In light of the Coronavirus crisis, please help make sure nothing stands in the way of social distancing in the VFX industry.

This petition asks the MPAA to act accordingly:


Sign the petition on Change.org

Be safe, everyone.

Psyop's Cryptomatte In Fusion

User avatar
SecondMan
Site Admin
Posts: 4229
Joined: Thu Jul 31, 2014 5:31 pm
Answers: 15
Location: Vancouver, Canada
Been thanked: 58 times
Contact:

Re: Psyop's Cryptomatte In Fusion

#256

Post by SecondMan » Tue Feb 11, 2020 11:29 am

bowserlm wrote:
Tue Feb 11, 2020 11:02 am
Why don't we just have the latest one on there?
Obviously because no one submitted it.

Tags:

User avatar
bowserlm
Fusionista
Posts: 344
Joined: Sat Apr 28, 2018 3:44 pm

Re: Psyop's Cryptomatte In Fusion

#257

Post by bowserlm » Tue Feb 11, 2020 11:49 am

Consider it submitted! How does one do that?

User avatar
SecondMan
Site Admin
Posts: 4229
Joined: Thu Jul 31, 2014 5:31 pm
Answers: 15
Location: Vancouver, Canada
Been thanked: 58 times
Contact:

Re: Psyop's Cryptomatte In Fusion

#258

Post by SecondMan » Tue Feb 11, 2020 11:59 am

Read the docs, look at the examples in the Submissions forum - there's an example submission as well. Finally, use Atomizer to at the very least check your Atom, if not build it.

As you pointed out, there's a version of Cryptomatte in Reactor already so you can use it as a template and adjust it where necessary.

User avatar
Kristof
Fusionista
Posts: 708
Joined: Thu Aug 07, 2014 1:30 pm
Answers: 1
Been thanked: 8 times

Re: Psyop's Cryptomatte In Fusion

#259

Post by Kristof » Tue Feb 11, 2020 1:31 pm

It is known that Crypto glitches in Fu9 and we're still trying to get to the bottom of this problem. I may have something concrete (it is a simple fuse suffering from similar behavior) but this setup works fine in Fu16 (same fuse, same node graph, different host). As I don't use Fu16 in production myself, I can't really replicate the crypto glitches. So I would like to ask other users if they experience glitching when using several cryptomattes in Fu16 standalone. Could somebody please verify this? Thanks in advance!

User avatar
Cedric
Fusioneer
Posts: 51
Joined: Tue Sep 13, 2016 7:26 am
Answers: 1
Location: Ghent

Re: Psyop's Cryptomatte In Fusion

#260

Post by Cedric » Fri Feb 14, 2020 5:07 am

Hi @bowserlm
bowserlm wrote:
Tue Feb 11, 2020 11:02 am
Perhaps I don't know how to correctly do the Path Map stuff...

The UserData one doesn't seem to be editable, and if I create my own and call it UserData as soon as I close the pref pane and open it again it's gone.

https://share.getcloudapp.com/RBud5l0b
The UserData: you select is the System one. As you see there are three groups of path maps.
System, Defaults and User. I recommend you, for the ease of use, to edit the Defaults UserData path map.
So from the screenshot you sent, this is what it should look like.
  1. UserData:;AllData:;Fusion:;Reactor:Delploy;/SOMEDIRECTORY/Cryptomatte/fusion
Also another way, imho a cleaner way, to set up these path maps is by creating User path maps and mentioning them inside the Defaults.
This is what my personal setup looks like at the moment.

Image
bowserlm wrote:
Tue Feb 11, 2020 11:02 am
Added in 14 minutes 31 seconds:
Also, is there a reason we're several versions behind on the Cryptomatte version on Reactor? Why don't we just have the latest one on there?
As @SecondMan mentioned, these things live separately from one another, so as long as no one reports it, as stated, nothing will happen.
Considering I do not personally use Reactor either, I am not really aware of the versions on there.

Cheers
Cedric

User avatar
Kristof
Fusionista
Posts: 708
Joined: Thu Aug 07, 2014 1:30 pm
Answers: 1
Been thanked: 8 times

Re: Psyop's Cryptomatte In Fusion

#261

Post by Kristof » Wed Feb 19, 2020 7:15 am

Kristof wrote:
Tue Feb 11, 2020 1:31 pm
It is known that Crypto glitches in Fu9 and we're still trying to get to the bottom of this problem. I may have something concrete (it is a simple fuse suffering from similar behavior) but this setup works fine in Fu16 (same fuse, same node graph, different host). As I don't use Fu16 in production myself, I can't really replicate the crypto glitches. So I would like to ask other users if they experience glitching when using several cryptomattes in Fu16 standalone. Could somebody please verify this? Thanks in advance!
A big, big thanks to @Droedel for testing this for me. Tests so far do indicate that Crypto in Fu16 is glitch free and that the bug was squashed (a bug in Fusion and not in the Crypto code). I am pretty confident my assumptions are correct but if @Droedel has new info to share that contradicts this, I'll share it in here. Fingers crossed!

User avatar
lucassfmoura
Posts: 4
Joined: Mon Dec 23, 2019 6:57 am

Re: Psyop's Cryptomatte In Fusion

#262

Post by lucassfmoura » Tue Mar 03, 2020 6:17 am

Hello everyone. Continuing a post about an issue with Clarisse iFX and Fusion Cryptomatte that said:

"Hello everyone.
I'm using some renders coming from Clarisse iFX in multi layers EXR with a cryptomatte pass included but can't seem to get the cryptomatte working properly.
I'm using the latest version from their Github and installed following their instructions. Tried using the test files and they are working properly. Asked the 3D artist to render me another scene with a simple cube and sphere and everything works too. When rendering the huge scene I'm comping with the same settings cryptomatte gets a error with this message from the console:

Cryptomatte1 failed at time 273
...agic Design\Fusion\Modules\Lua\cryptomatte_utilities.lua:880: table index is NaN
stack traceback:
...agic Design\Fusion\Modules\Lua\cryptomatte_utilities.lua:880: in function 'parse_manifest'
...agic Design\Fusion\Modules\Lua\cryptomatte_utilities.lua:732: in function 'initialize'
...agic Design\Fusion\Modules\Lua\cryptomatte_utilities.lua:912: in function 'create_cryptomatte_info'
...ta/Blackmagic Design/Fusion/Fuses/Matte/cryptomatte.fuse:583: in function <...ta/Blackmagic Design/Fusion/Fuses/Matte/cryptomatte.fuse:568>

Any clue on what might be wrong?

Thanks in advance :D"

In response to Kristof. Unfortunatelly I can't share the files since they are images from a feature film and I could get in trouble. I tried opening the same exact file in Nuke NC with the most recent cryptomatte files from their github and it worked just fine there.

User avatar
Kristof
Fusionista
Posts: 708
Joined: Thu Aug 07, 2014 1:30 pm
Answers: 1
Been thanked: 8 times

Re: Psyop's Cryptomatte In Fusion

#263

Post by Kristof » Tue Mar 03, 2020 9:17 am

Hey @lucassfmoura, try previewing a render in the Fusion buffer and enable the subview (v), then RMB in that subview and enable metadata. Do you see Crypto related metadata like in this example?

Image

User avatar
lucassfmoura
Posts: 4
Joined: Mon Dec 23, 2019 6:57 am

Re: Psyop's Cryptomatte In Fusion

#264

Post by lucassfmoura » Tue Mar 03, 2020 10:55 am

Hello Kristof.

I did that and apparently it's all there it's just that cryptomatte isn't getting them for some reason. When I opened it in Nuke it detected thousands of objects in there, could it be that the file is just too huge and Fusion probably can't handle it correctly?

Anotação 2020-03-03 184822.png
You do not have the required permissions to view the files attached to this post.

User avatar
Kristof
Fusionista
Posts: 708
Joined: Thu Aug 07, 2014 1:30 pm
Answers: 1
Been thanked: 8 times

Re: Psyop's Cryptomatte In Fusion

#265

Post by Kristof » Wed Mar 04, 2020 12:16 am

Thanks, @lucassfmoura. The keys in the manifest look different but I reached out to Cédric to chime in (he's the code guru).
Meanwhile, could a lighting artist shoot a render with primitives only, omitting assets used in production? That wouldn't violate the NDA. A single frame is enough and that would speed up the process of getting to the bottom of this.

User avatar
Kristof
Fusionista
Posts: 708
Joined: Thu Aug 07, 2014 1:30 pm
Answers: 1
Been thanked: 8 times

Re: Psyop's Cryptomatte In Fusion

#266

Post by Kristof » Wed Mar 04, 2020 1:40 am

I found some Clarisse renders elsewhere (discussing issues with Clarisse Crypto) and the one with the manifest does work in Fusion. The preview is impractical but that doesn't break the workflow. So ignore what I said above about the keys. Wanting to support Crypto with a missing manifest (which is not the case in your example) is new to me and I wonder why one would want to. Facilitate a workflow where you only want to pick to establish a matte, I guess. Will look into that.

So myeah, having a render coming from you would be needed I'm afraid.

User avatar
lucassfmoura
Posts: 4
Joined: Mon Dec 23, 2019 6:57 am

Re: Psyop's Cryptomatte In Fusion

#267

Post by lucassfmoura » Wed Mar 04, 2020 2:10 am

Hello Kristof. I sent you some files via PM hope you get them. It's just wierd since we did the test with the primitives with the same method to render out cryptomatte and the primitive one worked and the other don't.

User avatar
Kristof
Fusionista
Posts: 708
Joined: Thu Aug 07, 2014 1:30 pm
Answers: 1
Been thanked: 8 times

Re: Psyop's Cryptomatte In Fusion

#268

Post by Kristof » Wed Mar 04, 2020 2:35 am

Thanks for that! :) You did mention thousands of objects and it explains the sluggish response in Fusion when I try to load that file. Even without trying to decrypt it.
Am using Fu9 and we had issues with this which the devs fixed in the latest? 9 (so perhaps this is even better in 16, don't know). It still has an impact (which I can see), but at least it doesn't break. I can understand of course the need to be able to support a huge manifest, so I would advise you to look at the sidecar option. That would improve the experience. Nuke doesn't suffer from this.

I'm not saying this is the issue either, because I did work on a feature film with a humongous manifest embedded. And it worked, but loading times became slow.

Will keep you posted.

User avatar
lucassfmoura
Posts: 4
Joined: Mon Dec 23, 2019 6:57 am

Re: Psyop's Cryptomatte In Fusion

#269

Post by lucassfmoura » Wed Mar 04, 2020 3:25 am

We tried rendering the cryptomatte in a single file (instead of a multi-layer EXR) with both AOV and manifest but no luck either. You're probably right. The file is just huge for Fusion to process.
Sadly we can't find any option to get the sidecars in clarisse, we searched and found out the Katana actually has that option but no info on clarisse.