Magic Mask in Fusion?

I'm looking forward to trying the 17 Beta tonight, but I was looking though the release info and I can't seem to find out if the Magic Mask system in the Color page was also available in Fusion.

If it works half as well as they are showing, it would be a huge help for less precise rotoscope work in Fusion.

Does anyone know if it's in Fusion as well?

[Peter Chamberlain]

Color page only

I really hope it does come to fusion in the future. It would be a huge timesaver if it works as well as in the demo video.

You could export the mask as an alpha channel and use that in Fusion, until BM may port it.

Uli, could you point me to a summary of how to export an alpha channel to use in FUsion?

Page 863, "Using VFX Connect".

Thanks very much

Color page only

That's a bummer. Looks powerful. Hopefully that will be coming to Fusion later.

You could export the mask as an alpha channel and use that in Fusion, until BM may port it.

Thanks for the idea! Sounds like it could be a good workaround.

It somewhat defeats the purpose of having everything inside the same software if things can't be used in tandem with each other.

Like, just make a "Fusion input node" in the color page where you can output normal and alpha, and then have a "color input node" in the fusion page where the fusion output of the color page goes in.

How can that not even exist in the first place? It's such an obvious thing to exist. We have input in fusion, then output in fusion that goes into the color page. But what if we want to be able to compose together two clips where we use the color page as the grading for the different comp layers. Or use the much better tracking and masking tools in there. We literary have a software where we have these two pages not really working together, other than one comes before the other.

Make a fusion output node in the color page. Then a color page input node in the fusion page. Then have the ability to split the two in the color page so that you can use one grade or node structure that uses the input from the edit page and has an output into fusion, while being able to add another input that uses the output from the fusion page in order to grade the entire comp into an ouput.

I can't see how this would be hard to implement. Would be good if they even put that in before the main release of v17.

Well, while I admit that there should be better integration of masking or keys between those programs, for this specific feature it's not really needed.
It is far from what Roto Brush or similar functions, like those in Silhouette, can do. It'll just give you a very coarse mask, which you should soften considerably, and then you can use it for mild corrections of color/contrast/intensity in the Color page.

It's not a replacement for rotoscoping and serious masking (yet?).

Like, just make a "Fusion input node" in the color page where you can output normal and alpha, and then have a "color input node" in the fusion page where the fusion output of the color page goes in.

How can that not even exist in the first place? It's such an obvious thing to exist.We have input in fusion, then output in fusion that goes into the color page. But what if we want to be able to compose together two clips where we use the color page as the grading for the different comp layers. Or use the much better tracking and masking tools in there. We literary have a software where we have these two pages not really working together, other than one comes before the other.

Make a fusion output node in the color page. Then a color page input node in the fusion page. Then have the ability to split the two in the color page so that you can use one grade or node structure that uses the input from the edit page and has an output into fusion, while being able to add another input that uses the output from the fusion page in order to grade the entire comp into an ouput.

I can't see how this would be hard to implement. Would be good if they even put that in before the main release of v17.

I understand why you proposed that but it can't work. This is Resolve's processing pipeline.

1. Clip attributes are the attributes of the clip that are set in the Media Pool. This steps also converts all stills to short video clips. *
2. Fusion compositions import that media and possibly do more before sending it to the next step.
3. Input Sizing(?) scales the clip for the timeline.
4. Group Pre-Clip grade are applied to a group of clips specified in the Color page.
5. Edit FX and Sizing are the stacks of effects and transforms you assigned to a clip in the Edit page.
6. Clip grades applied to one specific clip.
7. Group Post-Clip is a grade applied to the same group of clips but after clip processing.
8. Clip blending determines how the current clip blends into the clip beneath it on the timeline.
9. A timeline grade gets applied to the entire timeline.
10. Output sizing does a last minute resize.

* You can and should bypass this step with a Loader when bringing stills into a composition. It gets you much better performance and save on memory.

That's difficult to figure out because Resolve isn't very good at conveying that pipeline in the software. Personally, I would remove the concept of all clips in a timeline being compositions and adopt a "Convert to Composition" approach that stores the composition in the Media Pool. Then I'd give Fusion access to all of the Color page's Color Correction tools. I'd also move the Edit FX and Sizing after the Group Post-Clip step and allow Fusion Effects to be dragged into node graphs on the Color page. That gets the same result as what you were suggesting and would prevent Edit effects from pre-empting clip and post-clip grades.

In order for your idea to work, a Fusion composition would either have to work like an effect that's applied to one clip or the Color page would have to manually route different clips to the same composition somehow. The latter would be the only way to allow the Color page to pre-process more than one clip before entering Fusion and, even then, any media added via Fusion would have to be color graded with Fusion's color correction nodes. It would ultimately become very difficult to keep track of and would be very error prone.

If you aren't concerned about performance with still images and you don't mind working with footage that's already been resized to the timeline's resolution, you can already pre-process with the Color page by encapsulating one or more clips into a Fusion Clip. That will put them into a kind of nested timeline that sends each track to a MediaIn in the Fusion Clip's composition. By viewing the Fusion Clip as a timeline, it will actually allow you to apply Edit FX and transforms, group grades and clip grades before they're sent Fusion and the Fusion Clip itself can still have a grade applied to it. I hate Fusion Clips and want them gone but, for now, they're there for you to use.

There is always the compound clip trick to use color page masks.
(add alpha output, connect blue line, edit in fusion with alpha).
I havent installed v17. Does the magic mask can be used that way in fusion?

I understand why you proposed that but it can't work. This is Resolve's processing pipeline.

That right there is the problem. Locking a system into a certain pipeline which limits the possibilities that should be logical workflow options and solutions.

This is the reason why Avid never gets out of innovation hell. Even when they reworked the code from the ground up, they managed to just update the UI, but kept all the tedious BS that keeps them behind Adobe and Blackmagic.

Hopefully BM will understand how powerful a "back and forth" workflow between the color page and fusion page will be. Magic mask is good for getting fast qualifiers, but they couldn't have created that without realizing how powerful it is as a rotoscoping tool. It's light years faster than the rotobrush in After Effects, but have no easy uses like it.

Hopefully BM will understand how powerful a "back and forth" workflow between the color page and fusion page will be. Magic mask is good for getting fast qualifiers, but they couldn't have created that without realizing how powerful it is as a rotoscoping tool. It's light years faster than the rotobrush in After Effects, but have no easy uses like it.

I'm don't really see the lack of ability to interweave the Color page and Fusion page as a problem. I'd prefer they get rid of the need to interweave between pages as much as possible.

For example, I don't see the purpose in routing the Color page to Fusion just to pre-process a clip or a mask. I think Fusion should have a superset of the nodes that are available to the Color page so that Fusion can do those tasks by itself. Then the purposes of the two pages would be this:

• Fusion would be for creating re-usable assets and effects with controls exposed for the editor and colorist.
• The Color page would be used to apply effects to clips, groups of clips, and timelines in the context of the edit. You can think of the timeline itself as a node graph full of merges and trimmed MediaIn nodes and the Color page is the page that modifies it.

I went into more detail about my idea in this topic.

I agree with you when you say...

That right there is the problem. Locking a system into a certain pipeline which limits the possibilities that should be logical workflow options and solutions.

...but when I think of that, I think about the Edit FX and how it relates to the Color page.

There isn't anything that the Edit FX stack can do that the Color page can't do better because of it's node graph. The problem is that Resolve can't just get rid of Edit FX and require all effects to be done with Color page. It's much easier to set keyframes for a blur effect within the Edit page so removing that ability would tick off a lot of people. The fact that the Edit FX precedes the Color page also makes it so that that keyframed blur effect would impede the colorists abilities to create a qualifier on that clip or track a power window. The Colorist could just jump to the Edit page, disable that blur, and then apply another at the end of the node graph but it wouldn't bring the keyframes with it.

If the order of the two were switched then that would be worlds better but only 99% ideal. I think a better option than that would be if the Edit FX just appended an OFX node to the clip's node graph. That way the Editor wouldn't mess up anything the Colorist needs to do and the Colorist could incorporate the effect into their node graph without undoing what the Editor did.

I bought Fusion Studio, it's a pity there is no similar feature. I hope that Fusion will soon have some automatic rotoscoping.

I got the idea to try Planar tracking for Rotoscoping. Looks like it might work

I've experimented, in some situations it's inaccurate and confusing. But at least something. I think BlackMagic could use this tool for tracking rotoscoping. Maybe you just need to modify the algorithm.

What I love most about Resolve is the software philosophy. I don't know if is intentional or not, but is very similar to the autodesk flame family. But one thing I would really like to see is something that Autodesk itself has done with flame in version 2012. Tear down the walls of the "modules" that divide the software. In a way, that a clip is a node tree and vice versa. Example, transform on the edit page is the same transform node in Fusion and is the same on the Color page. It's a rather confusing concept to explain without showing. But that would make the software infinitely more fluid and I believe that it would even make the development less complicated, in terms of bug fixes. For there would be no 3 transformations, magic masks, color corrections, etc., but the exact tool that would permeate different user interfaces.

As I cannot post URL´s I suggest a quick search on youtube for
Autodesk Flame 20th Anniversary Event, where autodesk demonstrates this concept.

Sorry for the english, google translator here.

Magic Mask is pretty young and I suppose BM will further develop it. Until then I suggest Mocha for commercial work.