Welcome to WSL!

Make yourself at home, but before posting, please may I ask you to read the following topics.


Posting 101
Server space, screenshots, and you

Thank you!

PS. please pretty please:


Image

Anyone successfully using Blender-generated Cryptomattes in Fusion?

Moderator: SecondMan

caspardavidfriedrich
Fusioneer
Posts: 52
Joined: Thu Dec 01, 2016 3:19 am
Location: Bracknell, UK

Anyone successfully using Blender-generated Cryptomattes in Fusion?

#1

Post by caspardavidfriedrich » Wed Mar 20, 2019 2:49 pm

I'm in Blender 2.8 and rendering out OpenEXR files with Cryptomatte enabled for objects and materials.
When I open the files in Fusion I can see OpenEXR channels called CryptoMaterial00.B for example. Most have no data but some seem like mattes (weirdly they all have the same ID like 65535). Must I put them in the ObjectID and MaterialID channels? Either way, Cryptomatte fails. Either I'm doing something wrong, or Blender is generating non-standard Cryptomatte data.
Has anyone successfully used Blender Cryptomatte exports?
Thanks,
Konrad

User avatar
ShadowMaker SdR
Fusionista
Posts: 866
Joined: Sun Sep 21, 2014 6:17 am
Answers: 7
Been thanked: 32 times

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#2

Post by ShadowMaker SdR » Wed Mar 20, 2019 3:02 pm

Did you add the Cryptomatte tools from Reactor to Fusion? I think they're required to make it work inside of Fusion.

User avatar
pingking
Fusionista
Posts: 744
Joined: Thu Aug 14, 2014 9:10 am
Been thanked: 1 time

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#3

Post by pingking » Wed Mar 20, 2019 11:45 pm

In blender 2.79 blendet didn't wrote the "cryptomatte manifest" into the exr files and cryptomatte failed with the error that it didn't found that.
I guess it is still the same in 2.8 but I didn't try for a long time

Edit: according to the blender docs,manifest is now in the Metadata

User avatar
SteffenD
Posts: 9
Joined: Wed Jun 28, 2017 3:55 am

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#4

Post by SteffenD » Thu Mar 21, 2019 7:40 am

Two further things to consider when dealing with Cryptomatte in Cycles:

- Render 32-bit full float. 16-bit half will give messed up results
- Don't use a lossy compression like e.g. DWAA because it'll also garble up the Cryptomatte info

User avatar
pingking
Fusionista
Posts: 744
Joined: Thu Aug 14, 2014 9:10 am
Been thanked: 1 time

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#5

Post by pingking » Thu Mar 21, 2019 1:38 pm

i tried to render a simple image with 2.8 and i did get the manifest into the exr.

but the cryptomatte node failed, it even failed with the official test images from the cryptomatte repository.

so i need to check if my installation of the cryptomatte fuse is correct (installing from reactor seems not to work)

caspardavidfriedrich
Fusioneer
Posts: 52
Joined: Thu Dec 01, 2016 3:19 am
Location: Bracknell, UK

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#6

Post by caspardavidfriedrich » Thu Mar 21, 2019 2:34 pm

Thanks for all the pointers guys! Yep, I did install Cryptomatte via Reactor and I'm using Fusion standalone. I'll take the 32bit float on board next time I render. Yes, @pingking, I've got a feeling in my bones that the implementaion of Cryptomatte in Blender somehow isn't playing nice.

Added in 14 minutes 59 seconds:
Right I've rendered an OpenEXR file at 32bit float and connected the Loader to the Cryptomatte node and I just get a stream of errors messages like below...
Anyone know what that means? Thanks.

...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: table index is NaN
stack traceback:
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: in function 'parse_manifest'
...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:461: in function <...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:449>
Cryptomatte1 failed at time 29182
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: table index is NaN
stack traceback:
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: in function 'parse_manifest'
...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:461: in function <...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:449>
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: table index is NaN
stack traceback:
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: in function 'parse_manifest'
...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:461: in function <...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:449>
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: table index is NaN
stack traceback:
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: in function 'parse_manifest'
...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:461: in function <...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:449>
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: table index is NaN
stack traceback:
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: in function 'parse_manifest'
...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:461: in function <...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:449>
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: table index is NaN
stack traceback:
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: in function 'parse_manifest'
...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:461: in function <...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:449>
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: table index is NaN
stack traceback:
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: in function 'parse_manifest'
...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:461: in function <...esign/Fusion/Reactor/Deploy/Fuses/Matte/cryptomatte.fuse:449>
...ion/Reactor/Deploy/Modules/Lua/cryptomatte_utilities.lua:379: table index is NaN

Added in 3 minutes 30 seconds:
pingking wrote:
Thu Mar 21, 2019 1:38 pm
tried to render a simple image with 2.8 and i did get the manifest into the exr.
Looking at the error messages I just posted, from what I can make of them, it seems like Cryptomatte can't parse the Blender-rendered cryptomatte manifest.

User avatar
Solluna108
Posts: 2
Joined: Sat Nov 25, 2017 11:00 am

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#7

Post by Solluna108 » Sun Mar 24, 2019 11:37 am

It works for me, just tried it. Cryptomatte set in Blender 2.8 with Material selected and Rendered 5 Frames to test in EXR Multilayer. In Fusion I use the Cryptomatte plugin form the reactor. Every Material is recognized correct.

User avatar
pingking
Fusionista
Posts: 744
Joined: Thu Aug 14, 2014 9:10 am
Been thanked: 1 time

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#8

Post by pingking » Sun Mar 24, 2019 1:01 pm

i cant make the cryptomatte fuse work correctly
tried to install from reactor, made sure that i have no older version installed, but nothing seems to work

User avatar
Solluna108
Posts: 2
Joined: Sat Nov 25, 2017 11:00 am

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#9

Post by Solluna108 » Mon Mar 25, 2019 1:02 am

Now i was curious. I tried a few weeks ago Cryptomatte with Blender and Fusion for the first time. My first testing was fine. After rendering my 700 frames animation it didn't work anymore in Fusion. It works for my only if I activate Material in Blender Cryptomatte. Activating Material and Object I have the same problem like you.
Seems more like a Blender issue. But to be honest I am not a professional so i cannot say if I use Cryptomatte the right way.

User avatar
Lester
Posts: 7
Joined: Fri Aug 30, 2019 4:25 pm

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#10

Post by Lester » Fri Oct 25, 2019 3:58 pm

I having the same issue here. The render from blender has 3 cryptomatte output layers, each with RGBA info. I can get it to work partially by assigning one of the layer's RGBA to the Color RGBA channels on the loader node, but one layer only contains parts of the mattes. Also, there is only one input on the node, but it looks like it at one point had multiple inputs as shown in this video of the developer showing it of in fusion. Maybe it is like that in the Fusion standalone, as I am using Resolve fusion. Anyone able to verify this with fusion standalone?

User avatar
Kristof
Fusionista
Posts: 760
Joined: Thu Aug 07, 2014 1:30 pm
Answers: 2
Been thanked: 15 times

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#11

Post by Kristof » Mon Oct 28, 2019 2:29 am

Hi Lester, that's an old video and it no longer works like that. The Fu developers helped us out and expanded fusion with an EXRIO module, giving us the ability to directly access the multiple layers and ranks in a Crypto exr. You're not supposed to shuffle the channels in your loader. I don't use Resolve, so I can't help you with that. My foucus is on Fu standalone.

What version of Cryptomatte are you using in Fusion? Try the one from Reactor (it is not the latest version, but it's easier to install). You have to use full float exr images and multilayered ones (containing all the ranks). And you can have a multilayered exr per layer ( asset, object, material) or all contained in one.

Do not save directly from a frame buffer or preview window, as these won't work in most cases. Anyway, go through this thread to see whatever applies to your problem: viewtopic.php?f=6&t=1027&start=225

And if you can, post a crypto render in that post and I'll have a look.

User avatar
gez
Fusioneer
Posts: 168
Joined: Mon Jul 16, 2018 6:21 pm
Answers: 1
Location: Argentina
Been thanked: 3 times
Contact:

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#12

Post by gez » Tue Oct 29, 2019 5:03 pm

An idea that might help:
Apparently Blender is applying destructive compression when used (DWAA/B, for instance) to all the passes in a multilayer render, not just the RGBA ones.
It is possible that compression doesn't play well with Cryptomatte.
As far as I know they are aware of this and it is planned to fix that, preventing losy compression from breaking technical passes.
(I'm not sure whether this is related to the problems cited above, but that's something I would check. Also make sure that render is full-float).

User avatar
jxxjxx
Posts: 25
Joined: Wed Jul 05, 2017 3:31 am
Been thanked: 1 time

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#13

Post by jxxjxx » Fri Nov 01, 2019 6:12 am

HELLO,I find a new tool can resolve this problem. i use the pick_cryptomatte tool read Blender 2.8 EXR files,as the trial,very fast and simple extract all ID.


https://youtu.be/xu7_YG9H__s?list=PLgnE ... BBPed4MDP5

User avatar
jxxjxx
Posts: 25
Joined: Wed Jul 05, 2017 3:31 am
Been thanked: 1 time

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#14

Post by jxxjxx » Fri Nov 01, 2019 7:35 am

Image

yes.
Faster and more steady - Pick_Cryptomatte v1.1 for Fusion

User avatar
Jack John
Posts: 9
Joined: Fri Nov 01, 2019 7:26 am
Real name: Jack

Re: Anyone successfully using Blender-generated Cryptomattes in Fusion?

#15

Post by Jack John » Fri Nov 01, 2019 7:40 am

I sad so long, I have resolve Blender EXR cryptomatte problem.
Yeah,PICK Cryptomatte v1.1 for Fu do best.
Image
Last edited by Jack John on Fri Nov 01, 2019 6:54 pm, edited 2 times in total.