Vvvv vs unity and all the rest (not clickbait)

[mrboni]

I was chatting to Tebjan about what it has been like using Unity for some projects recently and he suggested it would be useful to add my thoughts into open discussion.

From the conversation (all my comments) -

Having lots of premade functionality in one place is great
I like the entity / component model
I think I do anyway
at least it seems to provide a reasonable way to integrate an editor environment with scripting
animation was a total pain though
I used the timeline to try to do some basic things - swapping out images and some movement animation
it was insanely clunky
but at least it has one I guess :)
(I really want a VL or Xenko timeline)
Everything I tried to do was possible without much thinking, but much of it involved so much more fiddling than necessary
There’s a lot to be said for having all the standard tools in one place, with a conventional UI
I think the ideal is somewhere between vvvv and Unity in the UI sense
Unity is bloated to hell, but it has what you need, in a reasonably predictable layout
but it’s too clunky to ‘flow’ when using
which is what vvvv is good at
I wish vvvv was better at basic / standard tasks though
There are various jobs that it’s not getting used for right now that it could be
and we’ve been turning to things like Unity for those. ((my work buddy) even started learning Touch Designer!)
these are all the boring jobs
the exciting ones are where vvvv shines, because they are the ones that require originality
but it would be great for everyone I think, if the standard things were better catered for in vvvvland
I would be happier
and tbh those jobs are where the £ is
everyone is doing them
and they should be getting done in vvvv

Summary -

A regular conversation with the people I work with is what tools to use to get a certain job done. As I’m sure is the case for many people, not all those jobs are super fun art projects. There is a lot of bread and butter work using our skills that doesn’t require much originality or custom workflows, but it does require easy and fast, common tools. I would love to use vvvv for all this work, but it’s just not the best choice a lot of the time, and I don’t think it has to be that way. I would love vvvv to accommodate the more conventional workflows, as a development and design tool. This could benefit all kinds of vvvv project for the better (hello timeline!), and it would get vvvv being used more frequently = more £ rolling in for development.

[david]

thank you for bringing this up here. I do absolutely agree with especially this sentence
" I would love vvvv to accommodate the more conventional workflows, as a development and design tool."

We are all aware of the power that vvvv beta and gamma providing us to solve even the trickiest thing. But on the other hand we absolutely need an environment with a rock solid highlevel node set and the nescessary authoring tools that enable us to do our bread and butter design jobs with workflows that just work out of the box with everytime we start the vvvv application. And this set of functionalities and tools should be something that we can call “this is how you do it in vvvv”. For the future of vvvv this is absolutely essential.

[u7angel]

@mrboni , i can absolutely follow your post, except the unity timeline thing. but i guess we can not reduce unity to that and vvvv for not having a decent one.

actually, we switched the studio for the last 6-8 weeks from vvvv to unity (yes, its all my fault :), just to have a go at it. the result is mixed.

the expectation was that unity is a rock solid product, developed by a big company, with hundreds of developers. in reality, its the same buggy piece of software as anything else. with workflow stopper all over the place. but let’s look at some points.

unity pro

• really good weighted rendering with everything you wish
• cinema4D integration for instant testing/editing of models
• timeline, statemachine, animators (although not perfect)
• multiplattform support which magically works
• small executable project which start amazingly fast
• editor with preview
• C# freedom

unity con

• using git is horrible due to meta files and random changes
• git alternative “collaborate” is buggy as hell, especially with meta files and timeline
• project readability is bad due to component system, which code does what
• hidden menus all over the place
• compositing is only possible through cameras, weird system
• Unity developement is very slow, how often have we seen loading bars
• import of C# libraries not so easy, probably due to multiplatform export

hence, the grass is superficially always greener on the other side. vvvv beta and at some point gamma, is a lot faster to develop and sketch things. and that is the unique selling point. if gamma gets a decent rendering engine and multiplatform export support, it will be easily an competitor to unity.

one important thing i learned while using unity, i missed the dx11 pack “low level” way of compositing graphics. vux did an amazing job here. the beautyshot SSAO lightprobe softshadow blabla rendering engine is only good for certain things and can be a pain if you want to be more free in terms of building GPU things.

anyway, i guess unity can learn a lot from vvvv and vvvv can learn a lot from unity. i definetely missed vvvv while using unity a lot.

[catweasel]

For a recent project I’ve been using Notch, seamless integration with cinema4d, similar cloners/effectors concepts, timeline, media server integration which is incredibly good. It was actually quicker to make things in notch than c4d (after making the basic models) interactive and generative layouts need more work. The experience was fun and really quick to pickup.
Great render engine, fields/volumetric meshing, volumetric lighting, physical rendering built in, substance designer support in the next update, not so good if you want something not in the box as except for javascript and post fx shaders there is no customisation, but the box is pretty well stocked, expensive, but then if you add up the paid contribs for vvvv that are built in, probably on a par.
I’ve found unreal to be much more usable than unity, but again more of a pain if you need things that aren’t included.
If the dev’s wont to make some more money, I would suggest getting the api for integrating gamma in to media servers (ai, d3, hippo all use notch for generative content, and its a big ‘selling’ point in an industry based on novelty and also last minute changes) there are hundreds of users playing monthly subscriptions, and even more rental houses having to buy licences for their media servers, but probably doesn’t fit so well with the trust model. I’d suggest the export be a paid only add-on.

[beyon]

@mrboni

I wish vvvv was better at basic / standard tasks though

@mrboni

but it does require easy and fast, common tools.

@mrboni

I would love vvvv to accommodate the more conventional workflows, as a development and design tool.

What exactly are these easy/fast/common/conventional tools and workflows you miss? Just curious, since it’s not clear from your post (other than a timeline).

[tonfilm]

@beyon

What exactly are these easy/fast/common/conventional tools and workflows you miss?

good question, how about collecting a list?

[antokhio]

well, why don’t we start with proper geometry support gltf, FBX, vvvv and vl lacks this big time…

i mean all this stuff was discussed thousands of times, and it’s easy as what you are use vvvv/vl for?

and the answer is:
realtime genartive stuff
mutitouch
mapping

so what’s would we need? clean pipeline in thouse directions…

so realtime > geometry, import, export, handle materials, animations, cameras whatever…

multitouch > gui e.g.sliders, buttons etc easy custumisible, easy to use add node > loade graphics > done…

mapping > editors, timelines, whatever you need for a show…

vl gives you ability to work with wide array of custom libs on network, but still we are mostly use only ready made plugins…

forgot to mention maintainobility…

[david]

• keep up vvvv’s big advantage of making it relativly easy to scale up your installation to a multi renderer /multiclient system. Hence, provide an easy to use workflow to scale up your software to multiple Window/ clients with a robust clock sync and external sync sources. This is an important selling point as it is industry standard if you want to compete with notch/diguise/ventuz etc.

• A robust movie player for gamma to compete with other media servers in this regards.

• A robust easy to use preset editor / cue editor. Or at least provide a well integrated principle of how to expose parameters to external media controls or to our own timeline.

• That very own timeline with clips and keyframes. It would be good to see these basic functionalities first have some user testings and a testphase to make it stable for production. If this works, then add the things that makes it unique and special and maniac. But lets start with the basics.

• A 2d and a 3d Mapping/warping tool, with softedges etc. multiclient / multirender scaleable

• A texture effect node set to deal with texture based animations. With maybe the ability to preview your effects on the fly. Thats something people love about TD. The whole world of Texture Operator as an easy to use nodeset.

• Generally speaking: It makes sense to provide more ready to use tools that bundle complex stuff in an easy to use interface. Ship these with vvvv releases.

• Integrate these tools in the IDE with dockable windows. Just like we all know it from other sucessfull softwares.

• on the rendering pipeline antokhio is right: Geometry import, materials, assets, animations, camera pathes. An easy to use gui development workflow.

• I know we can create these things on our own now just by using vvvv gamma. it is great to build our own custom tools, yes. And it will work just like that. But how do we provide these tools for the rest of the community? And why should we? There is no way to monetize them other than using them only for your own projects and have a unique selling point with them towards your own clients. this brings up the “store” as a central missing piece for the future success of vvvv and most importantly for its great community.

[manuel]

I think there’s also a different paradigm between vvvv and unity/unreal

Remember a few years ago vvvv was only compared to processing of, max, etc. not really to game engines. Usually when question was “can i build a game” response was “use a game engine”

In vvvv there’s usually a culture of learn yourself how to do it. Which is great.

But not the best comercial approach.

Lets say 8 Years ago I wanted to do a projection mapping. I chose to do it in vvvv, so I had to learn:

how to work with video, codecs, patch your player, patch your controls
how to work with output setup, patch it
how to calibrate, patch calibration module / reuse some example, trial error, ask a lot, etc.

To do it now you have to go for the same process (if you just started), but If you want to do it in Resolume, Its just draggin a file, setup some menus and done.

For sure, dont ask resolume 50 projectors network sync etc. But most comercial project dont ask for that. Thats why readymade solutions are more suitable.

My concern is that this may affect new users joining, since most people dont start using a tool to solve a big complex problem, but to accomplish simpler stuff the easiest way…

Most of the graphics I see in TD projects look great, but completetly the same. I guess thats because It has more closed solutions directly deisgned to solve known problems in projects.

[antokhio]

Well speaking of what we are getting to the main question: my work was better by adding nodes is it better with having states and ability to edit nodes?

[catweasel]

All we can hope @antokhio is that the devs can now more easily make robust addons regarding the missing and my opinion vital features, that we have been waiting for for many years. I just pray we don’t get uncompleted addons like timeliner, and told the source code is there of you want to fix it, which is fine if you are a coder less so if not, in which case we disapear into ofx world where things become orphaned and need magic to get the right verisons of specific libraries to work. vvvv had always had the advantage that came with closed source, of reliable functionality, even if some of that functionality was missing, my fear is it becomes a visual tool for coders more than a visual programming tool for artists.
@manuel I was making what were esentially games as live visual instruments more than 10 years ago in vvvv, as soon as it got shaders it moved into that world, what has changed was games engines became more user friendly and got node graphs, so artists could use them instead of just coders. It was vvvv’s unique selling point, certainly for me!
I hope xenko helps in that regard, as long as that doesn’t become an orphaned open source project in a few years as slim dx has become.
Also regarding mapping, consider what would have happened if vvvv had supported robust video playback, 10 years ago (or even today…), everyone would be using vvvv instead of mad mapper (the devs used to be vvvv users…)

[antokhio]

Yea sorry for that drunken post.
Really looking in to what is VL as a production tool finnaly… And i proud of dev’s doin thier job…

[manuel]

@catweasel

Also regarding mapping, consider what would have happened if vvvv had supported robust video playback, 10 years ago (or even today…), everyone would be using vvvv instead of mad mapper (the devs used to be vvvv users…)

I think it is a robust playback in the sense that It can do almost everything what computers are able to, but you need to learn a lot how to make it work. and patch it yourself.

Same with (complex) games. can you develop a FPS in vvvv? sure…

I think It would be great to have more user adoption all these “readymade” tools we are talking about.

[mrboni]

Helloooooo

Sorry for silence

The time away gave me an opportunity to reflect. Possibly the main area in which I have gripes about the vvvv workflow is in dealing with assets. Getting them in, managing them, and getting them out.

It remains a total pain to deal with any 3D assets. Credits to everyone who has tried to tackle this to date (particularly to the creators of the recent SceneGraph), but it’s still prohibitively difficult to use vvvv in a 3d workflow in anything other than simple cases. I want to be able to import a variety of common formats - fbx, gltf, blend etc and for the contents to be visible and usable directly, without having to maps transforms and textures and hierarchical blah. I also want to be able to update that file externally an easily continue working. I want to be able to use the blend shape animations I’ve made, points, paths, cameras and the rest.

I want to be able to load video and images in all or most of the common formats, including pro ones, and access that data with minimal clicks. I don’t want to always patch dds(!) frame players because nothing else works well enough. I don’t want to have to patch transport controls at all.

Importing assets should be fast, and coherent. I want to know what assets are referenced in my project, and to update / change / gather them easily. When I import an image I should be able to preview that image either directly in the patch or in a multipurpose preview pane that’s part of the core UI. I make and delete dx11 (texture) preview nodes about once per minute while patching. This preview should either be in the UI, or available with a mouse hover. Similar data preview should be available for a wide variety of data types.

Rendering stills and video should be core functionality with smart and sensible helpers for naming conventions. This would make vvvv useful in post production workflows.

Related inextricably to useful asset export, is a timeline. An easily editable, repeatable method of frame sync time based animation is essential to make the this kind of workflow meaningful. The two options in use in my world to fill this gap (when using vvvv) are Ableton Live, and Cinema 4D. There are ways to make these work together but both approaches are restrictive in significant ways. I know many uses of vvvv do not require offline asset export, buy many of even these could still benefit from a way to purposefully animate parameters.

As with sequencing, I’ve found workarounds for most of the above in some form or another. These workarounds tend to get made custom for each project, and soak up a lot of what could be creative development time. They are all DULL to make, and sometimes difficult. I’ve tried incorporating vvvv numerous times with other design and development teams / workflows. A common experience is of people being really impressed at the unique things I can do with vvvv, but that then changes to frustration as it proves painful to do simple things like exchange assets. That can then provoke a switch to a ‘known quantity’ tool like Unity.

vvvv is beautiful and expressive. It has a freedom and flexibility that allows this to be. However it is obtuse, and unneccesarily so. I think it’s totally possible to have the best of both worlds. I love VL, and I’m very excited about the Xenko integration possibilites, but an extra step needs to be taken other than just opening up possibilities. There are solid tools & functions that we all need, to help us make money, and to make art. I’m pretty sure there is also a direct connection between the above and the possibility to new users.

[tonfilm]

@mrboni did you see my comment on asset management over at the gamma presentation blog post? it might not answer everything in your post, but hints the direction we are taking:

vvvv Gamma Presentation

and the good news is, we are basically working on all points you mentioned… details remain to be seen, of course.

[microdee]

@mrboni you’re describing a full fledged media server with notch inside them, not vvvv. There are plenty of options on that front, Green Hippo, Disguise or LightAct to mention a few from top of my head. They all support Spout and NDI so you can also have the crazy vvvv content too

[mrboni]

@microdee - I’m not. I’m just describing a software tool predominantly used for the creation of multimedia work, that plays nice with multimedia assets.

@tonfilm - Yes I did see that. I’m very hopeful about the benefits of the Xenko editor & asset pipeline. Though without knowing more about current & planned state of the implementation it’s hard to speculate on whether it will be comprehensive enough to deal with issues of workflow, rather than just functionality.

[Hadasi]

But @microdee, in-house I think that is what people build with vvvv, often with a view to making a streamlined and job-specific system, but always leaning on the same necessities as expressed here. Sometimes they have tried to build a general purpose system that they want to tailor to their own way of doing things, namely getting vvvv generated content onto a screen in a controllable manner. And @mrboni’s point is that working with other media and other software is a hassle. Its harder to collaborate because while all the media files can be brought into vvvv, it takes more work than should be spent on it.

I know we’re not the Blender Foundation here, but the TextureFX came out of a community discussion about applying effects a la Touch Designer. There is certainly the expertise here to design as components, a tool or suite to fit the above needs, and when the Xenko render engine drops sometime this Summer* I think it will be a lot clearer what gaps would need to be filled get this sort of meta-Notch/Unityvvv/UnVL scripted tool off the ground.

The discussion around timelines at Link last year produced a really interesting design (on paper) and maybe those final thoughts should be published, but I think it’s being worked on… If VL enables users to produce components in a more generalized way, so they can work across projects, maybe they could be developed by the community, or at least a design framework could be decided.

Should vvvv behave more like MaxMSP/Touch Designer with a media-specific workflow in mind or more like Visual Studio, with some designer-friendly extensions to the gui? I dunno. I know the devs are looking at ways to make the VL canvas more informative and some of those additions will be dropped in sooner or later, like datatype registration for designing type specific tooltips (say images as a thumbnail etc) and ‘looks’ to help improve the overview of data flow further down the line, but that’s the software development environment. I’d like to see a suite of tools programmed in VL that could work in or outside of vvvv.

Too slow?

Until there is a VL UI framework in place for making guis with a clear design pattern, the development of these tools collectively is rather out of the hands anyway. So let’s continue talk here about what we need, and then maybe one of us or a dev opens a thread about a specific area where we bring more detail to how we would like the implementation to work and feel.

*probably

[catweasel]

@microdee notch has a timeline built in. I’d be happy for vl to integrate with hippo or d3, as well having a timeline though, and I think it could gain a lot of users that way, as notch has many holes on interaction side of things.