Home Technical Talk

Looking for opinions from artists and technical folks alike on 3d software features

FlippedPoint
polycounter lvl 3
Offline / Send Message
FlippedPoint polycounter lvl 3
Hello everyone, I’m a coder and I do a little bit of 3d on the side.  I’ve been writing a piece of 3d software for a while now and I’d like some input from you guys. I hope this is ok to post here :)

What are the shortcomings of the software packages you use according to your experience?  
What kind of features have you needed in the past but could not find?
What kind of features would you like added now and why?
Generally speaking, what kind of feature would you like to see in a 3D application? Anything that may seem impossible to implement is welcome to mention, I’m looking to get the creative juices flowing and any idea is more than welcome!  Thank you.

Replies

  • Mink
    Offline / Send Message
    Mink polycounter lvl 5
    I wish blender had a better LOD workflow. That said, I highly doubt you can create a competing program worth anybodies time on your own, so don't worry about that right now.
  • poopipe
    Offline / Send Message
    poopipe grand marshal polycounter

    shape recognition is missing.

    If a package can identify the primitive shapes that make up an object then it can do a huge variety of stuff with the information -  not least intelligently LOD and unwrap them.
  • Thanez
    Offline / Send Message
    Thanez interpolator
    In 3ds max, I would want the possibility of parameters being instances of others, to link them to other parameters.
    For instance in modeling this m84, I made it all out of booleaned primitives. 

    In making the LODS for the green housing I would love if I could go back to the pre-boolean state, switch a 6 for a 4, 2 or 1, and get out my LODS only requiring a re-uv of the affected part.
    in short: if the parameters are cells in excel, I would want to be able to refer them to others instead of having their own unique values.

    And another one from the top of my head. The ability to right-click a UI element to assign a hotkey to it. Most of the elements are very hard to find in the hotkey settings as there are so many with similar or identical names.
  • poopipe
    Offline / Send Message
    poopipe grand marshal polycounter
    You can already hook parameters together like that in max
  • kio
    Offline / Send Message
    kio polycounter lvl 15
    Our check out Houdini if you want to have an application Which is just built around this paradigm.

    At the op, if this is just for fun, maybe better look into adding a feature to blender than to create yet another application which probably doesn't have enough appel on its own.


  • poopipe
    Offline / Send Message
    poopipe grand marshal polycounter
    I agree with that.. It is largely pointless trying to build a full 3d package - maya basically has it sewn up and has done for a long time. 

    It's far better to pick a niche, build something that does it well and make sure it talks to the big packages properly. 

    The biggest success stories in this vein include keyshot, substance and Zbrush but there are plenty more. 

    What they have in common is that they all either do something you can't do in maya/max or simplify something that's enough of a pain in the arse that you save more money than it costs to buy. 


  • FlippedPoint
    Offline / Send Message
    FlippedPoint polycounter lvl 3
    Thanks to everyone for your comments.
    As for whether this is feasible as a product or not, I’m yet to actually decide where to go with this, I’m merely exploring ideas for the sake of ideas themselves, I am fully aware I won’t be able to compete with Maya or Max, or heck even Blender.

    I simply wanted to know what people want in their packages as a way to give myself perspective as to what different people want and need, it’s something I’m enjoying, the journey is quite fulfilling.
  • CheeseOnToast
    Offline / Send Message
    CheeseOnToast greentooth
    Complex automated retopology still hasn't really been cracked yet, AFAIK. It's by far the most tedious part of my job and it feels like the tools to do it are lagging far behind those made for modelling, UV mapping and texturing/surfacing. The existing automated solutions are great for doing quick tests, optimizing geometry for further sculpting in a package like Zbrush or for organic, non-animated objects. They fall flat when you need to make precise edgeflow for animation, which always needs manual control. If you could create something that lets artists define the important parts of the mesh while automating the creation of the rest you'd have hit gold, IMO.

    Poopipe mentioned shape recognition above. If you could include this for sensible automatic retopology of hard surface models that would also be a big plus.
  • poopipe
    Offline / Send Message
    poopipe grand marshal polycounter
    The frustrating part is that most of the maths for this stuff has already been worked out but it's languishing in scientific papers or PhD projects and never really finds its way out into the creative toolsets. 

    Eg. Something as simple as generating median curves (or whatever the 3d equivalent is called) of a solid mesh could inform rig building, topology, even uv layouts. 

    Similar algorithms in 2.5d  give you the foundation for safe autobevelling

    The problem is that the people clever enough to invent this stuff often don't realise how what they've invented can be bent into a tool - with that in mind, it's quite heartening to see a developer coming straight to the source
  • FlippedPoint
    Offline / Send Message
    FlippedPoint polycounter lvl 3
    poopipe said:
    The frustrating part is that most of the maths for this stuff has already been worked out but it's languishing in scientific papers or PhD projects and never really finds its way out into the creative toolsets. 

    Eg. Something as simple as generating median curves (or whatever the 3d equivalent is called) of a solid mesh could inform rig building, topology, even uv layouts. 

    Similar algorithms in 2.5d  give you the foundation for safe autobevelling

    The problem is that the people clever enough to invent this stuff often don't realise how what they've invented can be bent into a tool - with that in mind, it's quite heartening to see a developer coming straight to the source

    That's exactly what I've thought about and what has lead me to think about this and play around with ideas.

    I saw so many potential "tools" and all the math worked out for it and it's basically being delivered on a gold plate for someone to work on and nobody does, especially with the way 3D packages are stagnating because of the whole "industry standard" stuff, every major studio having an army of programmers available for when their artists want a tool and bam gets implemented in house and never sees the light of day, it's a little sad, at least to me.   

    Which is why I'm asking you folks, if I manage to make one single tool and help one fellow artist do one single thing, this whole thing to me would be successful.  Which is why I came here.  To see what artists actually want!  I hope to get more feedback from you guys. :)
  • gnoop
    Offline / Send Message
    gnoop polycounter
       Imo the problem with art creation sofware is not that it's missing a few specific features but rather how monstrously inconvenient, unintuitive  and tremendously artist unfriendly   it is,  even in cases of   "flexible" ones like Houdini.     

      And nothing have changed with that during last 30 years.   We still wasting half  of our life time  trying to dig out what  software folks do and how it works instead of  focusing  on our own creative tasks and art related nuances.      The whole "uncanney valley" phenomena have surfaced because of  that  imo.   In traditional figurative art  such uncanney thing had been disappeared  gradually since 13 century  .  With just  general development of art school and traditions.
     
       I don't have a recipe of how to fix it.   Imo software folks usually have just fundamentally different cognitive approaches or something.   It's why probably it's so hard usually to explain something  to  a shader coder while a fellow artist understands you just after a first word.

       Thus when I see something  like "For 3d artists, by 3d artists"  in a software advertisement  I instantly know it's not true  and would be just another mess I would have to spend a year to figure out.    Some "artist focused" software companies couldn't even write a help that  really helps and not causing more questions then answers.  

    After so many years we still couldn't get an easy solution for an adequate visual feedback in 3d tools, close to a game, not even in a soft like Substance Designer  without lots of special voodoo.   So I don't believe things could change in close future.  Maybe in 200-300 years :)  With  neural something.  After all artists used animal bladders for saving their paints from drying during thousand years  before some bright guy invented paint tubes with a cap. 
     So far "convenient  3d/ 2d software " is an oxymoron  like "dry water" . 
  • FlippedPoint
    Offline / Send Message
    FlippedPoint polycounter lvl 3
    This is where you're mistaken, my friend.  I'm a 3D artist myself (sort of, not a professional but I'm more than an advanced sculptor and modeler).  I've suffered through all of the tools and I've managed to somewhat master ZBrush (by that I mean know what almost everything does and so on), but I've come to notice a huge gap between execution and MASTERY of execution, which is why I'm asking here now.

    If I ever implement something in a piece of software, I'll make it a few click solution, I'll try my best to do that or at least point artists in the right direction.

    And I'm afraid I have to disagree with you when it comes to "convenient 3d/2d software" is an oxymoron, it really is not.  All tools are somewhat simpler once you actually spend some time, some more than others ie Modo vs Blender.  
  • gnoop
    Offline / Send Message
    gnoop polycounter
    And I'm afraid I have to disagree with you when it comes to "convenient 3d/2d software" is an oxymoron, it really is not.  All tools are somewhat simpler once you actually spend some time, some more than others ie Modo vs Blender.  
    I might be exaggerate it a bit here and of course any tool implies some time spent to master. Still what a typical pro level software usually is:

    1. A space shuttle cockpit kind of  a mess on a screen , countless check-boxes ,buttons and sub-menu   organized in some random "crazy programmer" style of logic.    To do something you have to check in here, than there and there and don't forget to have a certain feature on in a  certain god forgotten  deep corner of a soft.    I have a huge list  of such  "hidden features" in my Evernote.      Oh, and things are still not working as expected anyway because  of   well,  you may have been done a mistake before and have to go back and figure out what it is  and  do it completely on your own, no hints form the soft.

    2. Help system that could be understood  only by person who wrote it .  

    3.  Lots of lacunas and shortages nobody would tell you about  before you would happily discover them after a year spent on learning.

    So to do something you have to fight with a software non stop usually  while searching for workarounds, searching for compatible 3d party solutions  and swearing badly  in the process.:)  Half of this forum is for that. 

    And finally when it works you just have no time any more to focus on actual art nuances.

    My guess we would get convenient tools only when such tools acquire a kind of AI behind them  and UI would reflect an artist styled   " from a general idea  to  a more specific one" way of thinking.
  • huffer
    Offline / Send Message
    huffer interpolator
    Here's an idea - try teaching a 3D program to a group of people or kids and gather ideas from there. You'll see how they expect the tools to work and get a pretty good idea on what is intuitive and what not. You can also ask them how would they want to make this or that operation - basically how a tool should work at the simplest level, and it will be way ahead than what we're using now.
Sign In or Register to comment.