Studio migrating from nuke to fusion

Moderator: SecondMan

conomara
Posts: 4
Joined: Thu Jan 10, 2019 2:00 am

Studio migrating from nuke to fusion

#1

Post by conomara » Thu Jan 10, 2019 2:17 am

Hello,
We are a small but growing studio. We are seriously thinking about migrating to Fusion. Now seems like a good time to consider this since we are still small.
The obvious advantage of licence fee wrt to nuke is great. There are other really nice features as well. We have our compositors doing a comparison now, however they have a known bias since they are already nuke orientated.
We are windows now, but I want to move to Linux in the future.
We would be using Fusion as a tool for slapcomps and comp.
We would not be using it for editing or anything else etc.
We would be using it on farm nodes as well.
We have diff. departments eg lookdev, anim etc, but we may be merging our lighting comp depts, since we
are an animation company (as opposed to VFX), while retaining pure comp leads/supe and developers as well.
This is basically the model I've worked with on feature in large companies. However at the moment lighting and comp are
separate.


We have virtually no pipeline built for our nuke people anyhow, so we don't loose much yet if we switch.
Hello,
I am interested to know from people who have a degree of experience with both packages, what your experience
of the pros and cons are?
I guess areas of interest are:

general comp functionality
3D functionality
Renderer
scripting
third party play nice ability-Deadline etc
How good is Package management in Fusion...
Support from community and BlackMagic (this site seems great, love the tagline)
Rate of development compared to Nuke, is BM focused on it?
Big one.......how hard is it to find compositors for Fusion, at all levels
Learning curve for new compositors and formerly nuke compositors.
General feeling about its future of Fusion and any other notes.
Where is a good place to look for artists?
Does Fusion play nice with Linux
Does Fusion play nice with audio
How well is OCIO integrated, general colorspace management compared to nuke?
....anything else

Big thanks in advance, if any of you can even answer one of the questions

Conor
Last edited by conomara on Thu Jan 10, 2019 3:38 am, edited 4 times in total.

User avatar
Kristof
Fusionista
Posts: 504
Joined: Thu Aug 07, 2014 1:30 pm
Been thanked: 10 times

Re: Studio migrating from nuke to fusion

#2

Post by Kristof » Thu Jan 10, 2019 2:49 am

Hi Conor,

Welcome to the Fu community. A bit more background info would help. What kind of OS are you using? What is your typical workflow like? I mean, are you guys pushing around heavy comps? Deep compositing? Do you guys have different departments ( editing, modeling, shading, rigging, animation, lighting, compositing, grading ) or do your artists do "a bit of everything"?

One thing I can say is that Fusion's (standalone's) future is very uncertain. This is a very delicate subject, but I suggest you take a look at the official BMD forum and make up your own mind. Compare the activity in the Fusion section of the forum vs the Resolve one (especially the level of engagement of official staff). The lack of response and all that jazz. Support is lacking for sure (am also hinting at how things like bug reports are organized for instance). But again, that depends on how far you're pushing things.

It is hard to find seasoned Fusion users too, but I have a feeling (so that's very subjective) that you'll find more Fu users in Europe compared to the States. Might be wrong.

Plugins are limited, but you'll have the basics covered ( DoF, de-noising etc).

Integrating Fusion in a pipeline is not on the same level as Nuke for instance. Again, that would depend on what you're automating. You can do a lot, but most TDs will favor Nuke as it plays nice and conforms more to industry standards. Well, Nuke is the industry standard tool for compositing after all.

But hey, it's only $299 or something like that. So that's not a bad deal, even considering all the bugs and limitations.

conomara
Posts: 4
Joined: Thu Jan 10, 2019 2:00 am

Re: Studio migrating from nuke to fusion

#3

Post by conomara » Thu Jan 10, 2019 3:33 am

Thanks alot Kristof for the advice, I will digest it, there's alot of info and suggestions for how to further my own investigations.
Thank you!

User avatar
SirEdric
Fusionator
Posts: 1741
Joined: Tue Aug 05, 2014 10:04 am
Answers: 1
Real name: Eric Westphal
Been thanked: 77 times
Contact:

Re: Studio migrating from nuke to fusion

#4

Post by SirEdric » Thu Jan 10, 2019 6:19 am

Kristof wrote:
Thu Jan 10, 2019 2:49 am
It is hard to find seasoned Fusion users too
There are certainly a few quite seasoned Trainers available though, that could help you migrating,
transferring your pipeline, getting your team up to speed, hitting the ground running...:-)

User avatar
ShadowMaker SdR
Fusionista
Posts: 606
Joined: Sun Sep 21, 2014 6:17 am
Answers: 2
Been thanked: 13 times

Re: Studio migrating from nuke to fusion

#5

Post by ShadowMaker SdR » Thu Jan 10, 2019 6:21 am

Yes, I could even provide a link to one of them, since he's too modest to do it himself. :-)

User avatar
SirEdric
Fusionator
Posts: 1741
Joined: Tue Aug 05, 2014 10:04 am
Answers: 1
Real name: Eric Westphal
Been thanked: 77 times
Contact:

Re: Studio migrating from nuke to fusion

#6

Post by SirEdric » Thu Jan 10, 2019 6:23 am

ShadowMaker SdR wrote:
Thu Jan 10, 2019 6:21 am
since he's too modest to do it himself
Me? Modest? Only a tiny little bit....:-)

User avatar
ShadowMaker SdR
Fusionista
Posts: 606
Joined: Sun Sep 21, 2014 6:17 am
Answers: 2
Been thanked: 13 times

Re: Studio migrating from nuke to fusion

#7

Post by ShadowMaker SdR » Thu Jan 10, 2019 6:29 am

You're the most modest person I know. You've told me so dozens of times ;-)

@conomara

Check out:

http://fusiontrainer.de/

User avatar
Midgardsormr
Fusionator
Posts: 1002
Joined: Wed Nov 26, 2014 8:04 pm
Answers: 1
Location: Los Angeles, CA, USA
Been thanked: 47 times
Contact:

Re: Studio migrating from nuke to fusion

#8

Post by Midgardsormr » Thu Jan 10, 2019 8:33 am

I'll hit those that I can say something about and ignore the ones where I don't know enough to give an intelligent answer.

conomara wrote:
Thu Jan 10, 2019 2:17 am
general comp functionality

For most tasks, Fusion is just as capable as Nuke, regardless of the level of complexity. 99% of compositing is doing the math, and that doesn't change no matter what software you're using. Fusion's newer tracking systems—3d camera tracker and planar tracker—are not yet production-ready, it lacks Deep, and it doesn't do the arbitrary channel handling that Nuke does. Deep may or may not be coming as a 3rd party plug-in (someone's been working on it, but given the uncertainty about Fusion's future, I'm not sure if he's still progressing or waiting to see which way things are moving). Handling for multi-part EXR and other channel-related stuff is supported, but the tools available currently are limited—you might have to build some of your own if that's important to your workflow. There are some Explode EXR scripts that will clone a Loader in order to give you access to all the channels in an EXR, but the Flow can get a little more complex since you don't have the ability to send all those channels down the same pipe. On the other hand, when I used Nuke I always used to break that stuff into its own visible flows, anyway, just for the sake of clarity in the graph.

3D functionality
Renderer

I never got very deep into Nuke's 3d system, but I'm very fond of Fusion's. It lacks any kind of ray-trace renderer, although it's technically possible to build a plug-in renderer. Someone did a proof of concept using, I think, 3Delight several years ago. It wasn't finished, but it showed that a 3rd-party renderer plug-in could be made. Fusion's ability to manipulate meshes at the point level is limited, and it has only a few primitives, although you can do some fancy extrusion stuff with the free Krokodove plug-in. It does support FBX and Alembic, though, and it can handle large point counts. Only spot lights can create shadows. I recommend looking at the work of Con-Fusion (aka Vito La Manna) and Max Seredkin to see what you can do with Fusion 3D in terms of lookdev and lighting.


scripting
third party play nice ability-Deadline etc

I never did any scripting in Nuke, so I don't know how it compares, but I do quite a bit of pipeline work with Lua in Fusion. I find scripting in Fusion to be flexible and powerful. We have automated comp creation, rendering, asset import, EDL reading and plate pulling, version control, pre-comp and matte generation, and many other things. I use Lua and the other pipeline guy uses Python. Both work just fine.

We use Deadline as our render manager and have little trouble. The most recent update gave us some problems because Deadline didn't wait long enough for Fusion to start up and report its status, so we had to add a wait command to the submission script. That took a couple of weeks to figure out since BMD and Thinkbox support did the usual bit where they each blamed the other instead of actually looking at the problem. But that's typical whenever you're trying to get one piece of software to talk to another, regardless of the vendor. We routinely pass data to Fusion from our Filemaker database, and we have a module to send information back the other way, but it's never been turned on, so I don't know how well it works.


Support from community and BlackMagic (this site seems great, love the tagline)
Rate of development compared to Nuke, is BM focused on it?

As others mentioned, development at the moment is uncertain. I personally remain optimistic, but Blackmagic keeps their cards close to their chest, so it's difficult to know what's going to happen. As for community support, it's small but enthusiastic. You'll find this site and the allied Con-Fusion Discord channel to be where most of it happens in the English-speaking world. The smaller community vs Nuke means there are fewer custom tools, but it also means that communication can be easier since you're more likely to find people who really know what they're talking about—I've seen a good deal of misinformation in Nuke-related conversations because there are so many voices, and an uncomfortably large percentage of them don't know how little they know.


Big one.......how hard is it to find compositors for Fusion, at all levels
Learning curve for new compositors and formerly nuke compositors.

Fusion compositors are thin on the ground. At Muse we generally don't specify that Fusion experience is necessary. It's easier to retrain a Nuke user than to find an experienced Fusioneer. Depending on the artist, retraining takes anywhere from two to eight weeks. We usually start a new artist off on junior-level tasks, even if they're experienced, and help elevate them as quickly as we can to their previous level of Nuke proficiency. Since it takes some time for that retraining to pay off, you won't be able to hire as many people on short-term contracts. You can bring a Nuke compositor in for two weeks and then let them go without much trouble, but if it's going to take two weeks just to get someone up to speed, you'd better hang on to them for several months to make it worth the cost.


Does Fusion play nice with audio

Not really, no. Audio is the biggest pain point for me in Fusion. It's pretty well the only thing that makes me go back to After Effects occasionally.

conomara
Posts: 4
Joined: Thu Jan 10, 2019 2:00 am

Re: Studio migrating from nuke to fusion

#9

Post by conomara » Mon Jan 14, 2019 3:49 am

Amazing info Midgardsormr and other respondents.
I appreciate you taking the time to write the reply.
Very substantive and gives us alot to think about.
Thanks
Conor

User avatar
SirEdric
Fusionator
Posts: 1741
Joined: Tue Aug 05, 2014 10:04 am
Answers: 1
Real name: Eric Westphal
Been thanked: 77 times
Contact:

Re: Studio migrating from nuke to fusion

#10

Post by SirEdric » Mon Jan 14, 2019 11:03 am

Midgardsormr wrote:
Thu Jan 10, 2019 8:33 am
Depending on the artist, retraining takes anywhere from two to eight weeks.
Au Contraire!
I converted Nukers in less than a week, AFXlers within a week or two.
It's not *that* bad when you do training on the actual job...:-)

Added in 2 minutes 17 seconds:
Well...yes...to be honest there also were those guys that went like "Yeah. <yawn> Cool setup. Show us another one."
But those are the guys you only want to convert to MS Office, and nothing else...:-)

User avatar
Kristof
Fusionista
Posts: 504
Joined: Thu Aug 07, 2014 1:30 pm
Been thanked: 10 times

Re: Studio migrating from nuke to fusion

#11

Post by Kristof » Mon Jan 14, 2019 1:55 pm

SirEdric wrote:
Thu Jan 10, 2019 6:19 am
Kristof wrote:
Thu Jan 10, 2019 2:49 am
It is hard to find seasoned Fusion users too
There are certainly a few quite seasoned Trainers available though, that could help you migrating,
transferring your pipeline, getting your team up to speed, hitting the ground running...:-)

Hey @SirEdric, yeah, you're one of those legends. :) And props to @Midgardsormr for taking the time to write such a detailed reply. Definitely agree on the math bit and such. Its mainly HR complaining about how hard it is to find Fusion people, but I always find that to be less of an issue. I've helped several people getting to know Fu, coming from Nuke and as long as they keep an open mind and are willing to give it a go, it works out okay.

User avatar
Chad
Fusionator
Posts: 1381
Joined: Fri Aug 08, 2014 1:11 pm
Been thanked: 8 times

Re: Studio migrating from nuke to fusion

#12

Post by Chad » Tue Jan 15, 2019 7:36 am

Kristof wrote:
Mon Jan 14, 2019 1:55 pm
as long as they keep an open mind and are willing to give it a go, it works out okay.
You're paying them right? :lol:

I mean, sure it makes headhunting difficult, but for normal hiring you won't see much of an issue because people who apply for jobs, generally speaking, want jobs.

Nebukadhezer
Posts: 5
Joined: Wed Jan 09, 2019 4:50 am
Been thanked: 1 time

Re: Studio migrating from nuke to fusion

#13

Post by Nebukadhezer » Wed Jan 16, 2019 2:18 pm

Just before I write down some thoughts about this topic, I used Fusion version 5-6 for some years in animated feature film projects and normal film projects.
We do have some compers now that, since fusion studio is 300 bucks, want to have it in the arsenal. Currently I am looking into integrating it into our pipeline.

And this is something I would consider a downside of fusion.
To me developing for fusion feels a bit like developing for 3dsmax. You can make almost everything work, but most of the time it takes longer getting there as with nuke or maya.
But that being said I used fusion mainly around 2010.
Getting back at it now it feels it matured in some areas, but it still lacks a well python/qt integration. (2 days of looking into it)

And to us this is a big deal as we share the same codebase accross most dccs. meaning the same code runs in nuke, 3dsmax, maya, syntheyes, photoshop etc...
I can run the same code in fusion too, but the python interpreter seems to be shutdown, after every call. then custom menus and the python call from within lua seems to be a restricted python session, which probably works for a hell lot of helper scripts, but context aware integration of tools, becomes difficult.
Of course I can open a seperate instance of a qapplication and dump everything on there, but having fusion as a qtapplication now I would want it to be the "host".

Back in 2012 when I moved away from fusion to nuke, stereo was the main reason. I can only emphasise how much more mature channels and views are integrated in nuke then fusion.
And views are not only fantastic for stereo but as the view system in nuke is not limited to two eyes, one can use it for configurators and what not.

Then there is another thread about the best default setup for fusion, and there seem to be a lot of LUT longings in there.
I was a bit shocked when I opend fusion and it still defaults to 8 bit and has no lut loaded.
Linear workflow is sth that nuke truly pushed and it is a bit of pity that Fusion makes it hard to begin with
OCIO in fusion is also a bit lazy, default roles are not propagated so again more clicks to actually start working.
Then one thing I could not find out is the working colorspace, where all transforms are made.
Maybe someone can help here? I guess it is srgb/rec709 primaries?

Particles in nuke are pretty much unusable, fusions particle system is so much better there are no words, I still have to look into nuke11 to see if they caught up a bit.
Things I miss in nuke are generators daysky and plasma for example.
But nuke has blinkscripts we just started using them and they are fantastic. (and they render the same on gpu and cpu)

Things that happened to me with fusion where black frames on the farm mostly due to ram. then the opengl renderer which on different cards could/can produce different results.

I guess I am biased towards nuke, it feels more consistent by design, but the current pricing and the constant lack of new features or the performance overall in nuke are big downsides.

One tiny thing that drove me mad in fusion was that an image sequence is not pushed to where its frames belong (frame 1001 became 0 or 1 cannot remember).
(easy to fix with some code, still why?)

just my 2 cents

User avatar
SecondMan
Site Admin
Posts: 3366
Joined: Thu Jul 31, 2014 5:31 pm
Answers: 2
Location: Vancouver, Canada
Been thanked: 65 times
Contact:

Re: Studio migrating from nuke to fusion

#14

Post by SecondMan » Wed Jan 16, 2019 6:38 pm

Nebukadhezer wrote:
Wed Jan 16, 2019 2:18 pm
just my 2 cents

Which may be worth a hell of a lot more than 2 cents further down the line. Can I suggest making user stories of each of your points and sending them to BMD support? Detail each and every one of them as much as you can.

Let me verify with them again what the preferred channels are. But in short, BMD needs to hear from users what the needs are. And it sounds like you have quite a few specific ones that may benefit a lot of others in similar situations.

User avatar
SirEdric
Fusionator
Posts: 1741
Joined: Tue Aug 05, 2014 10:04 am
Answers: 1
Real name: Eric Westphal
Been thanked: 77 times
Contact:

Re: Studio migrating from nuke to fusion

#15

Post by SirEdric » Wed Jan 16, 2019 9:29 pm

Nebukadhezer wrote:
Wed Jan 16, 2019 2:18 pm
Then one thing I could not find out is the working colorspace, where all transforms are made.
Maybe someone can help here? I guess it is srgb/rec709 primaries?
Fusion's 'internal' colorspace is (of course) linear, as it should be...:-)

Added in 3 minutes 27 seconds:
Some of the other things you describe (like 8bit CD or Loader start) could be set in the preferences.