By the way, I’m really interested into learning houdini for simulations like you do, and I would like your advice.
Fluid simulations are obviously 100 times better in houdini…
But what about Pyro, explosions, particles… ? Blender is IMO opinion not that bad in theses domains ? What is the real addition of using houdini (true question).
Thanks 
Hi rattle-snake
If i am not mistaken Houdinis team got academy award recently for technology behind Pyro (or mantra). Too lazy to check so don’t quote me on that. It renders very fast on CPU, it has very fast OPENCL( GPU based) simulation mode. It is defiantly one of the strongest solutions out there. I also believe we will be able to export Pyro simulations to Blenders .VDB when it becomes available so this is really big!
The general question is how can Houdini help Blender. One of the biggest problems I feel with Blender is that you cannot have .dll plugins like with other applications. Instead you need custom compilations that fragment the workflow and get out of date. But there is a FREE way for Non commercial users and 199$/year for indie users - Houdini
Virtually any functionality, any plugin you can imagine can be defined there through use of nodes and in theory hooked to Blender, Maya or other softwares. Maya has Houdini Engine for that reason, also Alembic. Infact houdini even has python module (for 2.8 i believe) which means provided Blender was on py 2.8, you could execute most sophisticated plugin functionality simply from Blenders Python script. Well we are not that lucky, there are a bit manual workarounds but it’s still indirectly doable*
From Houdini indie users point of view, one of the MOST REQUESTED feature is GPU RENDERING. Houdini Indie does not allow external renderers such as Octane (need full license). Houdinis MANTRA rendering solution is awesome for pro studios, but normal people do not have render farms at home. This is why Blender is such a GEM of a software as sending Houdini generated data to Blender and rendering it with Cycles is a dream come true. The only missing link is Alembic which would enable importing and exporting most sophisticated animations, simulations and FX. I really hope to see it happen
TL;DR: Blender + Houdini = no plugins needed. Missing link - Alembic.
Hey !
Thanks a lot for this great and detailed answer.
I’m really motivated to try it (I imagine there is a steep learning curve/ really different philosophy).
Blender should be fine for animation (characters…), rendering, compositing, modeling, sculpting, light simulations…
Houdini for heavy and complexe simulations.
I didn’t know Houdini Indie forbids users to get Octane render. It’s true that a GPU renderer can really speed up the thing in some situations (even if in my experience on heavy scenes it could becomes less powerful than CPU).
As you said missing link => Alembic 
Thank’s for the explanation @CGStrive, I hadn’t looked at it with that perspective.
I was also reading the thread about the Autodesk layoffs on CGTalk, and as it is, that’s not really big news; as a business strategy every big company has layoffs each year, and a bunch of people gets hired too.
But I find people’s reactions interesting. While AD products will continue to be the most used, users seem to be increasingly dissatisfied with the new releases, and a consequence of that is the search of alternatives, in which Blender often gets mentioned. Right now, even if people still think it is not a production ready software, it gets recognition as a powerful modeler that can play well with other packages, such as Blender+Houdini as you just said.
Cycles also becomes a good enough reason to look at Blender as an alternative. As stated here, a very good combo for a relatively cheap pipeline would be Zbrush + Blender + Houdini + Fusion…
The original post:
Blender is great, and free. It get’s greater all the time and stays free forever. It’s the best character animation app out there, after maya and softimage. It has a free renderman renderer to use, besides its already cool Cycles renderer. (Also vray etc)
For mindblowing VFX and general animation wizardry, there’s big H. With Houdini, you pretty much have a pipeline out of the box. Contrary to what people say, it’s is easy and cool to learn. You just have to think like a compositor who works in a node based environment. It has a free Aprentice version, a very affordable indy version and a good old perpetiual licence model.
As for modelling and texturing… I don’t know a serious modeller who doesn’t use Zbrush. But If you’re stuck in boxmodelling, there are more then enough alternatives out there, with Blender being my favorite. For texturing, you can use Zbrush along with Substance Painter/Designer for example. Modo and Cinema4D look very capable too, but I never used them.
For editors, colorists, mograph artists and compositors, it’s a no brainer. There are very capable free versions of DaVinci and Fusion from Blackmagic. And inexpensive fullfledged studio versions. All with a perpetual licence.
But the big showstopper is still interoperability. Mainly Alembic. I don’t doubt Alembic will be part of Blender in the future, and I don’t doubt it will be properly implemented. Just, as many others have said before, I hope it gets the attention it deserves to have it for the 2.8 release.
Having listened to this, I’m still not sure I understand the motivation behind using a foreign data format (flexible as it may be) to store cached computations that end up staying in the same application, after all. The Gooseberry artists didn’t exactly make it sound like an invaluable feature, either.
I would carefully suggest Alembic users to try and contact Sergey to describe how they use it in practice, in order to mitigate his concerns regarding interoperability. He mentions Collada as problematic, but Alembic was designed specifically to not belike Collada. It does not specify much beyond the very basics practically every 3D application employs. It’s then up to the user to “fill in the gaps” (e.g. to define Materials for the renderer), but that’s fine (i.e. “by design”).
I have my doubts about whether there really is such a big demand for this, but maybe others are less skeptical, so let me throw out this idea for a business:
Suppose there is a decent amount of professionals who use Blender as little more than a Cycles frontend. Alembic support would have to go through Blender to Cycles. Instead:
Cut out the middleman by creating an application based on Cycles that only imports Alembic geometry and allows for basic scene composition and material editing (like Octane, for example).
Sell this graphical frontend for money (Cycles uses Apache license, not GPL!), but also provide a backend that can be freely used on an unlimited amount of render nodes (this will be your key competitive advantage).
Factor out a part of your application into a standalone benchmark and try to get it adopted by major publications for some cheap publicity.
Move your business to New Zealand to pursue a romantic relationship that ultimately ends up fruitless (this step is optional).
Finally, add 360° panoramic rendering to your product and “VR” to your company name, then get bought out by Facebook for 5 billion dollars. Proceed to move into an obscenely expensive house in San Francisco. The house must come with an array of candy dispensers (this part is non-negotiable).
Important: If you implement this business idea, you owe me 19,95$/month + 5% revenue.
Thoughts?
That’s actually a really good business plan. People are already doing it even. There is a plugin for Rhino and the latest Poser is using Cycles as well. The first person to create a robust Maya -> cycles pipeline will indeed be able to peruse a romantic relationship in New Zealand and then later a home in San Francisco.
If anyone thinks this is already possible through FBX or this Alembic. . . needs to try it for themselves. It basically isn’t. It is way way better than before when we didn’t have FBX but in my case it still takes months of re-working materials, texture maps, geometry problems etc etc. Cycles is the only thing that brought me to blender. Cycles stand alone with a robust API that could allow lower level programmers to “hook in” with their studio’s already established pipeline would be a major boon for Cycles adaption. But from my understanding, this is already being considered and implemented. The future is bright.
There are already wealth of render engines(including standalone) that can read alembic. By itself the format is useless, it needs to operate within another DCC app. For instance if you need to do a tsunami scene - you first create 3D scene with full setup from modeling to rendering > export proxy data for example to Reafllow or Houdini > simulate -> import alembic cache > maybe add some secondary FX(eg splashes) to it using the data > set up materials and render.
That is why Houdini(Tech powerhouse) and Maya(artist friendly env) usually work hand in hand. Same software harmony could work with Blender.
Like julperado said, there are some pretty insecure users right now in camp of other software for variety of reasons (like a certain company trimming 10% of work force). It’s all over the internet. People WANT to have at least 1 foot in boat of Blender but many studios list lack of Alembic as major Con that prevents them. Blender has absolutely everything else to be the nr1 choice. It is extremely productive, full featured application with powerful GPU rendering as well as external production proven render engines like prman and vray.
Coming from max I can also tell you that there was no real alternative to using Realflow for fluids (well now Phoenix). It’s frustrating also with them as they havent updated their SPH code for years. 30 minutes of usage will result in mem leaks and slowdown (unless cmd). After RF I tested same production solutions in Blender and was blown away how stable, fast and accurate result was! Perhaps not as universal(all the daemons of RF are exceptional) but make no mistake, there are plenty of scenarios where Blenders physics and simulations would seem attractive to users of other softwares.
Without plugin API, if studio cannot deliver certain solution that might not be natively there or is a weak area - you might aswell close doors. This is why alembic is vital for production. Blender though open, is the last major 3d software not to support this standard. It’s a shame.
This really sums up my thoughts on the matter. I can’t help but think that Alembic would be THE killer feature that would suddenly get Blender into pretty much all pipelines, big and small, in some capacity. It would be a true Trojan Horse feature that would get a lot more users relying on Blender, even if only to gain access to Cycles. More pro users would then lead to more pro developers as studios would want to further develop Cycles and Blender in general. Would be a win/win.
After keenly following the Gooseberry development, mainly to get a glimpse into the Alembic development, it was very disappointing to then see it abandoned. I did try some of the dev builds, but could never get it to work. Although I realise it was created as a “hacky” production solution, wouldn’t having a temp solution now (purely for .abc import) be better than none at all?
There might not be a big demand for this from the Blender community, but it is essential if Blender is ever to be adopted by the broader industry for anything other than small jobs on the side. (same goes for USD, but that’s another topic, and apparently isn’t even on the radar right now - hope I’m wrong.)
Unfortunately it gets little attention. One of the big reasons for this is that neither Ton nor any of the other core developers really understand how alembic is used in professional production pipelines.
Alembic (and other caching systems) are used throughout the production process, NOT just at render time.
They are used to bake animation performances for further effects work, to improve viewport performance in massive scenes, to allow for fine tuned corrections free of rigging constraints, the list goes on.
Lukas unfairly gets a lot of blame for how the alembic code went in Gooseberry, but in truth he was misbriefed and mismanaged. The implementation he was asked to work toward was way too limited and missed the point of how caching is used.
Blender would really benefit from proper caching support, and the current options don’t fit the requirements of big production. Unfortunately this situation is unlikely to change any time soon.
The thing about Houdini is that there are a lot of different ways to accomplish things. It’s more of a development environment than a traditional 3d workflow. It often feels more like programming - but in a very artist-friendly way. There are these generic presets of course, but the real power in in it’s VOP/SOP nodes - which are essentially visual programming environments. At the same time though Houdini auto-routes a lot of stuff so that you don’t have to do all the tedious stuff.
It’s not unusual to for me to spend a great deal of time using not even looking at the viewport.
One point people seem to forget is that Alembic is good only for people who want to make movies. For gamedevs and archvizards out there I don’t see it having that much use. That’s my understanding.
Not disagreeing that it would probably make Blender more popular and bring in more talented, professionally oriented people.
Also regular users. I watched a couple of vids and it seems useful for movies, but I can’t find a use for gamedevs outside maybe baking large physics simulations.
Ambi is right, alembic isn’t really doing much for the game dev or arch vis scene at the moment. So being devil’s advocate one could say why take away development time from those areas to implement a change that doesn’t affect a large part of blender’s current userbase.
I think what we need to consider though, is whether we want to be developing blender to appease a current user base or to encourage new users to join. I think the latter is much healthier in terms of actual progress. Ideally, both happen and one encourages the other but since blender is funded by its current users it makes decisions like this incredibly hard to make when allocating the little development resources that are available, I’m sure.
For gamers and archviz obj and fbx covers most of their needs, for sure.
But if you freelance and get to do part of work for some studios, chances are that you will get your scene, of scanned environment, or tracked data, using Alembic or FBX. As a matter a fact nowdays in 90% cases you will get data in Alembic. There are many freelancers that would use Blender, specifically Cycles but can not as they are not ‘compatible’ with pipelines in most studios.
And these type of freelancers could easily pull Blender into larger pipelines from which we all can benefit at the end, that is why I think Alembic is essential for Blender.
+1000 !
And more experienced users is always good for Blender, more users, more donations, more donations, more developpement.
It’s a WIN/WIN !
I don’t know for gamedevs, but for Archviz Alembic would be also a great help!
Being smaller in size it would make the handling of huge scenes much more easier, and for exterior or city shots, where you also have the terrain, trees, cars, people, etc… having the extra help of something like Houdini (for the procedural creation and placing of assets) would make the job a lot easier.
My $0.02 - I’m convinced the reason you’re not seeing more in the way of notable short / indie films that include Blender as part of their pipeline is precisely BECAUSE of the interoperability woes, which a familiar caching format would begin to resolve.
I’m not working in film at the moment; I freelance, which keeps me all over the map and moving from one app to another. But having worked with Maya’s Alembic implementation to some degree, and seeing the ease with which I could get my scenes and deforming characters into lookdev and compositing applications, etc., without the muss and fuss of broken rigs / animation / parenting hierarchies, I tend to look on Blender’s lagging behind here as a major turn-off. And I’m just one contractor, with relatively simple requirements - imagine the frustration a larger team of people would experience, trying to accomplish more advanced ends using multiple applications. If I have to opt not to use Blender for my work, certainly a larger team would as well.
Edit: I should clarify - this is assuming a team of indies / professionals that would like to incorporate Blender because of its generally delightful character animation interface (this is another debate entirely), and not a group of teenage Cycles fanboys. 
What do you think would be the best way to contact and open a dialogue about this with Sergey and Ton that wouldn’t come off as more “feature begging”? 
Agreed, you kind of summed up the second half of my post, we should be attracting these people through developing such features. I was simply presenting another perspective.
Like a lot of people in this thread, my desire for alembic is from first hand interoperability issues. My work (minus simulation) is really quite simple most of the time, I do all my modeling in blende, the vray - blender integration isn’t bad and cycles is great for stills or if you don’t have a strict deadline. I’d love to keep the entire pipeline primarily within blender and nuke. I can’t however, use alembic to import sims so I stick with max for almost everything just to save the hassle of jumping between programs due to ‘loyalty’.
My point more simply is trying to look at it from the perspective of assigning budget. Developing something that a current minority is calling for, regardless of whether it attracts more studios to adopt blender into their pipeline in the future, can be seen as a difficult decision to make when presented with feature requests from a more vocal and numerous part of the community. A leap of faith that hopefully someone will take.
Since it is “feature begging”, it is pretty difficult to do that
+1000 !
And more experienced users is always good for Blender, more users, more donations, more donations, more developpement.
It’s a WIN/WIN ![/QUOTE]
This.
It about sums it up.
+1