What does DEEP in Nuke that we can NOT do in Flame?

This is a very serious question, as I’m starting to understand deep has some issues, it is not perfect, and we can replicate it in Flame, but it can’t be that easy, I am sure I am missing some features I don’t get or maybe my pals are not the most regular uses of deep. But what can DEEP do that Flame can’t?


Place something dynamically within CG, not just behind or in front of it.

1 Like

Yeah, but you can do that in flame with the alembics, the camera, and projecting the render, takes 2 min, so you have volume and you can easily add something in between playing with Zs. Deep is only that? Nothing more?. There must be something: Ease of use, or anything related to the Depth of Field, perhaps. Don’t know. Or how it sticks to the z position automatically, something you can not do in flame, but working with 3d viewers should make it easier.

There is an entire class on fxphd where Victor Perez goes into that.

From my understanding the difference is when you get rendered files that have z-level pixel information, which is different from doing this in a 3D model/render environment. Different assets and workflows.

Like fog volumes as an example.

Deep is awesome for DOF and deep holdouts are also great.

No idea how you would do that with CG alone.

lets say you have cg birds circeling a clocktower, and you want to be able to remove the birds behind the tower not in front, you could either use 2 seperate renders for fg and bg or just use deep and roto cg renders in depth because you can just remove stuff thats behind other stuff.

I guess there is always a workaround with other mattes abd whatever but really deep is very easy and solid.

its just like rendering the whole scene into slices, takes forever to render though and is heavy to work with, so its not … for everythibg


Deeps are very important for Big VFX on both Feature and Episodic. So much, that the CG departments includes them by default as part of their renders. It can reduce the amount of time you take to comp a shot by a big percentage. For big shots is a given that is not worth even trying without deeps.
Deeps also help on simple cases like comping volumes, particle fx and fur in between CG renders and 2d elements. There many examples.

Lately Deep Utility passes are been used more. Deep Normal, Position and RestP give better results that regular RGB as deep handles overlapping transparent samples better.

I have been meaning put out a little example of how to creates hold outs for Flame… but havent found the time yet. In case you CG artist says… but I gave you deeps for that smoke!


This is it !!!

And there’s no way Autodesk implement this into Flame?

Here are my two cents:
At my job, I use both Flame and Nuke. For CG comp integration, I use Nuke. The R&D team where I work has developed a z-depth node using deep, which yields results infinitely better and easier to use than anything I’ve tried in Flame. Additionally, Nuke has a native PGBokeh node that utilizes deep and is very good. I’ve given up on using Flame (or been forced to!) for integrating CG, not only because of the platform’s advantages in this environment but also because of the possibilities that deep offers. It’s extremely easy and fasts to create object links in 3D space using deep.


Thanks for your answer. I clearly have to go way deeper with it.


Jake Gyllenhaal Reaction GIF

1 Like