I’m doing my compositing in Nuke, so I enable lots of render passes, like diffcol, diffdir, difind, AO, cryptomatte, location etc.
I’m also rendering out different render layers.
In older projects, I plugged all these passes into an output node in Blender, where I manually had to rename all the passes, which took a lot of time.
This time I figured out, that when using the normal composite node, Blender saves all the channels into the multiEXR file. This does work nearly perfectly, except it seam to forget the alpha socket of each render layer.
I need this, to composite all the layers together. Especially when using diffInd etc, which don’t come with an own alpha channel.
In Blender, there is an alpha channel in the Image node, and even a seperate alpha output right underneath. Bot for some reason this does not get included into the EXR Multilayer file.
How can I get the alpha channel included into the multiEXR file, without having to plug everything manually into the output node, having to rename everything by hand?
I wanted a comped version of all the render layers, so that I can compare it in nuke to get the same result.
But I disabled the compositing nodes, but still same result.
Somewhere there is an alpha channel, which Blender can read, but Nuke can’t find it.
It appears I need to connect all the output manually and rename them in an output node.
Please raise this issue on a Nuke forum, providing them with the simplest possible Blender file which illustrates the issue. Then, come back here and tell us what they said. It’s not at all clear why Nuke would fail to identify a particular layer …
The alpha channel lives in the Composite_Combined and ViewLayer_Combined passes. It doesn’t show up in Nuke’s alpha breakout for the same reason the beauty doesn’t show up in the RGBA by default (because Blender doesn’t put it there. An annoyance many have been complaining about for ages).
But it’s right there, ready to be copied into other streams as needed.
G’day mate, so I assume it’s a similar issue that I ran into when I tried to export into Davinci Resolve Fusion compositor, I found there is no alpha channel that blender exports. to re-create the composite you have to use blender’s compositing structure
, then edit things in between those nodes, That was my experience anyway with Fusion, then you use all the other passes & cryptomattes to adjust in between this layer structure.
Edit: I export my image sequence as OpenEXR Multilayer
Here’s an example of my node tree in Fusion combining all the layer passes from Multilayer EXR (right image) to re-create the Rendered composition ‘combined passes’ from blender (Left Image)
Nah it doesn’t have an Alpha that it exports with MultiLayer on Fusion, you can only use the alpha calculation from the ChannelBoolean / Blend mode as in other softwares.
BUT if you’re wondering how you can combine the final scene in compositor, you don’t need an Alpha channel for it.
…what the hell are you talking about?
Of course Blender can output an alpha channel where appropriate. Do you think view layers and the transparency rendering settings are decorative? Or perhaps you don’t know that such parameters exist?
I understand your screenshot.
I use Fusion as well, but I don’t use the method of extracting the channels, I use the manual method. In this method, just choosing the Alpha layer/channel that is right That q I don’t remember in the nuke, if you can do it manually
The ‘alpha channel’ isn’t a pass that you have a socket for, it is a Blend Mode that all compositing software use when you set the blend on ‘screen’ etc…
Transparency rendering makes the render transparent until you put it into a compositing software where ‘transparency’ is just black space that you use a blend setting to get rid of.
and how does this change the fact that compositing software like Fusion & Nuke interpret ‘alpha’ as Black, not as a ‘Pass’ that you can edit separately without affecting the rest of the image that may have a bit of black colors in it.
I’m trying to explain how the software functions based on my experience to solve OP’s problem, if you can show where the alpha pass is located & how to work with it, please do so.
There isn’t an “alpha pass” because that would be silly. Why would you have an entire layer for a single channel which is only relevant to an existing layer? Blender’s compositor may create a ‘socket’ for the alpha but that’s for convenience. It’s not a discrete layer.
Alpha information is stored as a component of the ‘Combined’ pass because that’s the only place it’s relevant. Sometimes it’s called “RGBA” or “Beauty”. Most of a compositing process is addition and multiplication of layers which shouldn’t be using the alpha channel at all.
Using a channel boolean/shuffle node to split out channels is compositing 101. If alpha weren’t a component of the RGBA pass slap comps would be a pain. Also you should be alpha dividing to protect edges anyway.
