DIfferent types of Fusion nodes? (Resolve)

Hey there,

I was starting with my first BlackMagic product (Resolve) after having used nodes in other software (Mainly Houdini /Toon Boom Harmony).

And i was wondering which different kinds of nodes there are in the Fusion Tab and what their differences is in working with them. For example, there are nodes with this symbol (attachment). Or what the color coding of the nodes mean.

I do not understand yet why some nodes work different then others. For example: when i put a Paint node in a single plain merge, i can't paint on it. But when i put a shape into a plain single merge node, it does show up.

I am having a hard time trying to understand the structure of how nodes are and i can't find a Resolve 15 manual, so help is really appreciated.

Thanks in advance!

The manual is under the Help menu in Resolve, Help>Davinci Resolve Reference Manual

Or this: https://www.blackmagicdesign.com/produc ... e/training (scroll for Fusion).

Fusion has some specific workflows that need to be learned. Mostly, when and where does "this" node work. When I started on fusion back in the Eyeon days, that was what threw me for a loop.

In your example of putting a paint node onto a merge vs putting a shape onto a merge, it sounds like you need to understand what the circle is doing, the paint is supposed to do and the merge is used for.

Paint would typically be used in a straight line from a source node. (not that it has to be second in line, but it should be in that line) Think of it in terms of what you want to do. (I have a source, and want to resize it (add the resize node) I want to color it (add the color node) I want to paint out a pimple (add the paint node)...all in a straight line.

Merge is used to Merge two different node trees together. It has a bg and a fg input, as well as a mask input. It sounds like you are trying to put a paint node into the fg input, without anything feeding the paint node itself? Something has to feed that paint node, before it can plug into the fg of the merge node (or the bg) Also, you can send the output of the merge node into a paint node. In that case, everything that is done before the merge node, becomes the input to the paint node, which you should then be able to paint on.

The shape you were adding was probably defaulting to the mask input on the merge node, which is one possibility of it's design...but again you cant plug a paint into a shape(at least i don't think you can) as it serves no purpose. You can't paint on a shape, but you can paint on a source that you are also masking with a shape.

The more you play, the more your will figure out the operations and how they can work.
Sometimes it helps to learn nodes, buy simply thinking of them as layers on a timeline at first.
As you figure them out, then you start to see them more as the powerful workflow they can become.

I too have a bit of a hard time understanding some logic here. For example, if a node is a 2D operator, meaning it takes (usually) raster input(s) and outputs a raster image, how come that you can't plug a paint node into merge (why doesn't paint node simply default to project format if it has no input) or why can't you paint on result of shapenode ? Isn't shape output rasterized before it is processed in following nodes? And if it isn't does it mean that it is outputting vector data directly? When is its result rasterized in this case?

Other than for After Effects, a lot of functions in Fusion are resolution independent or store single actions.
So, even for 2D, you need rasterizers in many cases.

Other than for After Effects, a lot of functions in Fusion are resolution independent or store single actions.
So, even for 2D, you need rasterizers in many cases.

And how does it have something to do with paint not working on shape output? Every 2D node rasterizes its output anyway, unless downstream nodes can actually request vector shape data (like 3D ops do), which I doubt is happening. Anyway merge, paint etc request raster data, so even if they do allow vector output, merge node doesn't ask for it, unless I completely misunderstand the logic of Fusion engine.

Well, Paint is fully vector-based, editable, undoable and animate-able. How much computing power would you like to throw at it to be able to work smoothly? Even in AE you can switch the rasterizer on and off.

Paint op is vector based in the sense that each stroke is stored as vector object, but it takes raster data as input, does it not? What has computing power got anything to do with it anyway? I'm talking about what is technically going on (what goes in, what is processed and how, what goes out), not whether it happens smoothly or not on someones desk. You have to rasterize vector data at one point anyway, there is no way around it unless we get vector displays.

Makes sense. But i don't know about Fusion that much so i will stay silent on that..

Another thing: is it possible to merge more than 2 things in one merge node? For example: in Toon Boom you have one 'composite' node and the most right input is the foremost and the most left the thing farest in the back. Compositing 2 things at a time feels a bit odd to me, but i'm sure it has a reason.

One merge per item in 2d. If you are working with 3d tools then multiple items can merge in one 3d merge operation.

Unless you install Reactor for Resolve which contains - if you choose to install it - the MultiMerge tool which allows for up to 64 inputs merged on top of each other.