Flame 2024 is here! ❤️

@TimD Could you give us some examples of these ‘blue sky’ features you would like to see. Or at least in which area of the software you would expect them to be.

2 Likes

Using a new graphics API is HUGE and really important so I definitely applaud the time being spent on that, especially with the depreciation of OpenGL. This is something that simply needed to happen. Hopefully stability will improve.

Can I ask though, had the steppy playback on Linux been fixed?

2 Likes

Surprise us! We like surprises.

Like, just one small example. I don’t think anyone expected Action Gmasks or requested that. But it was a surprising “new feature” which now I’d say a lot of us wouldn’t want to live without. It wasn’t a fix, it was a new idea.

1 Like

Noticed that price seems to have gone up alongside. It’s now $4,870/yr.

1 Like

It’s a monthly payment more than many nice cars.

Amazing Release !! Love the Native Apple silicon support / Paint improvement and other features, that make our life better !! Thank Autodesk Team for your hard work :slight_smile:

3 Likes

I’ll take that challenge…

Batchion

  • NeRF implementation in batch/action
  • Real-time path tracing in action

Timeline

  • AI capable tools for matching–transforms, specific frame to offline, etc.
  • Node based editorial tools for timeline versioning (requires a conversation)

Animation

  • Node based animation tools for expressions/particles

…just to name a few that would get heart-rates up. I’ve tried to stay away from features already available in Nuke or Resolve but there are quite a few that I would love, staring with full Multi-Channel support. Again… super grateful for 2024 but since you put it out in the world. Nothing like a bunch of sweaty, bothered Flame artists.

11 Likes

Some Blue Sky Features…

An AI/ML particle system, including text to gen, with control sliders.

An incredible update to the text module.

A native AI texture generator

A native interface to hook into/ interface with the likes of Runway.ml

An AI assisted snap to object function in GMask tracer to help with roto. Let AI do the heavy lifting but then have the ability to make it human editable/tweak-able afterwards.

Considering the amount of Freelance Flame Artists with their own kit out there, some incredibly dreamy way of getting shots to and from them in the easiest and most user friendly way.

Media IO: IAB audio support, Native IMF export and supplemental creation, proper audio metadata tagging

As someone else said, a much improved multi channel exr interface.

6 Likes

hey Adam , could you please elaborate on what depreciation of OpenGL means? or if it’s not much trouble post a link to learn about it.
thanks

In layman’s terms, OpenGL is a language for the computer/CPU to talk to the GPU and explain what it wants/needs the GPU to do. Open GL has been around a long time and there are newer languages that can talk to a modern GPU and better utilise its modern features. Two of the languages (aka APIs) are Metal (MacOS specific) and Vulkan. Because it is a more modern language, the communication between CPU, Memory & GPU is more efficient so therefore faster and more stable.

Because Open GL is an older language, Apple as a company decided that whilst they would include it as part of MacOS, they weren’t going to update, optimise or test it any longer to see if it works with the OS. So it is slower and less stable.

A Google search will find a whole lot of articles on this.

3 Likes

Having experimented with that recently - I wonder from an effort perspective, if it makes sense to fully build out the particle system in Flame, or if it should just be easier to import particles into a Flame comp from an external app? At some point you have to pick your battles with available dev time. Particles are useful to some, but not everyone using Flame, and it’s a massive and evolving subject matter when you go deep on it.

Cine4D and Insydium X-Particles are incredibly powerful. But at present there doesn’t seem to be any scalable way of bringing a model into Flame. That may be an easier problem to solve, at least for advanced uses of particles beyond what is presently possible in Flame. And there are other particle systems out there, this is just one I’ve worked with.

5 Likes

I would definitely second this. The 3d environment within action is an insanely powerful tool that is insanely painful to navigate. Improving the general performance and UI/viewports of the 3d environment in action would be amazing.

Also I would love a revisit of the semantic keyer. It has some of the best potential of any tool released in years, but the implementation is mostly unusable because of the edge-of-frame issue and the fact that it is painfully embedded in the image node and requires a matchbox to even get the matte out. A standalone semantic keyer which could take a roughly tracked gmask tracer as a region of interest and had even moderately improved temporal coherence would be useful in just about every comp.

3 Likes

I dream of this every night.

7 Likes

Yeah. I’d of rather see improvements to the exporting Action scenes so they’d easily load up in Blender, Houdini, Maya et al. Then you can do particles properly, but still retain whatever images, geo, and camera tracks you’ve built in flame, ensuring the renders will line up.

Making the software more round-trip friendly would open up a lot of avenues that are currently just a little too cumbersome. For example, if Neat Video required you to export and import plates it would get way less use, but because it’s so easy to move into and out of the program we all love it.

More of that.

5 Likes

I falsely thought it would affect linux boxes. Is there a hope that sometime Flame will use Vulkan API? What would it mean for Flame? AMD gpus support? or any other gpu for that matter?
Thanks a lot for your time.

Hey Fred,
All great suggestions coming in from Logik. For me, particle generation is always a thing… the time required to get anything looking good out of action particles is way too costly/risky to take on. Let’s have great smoke/fire sims with manipulators/forces which are easy to use, even using geometry as emitters.
Or a ‘warp stabiliser’ like premiere, no motion vectors to think about, just a simple node with similar controls.
KeenTools Facebuilder/tracker in nuke has always been interesting in being able to use a mesh to track & unwrap face uv’s.
To be honest, it doesn’t have to be anything that big, just a small token of fresh, bleeding edge tech. We now have more playtime time cos silicon just got quicker!
Don’t get me wrong though, I realise it’s not always that easy & i really do appreciate the stability/workflow advances you guys are doing. Really, thank you.

1 Like

I think this was discussed in some detail during a LogikLive in the Fall when the Flame Devs were on…

If I remember right it was this episode: Logik Live #93: Summer of Flame 2022 with Autodesk’s Jasmin Frenette — Logik.tv

3 Likes

Not just EXR, Flame’s whole pipeline need to be multi-channel, something Nuke got right from the start.

I really hope that whatever replaces OpenGL improves render quality massively and does away with archaic stuff like Action’s manual z-buffer nonsense. Maybe even, dare a boy dream, some GI features.

2 Likes

Now I’m almost sorry I brought this up. Yikes.

1 Like