.

    Sorry for sounding obtuse, but can you show me an actual example of something that would be better produced in CAD over polygons??



    Good questions. 


    the answer to your last question is this 



    You could build that with subdivided polys,  but I'm guessing you'd rather not have to.  


    My point above was that the media used to create something will in some way define it's form and if you're tasked with reproducing the object, then using the same method will likely lead to less ballache. 


    It makes no difference to the end result - this is true. The impact comes in terms of whether you're using a tool that lends itself to the process or not - if you can achieve the same result in quarter of the time by switching tools, it means you can make either 4 times as many things or take much longer poop breaks.




  • JacqueChoi
    Offline / Send Message
    JacqueChoi polycounter
    No Loops:

    AmsterdamHIltonHotel kinda pioneered this method:
    http://polycount.com/discussion/168610/3ds-max-zbrush-proboolean-dynamesh-hardsurface-workflow-tutorial/p1
    (Which is the methodology behind the Zbrush hardsurface pipeline).
    Boolean to your hearts content, and then just dynamesh/polish.


    The biggest advantage for using Max, is adjusting boolean operands for a very quick lowpoly:




  • Unknown_Target
    Offline / Send Message
    Unknown_Target polycounter lvl 6
    I regularly blend CAD and SubD in my workflow, using Modo's PowerSubD plugin, which is more powerful than Fusion (it supports edge weighting, whereas I don't believe Fusion does). However, I do mainly product design, and whenever possible I lean on MeshFusion in Modo.

    That being said, CAD modeling will be superior for most mevhankcal models - you're basically working on it in the same system that was used to create it. However there is currently no way to get clean polymeshes out of CAD - whatever you do it'll have to be retopologized. Or you can use it for a normal bake. :)
  • oglu
    Offline / Send Message
    oglu polycount lvl 666
    mehses from fusion are really good and clean... its not that bad... its triangluated but ready for baking...



    Model done by:: Carlos Fuentes
  • bakamund
    Offline / Send Message
    bakamund polycounter lvl 4
    oglu said:
    mehses from fusion are really good and clean... its not that bad... its triangluated but ready for baking...

    -----------------------------------------------------------------
    For detailed HPs with realistic details it's very fast. However, I'm still trying to find a way to quickly generate the LP to bake down the HP into. 
    Zbrush's zremesher isn't really giving me optimized LPs to work with. I rather not retopo hard-surface stuff manually, seems like I will remove all that CAD precision in my LP topology.

  • poopipe
    Offline / Send Message
    poopipe grand marshal polycounter
    ^^
    Not sure if maya can do this sort of thing but in Max you can bring in the objects as CAD data and adjust tessellation etc. as much as you want.  
  • napelazam
    Offline / Send Message
    napelazam polycounter lvl 6

    in this case its better to let the software make as little tris as possible or?

    u dont need quads in this objects?

  • Neox
    Offline / Send Message
    Neox veteran polycounter

    as little as in as few or as small?

    because if as few, why do you think you would need quads in this mesh?

  • napelazam
    Offline / Send Message
    napelazam polycounter lvl 6

    i mean reducing amount of tris. u dont need quads for exporting

  • Neox
    Offline / Send Message
    Neox veteran polycounter

    but this isnt supposed to be the lowpoly. this gets gaked down onto the lowpoly. so add whatever geometry/triangles you need to get nice and smooth shading.

Sign In or Register to comment.