Jump to: Board index » General » Fusion

Fusion 18 feature requests

Learn about 3D compositing, animation, broadcast design and VFX workflows.
  • Author
  • Message
Offline

Alaz Soytemiz

  • Posts: 129
  • Joined: Tue Nov 11, 2014 8:26 am

Fusion 18 feature requests

PostThu Dec 10, 2020 2:31 pm

Please request something that you need or you think that would help you here.

+ create mesh from dense pointcloud
+ vector tracking
+ AI based human roto

Without workarounds.


Tapatalk kullanarak iPhone aracılığıyla gönderildi
Windows 10 Pro 64Bit
Processor: Intel i7-6800K 3.40GHz
Ram: 128GB
Graphics Card: NVIDIA GeForce GTX 1070
Disks: Samsung SSD 970 PRO, Intel® SSD 750 Series (400GB, 1/2 Height PCIe 3.0, 20nm, MLC)
Offline
User avatar

Bryan Ray

  • Posts: 2080
  • Joined: Mon Nov 28, 2016 5:32 am
  • Location: Los Angeles, CA, USA

Re: Fusion 18 feature requests

PostThu Dec 10, 2020 7:02 pm

Updated Fuse and scripting API documentation.

Overhaul of the 3d system, preferably built on Universal Scene Description (USD) and supporting Hydra.

Support for arbitrary channel data, passed through all tools (Nuke-like).

Bring back flexible and customizable layouts.

An option for Text buttons on toolbars.

In-viewer buttons for Normalize, display DoD/RoI, Show Controls, Gain/Gamma, Show Checker Underlay.

Ability to maximize a pane (F4 in Fusion 9 and earlier).

Full 8-bit node tile colors.

Better whitespace management in the UI.

Improved hotkey manager.

Configurable hotkeys for the functions currently on the arrow keys—navigating the flow by arrow, nudging points in the viewers, and stepping through the timeline. Let us configure the controls the way we like instead of moving them around on us with every release.

Customizable 3d navigation controls. Default control scheme should preserve the current behavior, but it would be nice to be able to change it to conform to other tools an artist might use more frequently.

Improved VFX Connect workflow with Resolve, with support for versioning and user-configured paths.

Improved support for arbitrary data handling in the Flow—see Krokodove, Spicy Acorn (https://github.com/Spicy-Acorn).

CameraTracker improvements: Supervised tracking—maybe make it possible to add point tracks from Fusion's standard Tracker node. Focal length estimation by drawing perspective lines.
Bryan Ray
http://www.bryanray.name
http://www.musevfx.com
Offline
User avatar

Chad Capeland

  • Posts: 2742
  • Joined: Mon Nov 10, 2014 9:40 pm

Re: Fusion 18 feature requests

PostThu Dec 10, 2020 8:55 pm

v18:
  • Black box all Color nodes, including ResolveFX. Allow VFX Connect to export full pipeline to Fusion.

v19:
  • Interactive Color nodes including ResolveFX.
Last edited by Chad Capeland on Fri Dec 11, 2020 7:09 pm, edited 1 time in total.
Chad Capeland
Indicated, LLC
www.floweffects.com
Offline

UserNoah

  • Posts: 336
  • Joined: Fri Oct 04, 2019 3:32 pm
  • Location: Berlin
  • Real Name: Noah Hähnel

Re: Fusion 18 feature requests

PostFri Dec 11, 2020 12:21 am

While I was disappointed with the number of updates that Fusion, especially Fusion Studio, saw with 17, the ones we got seemed overall beneficial to me, so I'm hoping for further good updates to Fusion as well.

I can fully get behind everything that has been said before! I'm simply adding things that haven't been mentioned yet and that I remember from the top of my head.
They are not ordered by importance to me, besides the red highlighted which are requests I believe are very reasonable and close to my heart :D . I also understand that some of these aren't and shouldn't be high priority like many of the things that Bryan said, but I don't think one should "hold back" on a feature request thread :lol:

UI:

Bigger and more visible render indicator.

Let me hide the toolbar (generally bring back UI modifications)

Make keyframes easier to see in the Playbar and Keyframes Pane.

Give modifiers a visible selection state. So you can select and delete several at once.

Make modifiers available as nodes and support connecting more data types in the flow. (Being able to connect a number data type to an input slider control for example, through the flow)


Updates to the Particles which would also tie into updates to the 3D workspace in general:

Bug fixes for issues that have been around for some time, now. For example, behavior change even though the same seed is used or the rgndist function in the pCustom tool.

An out of the box particle turbulence that doesn't copy the velocity to the Particles and thus allows for gradual changes of direction.

Render time style generation so particles stay efficient points until rendered by the Renderer3D.
Updated point rendering of particles, like most modern render engines. Very often you only need particles to be spheres/dots with varying sizes and colors, yet the current methods take surprisingly long to render and make the 3D Viewport sluggish. I assume this is because it's currently not instancing but a bitmap for each particle (?)

Bring back support for Particle Fuses.

Make branching available in particles (although that would probably require a complete change to the current way of simulating them in Fusion, likely completely out of scope)

Proper instancing + multithreaded replicate tool.

As mentioned before, updates to the 3D workspace. Preferably built on USD.
I don't need Fusion to have an inbuilt path tracer if the OpenGL renderer would receive updates. Shadows are not available on both light types and soft shadows are not in the OpenGL mode for how long now?
USD could lead to many path tracers in Fusion.

Navigation in 3D could use a pivot on mouse pointer mode instead of center of the focused object only.


General ideas:

RAM usage while final rendering. There is no reason to keep 20 previous frames of the last 30 nodes in the RAM if there are no time modifying tools in the flow. I want to give Fusion all the RAM while working interactively, but it would be nice to be able to work in other software while Fusion renders. Most of the time the CPU and GPU usage would allow that, but the RAM is filled to the Max. (Without manually limiting it before rendering).

and/or:

Increase hardware utilization for faster render speeds.

OpenCL CPU fallback. I'm unsure if this is possible with Metal but it would be great to have a fallback for tools like the new OpticalFlow to be computed through CPU OpenCL (I only know that this exists in Houdini for Vellum sims). There is currently no automatic switch to the Legacy CPU version of this tool when the tool fails to initialize on the GPU. Which happens to me quite often.

"Compiled Macros/Groups". Looking at Houdini, you can use Compile nodes that encompass a set of nodes, limiting the copying between CPU, RAM, and GPU. This makes certain operations much faster as a block of nodes acts as if it was a single node. Something like this might yield speed gains for Fusion, too.

Support for the ResolveFX in Fusion Studio. There are a few tools that Fusion Studio desperately needs, like a good denoiser. Also, the two products should be able to exchange compositions without issues.

Updated imports for Alembic, FBX, and SVG. They don't support current standards/versions.

Smart vectors type of tool. Playing around with the magic mask in Resolve, it seems like it's almost there already. The dots and lines you draw track very well with a face, for example, only deformation is missing. I assume this also works through the same optical flow vectors that Fusion uses now?

Interactive tracking for 2D tracks. (Native) Single frame advances in the tracker would be great, but even greater would be an interactive mode that when activated, tracks to the time you advance the playhead, or advances frame by frame if you use the keyboard to go 1 frame at a time. Results in less clicking throughout the UI and more tracking.

Please separate viewer control movements and timeline movements again, this shouldn't be context-sensitive. There is so much clicking required in F17.

Network renderer/Render manager:

Network renderer could use more error messages.

Force render continuation instead of disabling everything. One missing frame is better than hundreds.
Offline
User avatar

BrianHanke

  • Posts: 52
  • Joined: Fri Jul 06, 2018 3:38 am
  • Real Name: Brian Hanke

Re: Fusion 18 feature requests

PostFri Dec 11, 2020 5:32 am

Honestly, for me a lot of it comes down to interface. Fusion 9 was VERY customizable, but ever since they put Fusion into Resolve it has gone with a more "locked-down" interface. I've been doing research recently to develop a workflow for some upcoming projects and I'm going to go with Nuke for compositing. You can make Nuke's interface into anything you want, making your specific tasks as streamlined as possible.

Aside from all that, I'd love to see a better ZDefocus-type tool in Fusion. Both Depth Blur and Defocus are pretty bad.

The Fusion tab interoperability is also a little buggy, making it hard to smoothly switch back and forth between editing and compositing. Nuke Studio gets this right, more or less. If I did go for Fusion for a project I'd definitely want to do all my work in standalone and then bring the renders into Resolve.
youtube.com/brianhanke
twitter.com/brianhanke
instagram.com/brianrhanke
Offline
User avatar

Chad Capeland

  • Posts: 2742
  • Joined: Mon Nov 10, 2014 9:40 pm

Re: Fusion 18 feature requests

PostFri Dec 11, 2020 7:22 pm

UserNoah wrote:Render time style generation so particles stay efficient points until rendered by the Renderer3D.


Particles don't change any of their properties when the style changes. If you're concerned about disabling caching, that can be done with a pChangeStyle.

UserNoah wrote:Updated point rendering of particles, like most modern render engines. Very often you only need particles to be spheres/dots with varying sizes and colors, yet the current methods take surprisingly long to render and make the 3D Viewport sluggish. I assume this is because it's currently not instancing but a bitmap for each particle (?)


UserNoah wrote:Proper instancing + multithreaded replicate tool.


Replicate3D does do instancing. The issue is what the downstream tools do.


UserNoah wrote:RAM usage while final rendering. There is no reason to keep 20 previous frames of the last 30 nodes in the RAM if there are no time modifying tools in the flow.


How would Fusion know that? At the time that the current frame is processed, Fusion doesn't know if other tools might require other frames. In Fusion, every tool is potentially a "time modifying tool" because SimpleExpressions, Modifiers, and InTool scripts can request other times independent of the tool using them.

A pre-render that maps out memory usage might be possible, like rendering out a low resolution variant and determining the cache patterns.

UserNoah wrote:I want to give Fusion all the RAM while working interactively, but it would be nice to be able to work in other software while Fusion renders. Most of the time the CPU and GPU usage would allow that, but the RAM is filled to the Max. (Without manually limiting it before rendering).


Would be a nice parameter to set, though. The render dialog could have an option to limit memory.

UserNoah wrote:OpenCL CPU fallback... There is currently no automatic switch to the Legacy CPU version of this tool when the tool fails to initialize on the GPU. Which happens to me quite often.


From the user perspective, that's a bug.
Chad Capeland
Indicated, LLC
www.floweffects.com
Offline

UserNoah

  • Posts: 336
  • Joined: Fri Oct 04, 2019 3:32 pm
  • Location: Berlin
  • Real Name: Noah Hähnel

Re: Fusion 18 feature requests

PostFri Dec 11, 2020 7:44 pm

Chad Capeland wrote:
UserNoah wrote:Render time style generation so particles stay efficient points until rendered by the Renderer3D.


Particles don't change any of their properties when the style changes. If you're concerned about disabling caching, that can be done with a pChangeStyle.



I might've been unclear. I want the option to display the particles as the default points as long as they are in the 3D workspace. So from to the pRender to the last 3D node before the Renderer3D. I want that the particles get their "Style" at the moment the Renderer3D creates a 2D image. If this would be coupled with a more efficient point/dot rendering it could make the whole workflow faster. Displaying hundred of thousands of particles as points is no issues but adding any kind of style to it will destroy any interactivity that you have.

Chad Capeland wrote:
UserNoah wrote:Proper instancing + multithreaded replicate tool.


Replicate3D does do instancing. The issue is what the downstream tools do.


Interesting, I never was able to use it with particles because it would increase memory too much and because it was single threaded it was simply too slow.

Chad Capeland wrote:How would Fusion know that? At the time that the current frame is processed, Fusion doesn't know if other tools might require other frames. In Fusion, every tool is potentially a "time modifying tool" because SimpleExpressions, Modifiers, and InTool scripts can request other times independent of the tool using them.

A pre-render that maps out memory usage might be possible, like rendering out a low resolution variant and determining the cache patterns.



This was pretty much what I envisioned but I lack the actual technical knowledge and didn't want to make assumptions. This is definitely nothing easy and I honestly don't expect this to happen. A simple slider to reduce Rendertime RAM usage would be a great start.
Offline
User avatar

Chad Capeland

  • Posts: 2742
  • Joined: Mon Nov 10, 2014 9:40 pm

Re: Fusion 18 feature requests

PostFri Dec 11, 2020 7:50 pm

UserNoah wrote:This was pretty much what I envisioned but I lack the actual technical knowledge and didn't want to make assumptions. This is definitely nothing easy and I honestly don't expect this to happen. A simple slider to reduce Rendertime RAM usage would be a great start.


Yeah, I think it would have to be in the render dialog for it to be useful.

For instance, if it was only a slider in the preferences, you might accidentally clear out 80% of your cached frames just because you made a preview render of a few frames.

If the process of rendering clears out your cache, you'd want the user to be very explicit about doing it.
Chad Capeland
Indicated, LLC
www.floweffects.com
Offline

Kel Philm

  • Posts: 505
  • Joined: Sat Nov 19, 2016 6:21 am

Re: Fusion 18 feature requests

PostFri Dec 11, 2020 9:51 pm

My hitlist

* UI customisation

* UI Scale options (150% at least)

* Add option to crop Merge node to BG DoD or Frame size (Would improve comp speed in a lot of situations)

* +1 for Chads import colour nodes from Resolve

* Show VFX Connect some love (Configuration, improve file layout, add hooks into the Resolve DB to manage comps from Fusion UI)

* Optimise Fusion engine to maximise CPU/GPU processing.

* Allow the speed editor panel to be re-purposed in Fusion (Jog wheel in particular)

* More interaction with users on the Forum
Offline

Duca Simone Luchini

  • Posts: 59
  • Joined: Thu Jun 13, 2019 6:29 am
  • Real Name: Duca Simone Luchini

Re: Fusion 18 feature requests

PostSat Dec 12, 2020 7:08 am

More camera raw file IMPORT as Canon CR2, please! :P
Offline

Dazzer

  • Posts: 211
  • Joined: Sat Aug 24, 2019 8:31 am
  • Real Name: Daz Harris

Re: Fusion 18 feature requests

PostSat Dec 12, 2020 7:37 pm

I'd like to see all FX which are available for Resolve Fusion also to be available in Fusion Studio.

For example Deflicker & Motion Trials.

And i'd like to be able to get a live feed from another application into the node graph.

I'm creating stuff in Unreal Engine and it would be great to be able to have a still output from Unreal (either via a hardware loop or by doing a screen grab) to get that into Fusion without having to export / import. So if the image is changed in Unreal it automatically updates in the node graph.
Offline

Frank Engel

  • Posts: 736
  • Joined: Thu May 17, 2018 11:09 pm
  • Real Name: Frank Engel

Re: Fusion 18 feature requests

PostTue Dec 15, 2020 9:48 am

Fixing the UI to be more flexible again and improved behavior in a low-memory situation would definitely be at the top of my list as well.
Offline

wfolta

  • Posts: 598
  • Joined: Fri May 15, 2020 1:12 pm
  • Real Name: Wayne Folta

Re: Fusion 18 feature requests

PostTue Dec 15, 2020 3:16 pm

I think the things listed so far are great, and I add +1.

A smaller issue (for me) would be better network rendering. It's not super-critical for me -- a one-man shop with one computer -- but I did experiment a bit with "network" rendering on my single machine so I could continue to work while rendering, which is a feature that Fusion Studio has that Resolve does not.

But it's a bit confusing and in the end I never got it to work. The renderer gets an error trying to open the comp on the same machine. MacOS, and both Fusion and the renderer have full disk access permissions, so I can't figure out why it doesn't work. I did some forum searching and it seems that people who heavily use network rendering also have reliability issues -- sometimes it works other times it doesn't, sometimes renderers drop out, etc.

Maybe really serious shops have 3rd party orchestration and so don't use Fusion's. Maybe the provided orchestration is more of a checkbox than an actual feature. And, again, I'm not a heavy user and I think this request is lower priority than others, but I wanted to throw it out there.
Resolve Studio 17 latest, Fusion Studio 17 latest, macOS Big Sur latest, MacBook Pro 2020 64GB RAM and Radeon Pro 5600M 8GB VRAM
Offline
User avatar

Bryan Ray

  • Posts: 2080
  • Joined: Mon Nov 28, 2016 5:32 am
  • Location: Los Angeles, CA, USA

Re: Fusion 18 feature requests

PostTue Dec 15, 2020 3:24 pm

Speaking of network rendering: Bring back clustering!
https://www.steakunderwater.com/VFXPedi ... references

I'm not sure how simple that is now that everything is GPU accelerated, but we've got this nice big GPU farm for Redshift. Would be amazing to let Fusion take advantage of it, too!
Bryan Ray
http://www.bryanray.name
http://www.musevfx.com
Offline

Joël Gibbs

  • Posts: 97
  • Joined: Wed Nov 12, 2014 9:18 pm
  • Location: Nashville

Re: Fusion 18 feature requests

PostTue Dec 15, 2020 8:16 pm

Bryan Ray wrote:Speaking of network rendering: Bring back clustering!
https://www.steakunderwater.com/VFXPedi ... references


yes!! +1
Offline

UserNoah

  • Posts: 336
  • Joined: Fri Oct 04, 2019 3:32 pm
  • Location: Berlin
  • Real Name: Noah Hähnel

Re: Fusion 18 feature requests

PostTue Dec 15, 2020 11:08 pm

Another one from me:

Display LUTs for the Studio Player.

It's faster to see your shots in sequence through the Studio Player than waiting for Resolve to open, but there are only manual color correction sliders like Gain or Gamma. Currently no way to select a Display LUT like OCIO ACEScg to Output-sRGB.
Offline
User avatar

GregAusina

  • Posts: 164
  • Joined: Thu Aug 23, 2012 7:30 pm
  • Location: Paris - France

Re: Fusion 18 feature requests

PostThu Aug 05, 2021 8:19 pm

- give Krokodove the place it deserves. This suite is brillant but a lot of very interesting tools are deprecated due to the lack of support.
I don’t understand why Bmd do not take that chance of collaborating with people like Raf Shoenmakers or the wsl community .Theses people have been developing/using fusion for decades, they are already offering Reactor for free.
Please BMD give them the credit they deserve. Hire them or make any necessary deal. Give fusion a real boost.
Grégoire AUSINA / Gisèle productions
Davinci 17.4.1 on Windows 10 (21H1) / Nvidia studio driver 472.39/ Dual Xeon E-2650v4 / supermicro X10DRG-Q / 128 Go RAM / GUI GTX 1070 / Cuda GTX 3090 - FSI CM250 - Tangent Elements - Accusys A12S3-p+
Offline
User avatar

TheBloke

  • Posts: 1904
  • Joined: Sat Nov 02, 2019 11:49 pm
  • Location: UK
  • Real Name: Tom Jobbins

Re: Fusion 18 feature requests

PostFri Aug 06, 2021 7:27 am

wfolta wrote:A smaller issue (for me) would be better network rendering. It's not super-critical for me -- a one-man shop with one computer -- but I did experiment a bit with "network" rendering on my single machine so I could continue to work while rendering, which is a feature that Fusion Studio has that Resolve does not.

But it's a bit confusing and in the end I never got it to work. The renderer gets an error trying to open the comp on the same machine. MacOS, and both Fusion and the renderer have full disk access permissions, so I can't figure out why it doesn't work. I did some forum searching and it seems that people who heavily use network rendering also have reliability issues -- sometimes it works other times it doesn't, sometimes renderers drop out, etc.
Maybe make a separate post about this and post what you're doing and what you're seeing?

It's definitely possible to get Fusion Studio network rendering working. I did it last year across my three (crappy) machines, all running macOS. There are reports of problems, and I had some stalled renders myself. However I don't believe that was specific to network rendering. But it worked for me in the end, and was definitely useful.

Anyway back on topic, definite +1 for me on improving network rendering, and adding it to Resolve Fusion as well. And +1 to most of the other suggestions as well - especially restoring the fully flexible UI, and improving the flexibility of the Resolve Fusion UI. Some work was done in that area in v17 (restored UI layouts, customisable taskbar), so I hope this continues.
Resolve Studio 17.4.3 and Fusion Studio 17.4.3 on macOS 11.6.1

Hackintosh:: X299, Intel i9-10980XE, 128GB DDR4, AMD 6900XT 16GB
Monitors: 1 x 3840x2160 & 3 x 1920x1200
Disk: 2TB NVMe + 4TB RAID0 NVMe; NAS: 36TB RAID6
BMD Speed Editor
Offline

Rick van den Berg

  • Posts: 970
  • Joined: Tue Jun 02, 2015 7:47 am

Re: Fusion 18 feature requests

PostTue Aug 10, 2021 6:48 pm

a reset of the incredibly messed up UI


when was the last time fusion got a real update? My hopes are not too high..
resolve 17.1
windows 10
i7 6950x
2x gtx 1080 8gb
96gb DDR4 3200
Offline

Jeff Ha

  • Posts: 157
  • Joined: Tue Apr 14, 2015 10:38 am

Re: Fusion 18 feature requests

PostWed Aug 11, 2021 12:32 pm

Rick van den Berg wrote:a reset of the incredibly messed up UI


when was the last time fusion got a real update? My hopes are not too high..



that's the last time. It's when they wanted to make the UI look like Resolve, and then force-f it into Resolve. Oh those were the days. Hopes haven't been high for Fusion for a long time. This is why I still use Nuke, although expensive it's still more feature rich and stable. BMD devs could learn from it.
Offline

GalinMcMahon

  • Posts: 152
  • Joined: Sun Sep 29, 2019 10:14 pm
  • Real Name: Galin McMahon

Re: Fusion 18 feature requests

PostWed Aug 11, 2021 4:15 pm

I just have a few requests. I'm still pretty new to Fusion so my suggestions may be pretty elementary or may already exist.

1. UI. As others have mentioned. One of the very few things I liked more about Premiere was its UI customization. I love what Resolve does have but things like undocking or hiding unnecessary (to those who use shortcuts) menu bars would be very helpful. This would help greatly with multiple monitors as the default layout is typically not one I would use in my flow.

2. General rendering / memory management / reliability improvements. I use an 8GB video card and my memory runs out very fast even with simple node trees. I get regular hard crashes. My comps never seem to render to the point that I can view them at full frame rate. I would say that this is the most important upgrade in my book.

3. Smarter keying. This may be user error but I am having issues pulling a key from a (admittedly) poorly-lit blue face mask against a rust red background. The white fringe of the blue will not key out. Perhaps Fusion could see that this fringe moves with the blue and then assume that the fringe therefore is part of the blue and should be keyed. I do understand in/out, blur etc but that affects the whole mask and in at least this instance, part of the mask needs to be as hard a line as possible.

4. Some simplification / dumbing down. Please don't hate on me for even suggesting this! Fusion is very intimidating for newer users (like me.) Things are finally starting to click but there are some actions that just seem illogical and concrete in the way they need to be performed. The big one for me is masking multiple spots on the screen at the same time. It would make the most sense to allow multiple polygons to go into a single MediaIn but that doesn't seem to be an option. Then when masks intersect things really get weird (this is probably a feature for the more advanced of "us.") I think a good way to look at this would be to have a focus group of novices try to do certain tasks to see how other people think and then implement that at least as an alternative method.
Windows 10
i7+-8750 @2.20GHz
32GB optane RAM
8GB Geforce 1070
500GB NvME main - 1TB Evo 860 SSD scratch
Resolve 17 Studio
Offline
User avatar

TheBloke

  • Posts: 1904
  • Joined: Sat Nov 02, 2019 11:49 pm
  • Location: UK
  • Real Name: Tom Jobbins

Re: Fusion 18 feature requests

PostWed Aug 11, 2021 4:48 pm

GalinMcMahon wrote:1. UI. As others have mentioned.
Yes definitely.

GalinMcMahon wrote:2. General rendering / memory management / reliability improvements.
Yes absolutely. There are some major performance/stability problems at the moment. Particularly in Resolve, but sadly also in Fusion Studio - particularly related to GPU acceleration of tools.

GalinMcMahon wrote:3. Smarter keying.
This is primarily user lack-of-knowledge. There is an alternative keyer in Fusion Studio - Primatte - which gives another set of tools which, sadly, are not available in Resolve Fusion (likely because Primatte has licensing costs that BMD did not want to pay in Resolve.)

However I've not heard many Fusion experts talk about using Primatte anyway. If you've got a sub-optimal keying situation, no tool will be able to do it all immediately. It always takes a bit of work and manipulation of the source image - sometimes a lot of work - and you can definitely do extremely advanced keying with the tools available in Resolve Fusion. And there's also various free tools and plugins available which can make life easier.

As always, it's a case of knowing how to do it. Here are two tutorials - one beginner friendly, one more advanced.





There are many others as well, but these are two I have watched and can vouch for as someone who is both a newbie, and has had to do a bunch of difficult green screen keying. The second one in particular has given me excellent results with some my poor quality footage (shot on webcams in bad, uneven light with creased green screens.)

GalinMcMahon wrote:4. Some simplification / dumbing down.
Again I would say this is just a case of learning. As well as many YouTube videos, there's also a free training PDF from Blackmagic Design which is pretty good I believe (haven't looked at it myself yet, but their Edit and Fairlight training books are good, so I imagine this is too)

GalinMcMahon wrote:The big one for me is masking multiple spots on the screen at the same time. It would make the most sense to allow multiple polygons to go into a single MediaIn but that doesn't seem to be an option.
Simply pipe one mask node into another, making a chain of masks. Then, optionally, use the "Paint Mode" control in one or more of the masks to choose how it combines with its predecessor(s).

Image Image

I do agree that there are things BMD can and should do to make it easier to use Fusion - especially in Resolve. The way Fusion has been integrated into Resolve is currently quite confusing - in terms of where compositions live, how they get media, how they are accessed again - and could be simplified.

There could also be a lot more help in-UI, with better tooltips and in-app documentation. This is a problem throughout Resolve, not just in Fusion.

Fusion is also notorious for having 'hidden' right-click menus all over the place, often including vital features. New users are often recommended to 'right click everywhere!' just to see what features are hidden there.

All this could be improved. But I don't agree that the fundamental concepts should be dumbed down or simplified - in my view that would be taking us backwards. Better to spend that time on making it easier for users to learn how things work now, like with UI improvements.
Resolve Studio 17.4.3 and Fusion Studio 17.4.3 on macOS 11.6.1

Hackintosh:: X299, Intel i9-10980XE, 128GB DDR4, AMD 6900XT 16GB
Monitors: 1 x 3840x2160 & 3 x 1920x1200
Disk: 2TB NVMe + 4TB RAID0 NVMe; NAS: 36TB RAID6
BMD Speed Editor
Offline

Charles Reilly

  • Posts: 14
  • Joined: Tue Nov 10, 2015 9:15 pm

Re: Fusion 18 feature requests

PostThu Aug 12, 2021 12:13 am

Blender

I see the power of Fusions 3D software however I struggle to find my way around it....it would be great if there were an option to use the same controls as blender?

Blender is free and I would also like to see more connection with Fusion ie...taking a blender scener (or object) between the two programs....... currently I cannot get any Blender objects into fusion without losing textures... and yes I have tried..FBX, OBJs etc

My biggest wish is to be able to to take 3d tracking info into Blender ......and Vice Versa

Charles
Offline

Sander de Regt

  • Posts: 2456
  • Joined: Thu Nov 13, 2014 10:09 pm

Re: Fusion 18 feature requests

PostThu Aug 12, 2021 7:06 am

My biggest wish is to be able to to take 3d tracking info into Blender ......and Vice Versa

That works fine with FBX at the moment as far as I'm aware of. Just make sure your framerate and resolution match and you should be able to go back and forth as much as you want.
Sander de Regt

ShadowMaker SdR
The Netherlands
Offline

GalinMcMahon

  • Posts: 152
  • Joined: Sun Sep 29, 2019 10:14 pm
  • Real Name: Galin McMahon

Re: Fusion 18 feature requests

PostFri Aug 13, 2021 9:48 pm

All this could be improved. But I don't agree that the fundamental concepts should be dumbed down or simplified - in my view that would be taking us backwards. Better to spend that time on making it easier for users to learn how things work now, like with UI improvements.


Great reply, thank you! I will try some of the methods on those videos. Right now I'm looking at frame by frame rotoscoping up to 9 masks per frame for about 9 minutes! That might take a while at this rate! Thanks again :)
Windows 10
i7+-8750 @2.20GHz
32GB optane RAM
8GB Geforce 1070
500GB NvME main - 1TB Evo 860 SSD scratch
Resolve 17 Studio
Offline
User avatar

TheBloke

  • Posts: 1904
  • Joined: Sat Nov 02, 2019 11:49 pm
  • Location: UK
  • Real Name: Tom Jobbins

Re: Fusion 18 feature requests

PostSat Aug 14, 2021 12:02 pm

GalinMcMahon wrote:Great reply, thank you! I will try some of the methods on those videos. Right now I'm looking at frame by frame rotoscoping up to 9 masks per frame for about 9 minutes! That might take a while at this rate! Thanks again :)
If you've got a green or blue screen I'm sure it can be done more easily than rotoing the whole thing. If you're still having trouble, maybe make a new topic and show an example frame or two. Then we can try and help.
Resolve Studio 17.4.3 and Fusion Studio 17.4.3 on macOS 11.6.1

Hackintosh:: X299, Intel i9-10980XE, 128GB DDR4, AMD 6900XT 16GB
Monitors: 1 x 3840x2160 & 3 x 1920x1200
Disk: 2TB NVMe + 4TB RAID0 NVMe; NAS: 36TB RAID6
BMD Speed Editor
Offline

joymonger

  • Posts: 47
  • Joined: Wed Apr 15, 2020 3:28 pm
  • Real Name: Paige Saunders

Re: Fusion 18 feature requests

PostTue Dec 07, 2021 9:48 am

VFX Connect feels like a minimum viable product that no one got around to building out. It'd be awesome if you could:
- Create without sending footage
- Connect a fusion comp that already exists
- Have it render a wav file and connect it so the audio is synced
- Allow relinking and folder moving
- Create folder with name based off the composition name
- Trigger renders from inside resolve

I have a template fusion project and system for all this but basically every comp is 5 minutes of rendering wavs, renaming files, rendering blank exf placeholders, going back and forward between resolve and fusion making sure there aren’t any typos and that I didn’t accidentally link the previous composition.
Windows 10 / RTX 2060 / 32GB Ram
Offline

tomhome

  • Posts: 24
  • Joined: Wed Jun 26, 2019 2:21 pm
  • Real Name: Tom Bremer

Re: Fusion 18 feature requests

PostTue Dec 07, 2021 6:40 pm

Better multichannel support.
1. The ability to manipulate any channel, whether its in the RGBA channels or not, would be amazing.
2. Abililty to use any available embedded channel as a mask without having to boolean it out and bitmap mask it.
3. A proper shuffle type node so that duplicating the same reader isn't necessary for different AOV channels
4. Create and name arbitrary channels

5. UI Customization, for sure.

That's really the only thing I miss about Nuke.
Offline

Jacob Danell

  • Posts: 143
  • Joined: Tue Aug 28, 2012 2:25 pm
  • Location: Sweden

Re: Fusion 18 feature requests

PostWed Dec 08, 2021 9:39 am

tomhome wrote:Better multichannel support.
1. The ability to manipulate any channel, whether its in the RGBA channels or not, would be amazing.
2. Abililty to use any available embedded channel as a mask without having to boolean it out and bitmap mask it.
3. A proper shuffle type node so that duplicating the same reader isn't necessary for different AOV channels
4. Create and name arbitrary channels

5. UI Customization, for sure.

That's really the only thing I miss about Nuke.


Sorry to say but that won't happen with Fusion as the way channels work are hardcoded into the SDK, meaning everything using channels would need to be re-written, including custom fuses and so on.

The way Fusion works isn't worse tho, it's just different. Secondman over at WSL have written a good text about it: https://www.steakunderwater.com/wesuckl ... 189#p30189
The whole thread is pretty interesting for Nuke people going over to Fusion :)
Offline

UserNoah

  • Posts: 336
  • Joined: Fri Oct 04, 2019 3:32 pm
  • Location: Berlin
  • Real Name: Noah Hähnel

Re: Fusion 18 feature requests

PostWed Dec 08, 2021 9:56 am

I think the biggest issues about channels is simply the importing and shuffling part. I personally would think an approach similar to how multipart EXRs are handled could work in fusion.
Have a specific loader that can read and pass through all multiparts and send all parts to a new shuffle node which can select between all these multiparts exactly like the Loader can right now and puts these into rgba. From there you can use a Channel Boolean to bring this into the data channels if necessary.
But I would assume that 9 out of 10 times this will be used for multi channel compositing where you’re layering different rgb channels to recreate the beauty.

The benefits: Only one loader to keep track of per exr sequence and possibly faster rendering because there’s only one I/O call per frame instead of loading each part individually.
This also keeps comp data small because you don’t pass through hundreds of channels you don’t need and only two new nodes would be needed while the rest of Fusion can be left untouched.
Maybe there is even a better idea that I haven’t thought about but that would be my current idea on how to implement that without breaking everything.
I wonder if something like that would already be possible as the cryptomatte fuse is already reading in and processing several more channels I believe.
Offline

Jacob Danell

  • Posts: 143
  • Joined: Tue Aug 28, 2012 2:25 pm
  • Location: Sweden

Re: Fusion 18 feature requests

PostWed Dec 08, 2021 10:08 am

Yes, there are a lot of improvements to be done on the multi-channel exr workflow in Fusion! It won't just be arbitrary channels.
A custom EXR loader with multiple outputs could maybe be something so you just pipe out what ever it is you need.
Or a loader like that accompanied with a special wireless node where you select what channel you wish to read :)
Offline

Hendrik Proosa

  • Posts: 2001
  • Joined: Wed Aug 22, 2012 6:53 am
  • Location: Estonia

Re: Fusion 18 feature requests

PostWed Dec 08, 2021 11:32 am

Jacob Danell wrote:The way Fusion works isn't worse tho, it's just different. Secondman over at WSL have written a good text about it: https://www.steakunderwater.com/wesuckl ... 189#p30189
The whole thread is pretty interesting for Nuke people going over to Fusion :)

Interesting thread. The argument about explicit vs implicit doesn't make much sense to me though. I don't see how implicit hard-coded use of channels in Fusion that user has no control over and where use of channels is all assumed is more explicit than user being able to arbitrarily add, remove, rename etc and select for any operation explicitly in Nuke, at any point in the flow. Nuke does not calculate any and all channels in all nodes, it only does something with channels that are actually requested. I have read sometimes that people think all channels get processed all the time. They don't.
I do stuff.
Offline

tomhome

  • Posts: 24
  • Joined: Wed Jun 26, 2019 2:21 pm
  • Real Name: Tom Bremer

Re: Fusion 18 feature requests

PostWed Dec 08, 2021 2:28 pm

Jacob Danell wrote:
tomhome wrote:Better multichannel support.
Sorry to say but that won't happen with Fusion as the way channels work are hardcoded into the SDK, meaning everything using channels would need to be re-written, including custom fuses and so on.

The way Fusion works isn't worse tho, it's just different. Secondman over at WSL have written a good text about it: https://www.steakunderwater.com/wesuckl ... 189#p30189
The whole thread is pretty interesting for Nuke people going over to Fusion :)


Yes, I read somewhere that the channels were hardcoded, and that's fine and all, but I also read that the loader was created and unchanged since before the multi channel workflow happened/caught on so perhaps its time to do an overhaul.

But regardless, it seems like since fusion can see the channels in the loader, that there could be a node that can see said channels and put them into different arbitrary channels. We can kind of do this by filling in all the AOVs into the "extra channels" (depth, normals, etc) and then using custom tool or channel boolean to "shuffle" them to the RGB channels without having to use the splitEXR script. There is also the AuxChannel tool I found on VFXpedia that adds and removes channels, but those are the hardcoded ones only.

So what I'm saying is it seems like everything that's required to make that node is there, it just needs the node.
Offline
User avatar

Pieter Van Houte

  • Posts: 558
  • Joined: Wed Nov 05, 2014 1:04 am

Re: Fusion 18 feature requests

PostTue Dec 14, 2021 2:25 am

Hendrik Proosa wrote:
Jacob Danell wrote:The way Fusion works isn't worse tho, it's just different. Secondman over at WSL have written a good text about it: https://www.steakunderwater.com/wesuckl ... 189#p30189
The whole thread is pretty interesting for Nuke people going over to Fusion :)

Interesting thread. The argument about explicit vs implicit doesn't make much sense to me though. I don't see how implicit hard-coded use of channels in Fusion that user has no control over and where use of channels is all assumed is more explicit than user being able to arbitrarily add, remove, rename etc and select for any operation explicitly in Nuke, at any point in the flow. Nuke does not calculate any and all channels in all nodes, it only does something with channels that are actually requested. I have read sometimes that people think all channels get processed all the time. They don't.


Perhaps my terminology (implicit vs explicit) wasn't the clearest. I mean that mostly from the point of view of reading the comp, or even more so someone else's comp.

I make no reference to either Fusion or Nuke calculating something that's not required.
Last edited by Pieter Van Houte on Tue Dec 14, 2021 2:27 am, edited 1 time in total.
We Suck Less
Offline
User avatar

Pieter Van Houte

  • Posts: 558
  • Joined: Wed Nov 05, 2014 1:04 am

Re: Fusion 18 feature requests

PostTue Dec 14, 2021 2:26 am

tomhome wrote:So what I'm saying is it seems like everything that's required to make that node is there, it just needs the node.


This topic on WSL may be of interest to you: https://www.steakunderwater.com/wesuckl ... php?t=4176
We Suck Less
Offline

Hendrik Proosa

  • Posts: 2001
  • Joined: Wed Aug 22, 2012 6:53 am
  • Location: Estonia

Re: Fusion 18 feature requests

PostTue Dec 14, 2021 8:39 am

Pieter Van Houte wrote:Perhaps my terminology (implicit vs explicit) wasn't the clearest. I mean that mostly from the point of view of reading the comp, or even more so someone else's comp.

I make no reference to either Fusion or Nuke calculating something that's not required.

I somewhat agree with this aspect, from channel usage perspective :) That what gets calculated bit was just to emphasize something I tend to stumble upon time and again in different places.
I do stuff.
Offline
User avatar

Chad Capeland

  • Posts: 2742
  • Joined: Mon Nov 10, 2014 9:40 pm

Re: Fusion 18 feature requests

PostTue Dec 14, 2021 6:04 pm

tomhome wrote: it just needs the node.


So you have a node that outputs arbitrary channels. Then what? Unless you have a node that can operate on them, what does that get you?

Likewise, how would those arbitrary channels be handled in Resolve?
Chad Capeland
Indicated, LLC
www.floweffects.com
Offline

tomhome

  • Posts: 24
  • Joined: Wed Jun 26, 2019 2:21 pm
  • Real Name: Tom Bremer

Re: Fusion 18 feature requests

PostWed Dec 15, 2021 8:25 pm

Chad Capeland wrote:
tomhome wrote: it just needs the node.


So you have a node that outputs arbitrary channels. Then what? Unless you have a node that can operate on them, what does that get you?

Likewise, how would those arbitrary channels be handled in Resolve?



If I had a node that read the AOVs from a read node and allowed them to be put into any of the built in channels, then I can shuffle them how I want. Its not as good as nuke, but I could use one loader instead of 5 or more. Even if the node could allow me to shuffle any AOV into RGB channels, I could use those the same way as I have to now except, again, using one loader instead of any number of them.

If I go into the loader and set all the channels properly with the depth going to the correct channel, connecting the UV AOV to the UV channels, etc, I can do it with a custom tool. Just instead of r1, g1, b1, I would tell them to use nx1, ny1, nz1 and it would do what I'm explaining above.

The problem is when I have a bunch of lighting AOVs, theres no place for me to "put" them. So I have to bring in new loaders with the RGB channels set to the particular light AOV channels. I have played with bringing ONE more loader and setting the light AOVs to RGB, A, Z, U,V, NX, NY, NZ, etc, and then using the custom tool to split them out. This works but that gets confusing since the names dont match up at all with the channel.

TL;DR If getting a proper multichannel workflow is not going to happen as everyone says, at the very least I would want a node that can read the AOVs from the loader and allow me to shuffle them to RGB A channels for me to use 1 loader, instead of how ever many the shot requires.
Offline

Theodor Groeneboom

  • Posts: 291
  • Joined: Fri Jun 26, 2015 12:51 pm

Re: Fusion 18 feature requests

PostThu Dec 16, 2021 10:17 am

tomhome wrote:
Chad Capeland wrote:
tomhome wrote: it just needs the node.


So you have a node that outputs arbitrary channels. Then what? Unless you have a node that can operate on them, what does that get you?

Likewise, how would those arbitrary channels be handled in Resolve?



If I had a node that read the AOVs from a read node and allowed them to be put into any of the built in channels, then I can shuffle them how I want. Its not as good as nuke, but I could use one loader instead of 5 or more. Even if the node could allow me to shuffle any AOV into RGB channels, I could use those the same way as I have to now except, again, using one loader instead of any number of them.

If I go into the loader and set all the channels properly with the depth going to the correct channel, connecting the UV AOV to the UV channels, etc, I can do it with a custom tool. Just instead of r1, g1, b1, I would tell them to use nx1, ny1, nz1 and it would do what I'm explaining above.

The problem is when I have a bunch of lighting AOVs, theres no place for me to "put" them. So I have to bring in new loaders with the RGB channels set to the particular light AOV channels. I have played with bringing ONE more loader and setting the light AOVs to RGB, A, Z, U,V, NX, NY, NZ, etc, and then using the custom tool to split them out. This works but that gets confusing since the names dont match up at all with the channel.

TL;DR If getting a proper multichannel workflow is not going to happen as everyone says, at the very least I would want a node that can read the AOVs from the loader and allow me to shuffle them to RGB A channels for me to use 1 loader, instead of how ever many the shot requires.


Agreed.

Fusion
+1 for updated external libs (ocio specifically)
+1 Proper USD workflow
+1 Hydra Render workflow
+1 flexible UI back
+1 resolve ML tools (like magic mask) in fusion.

Resolve/Fusion
+1 General performance and stability improvements. (But seriously, having used Fu in resolve its painfully sluggish compared to standalone)
Offline
User avatar

Chad Capeland

  • Posts: 2742
  • Joined: Mon Nov 10, 2014 9:40 pm

Re: Fusion 18 feature requests

PostFri Dec 17, 2021 1:00 am

tomhome wrote:
Chad Capeland wrote:
tomhome wrote: it just needs the node.


So you have a node that outputs arbitrary channels. Then what? Unless you have a node that can operate on them, what does that get you?

Likewise, how would those arbitrary channels be handled in Resolve?



If I had a node that read the AOVs from a read node and allowed them to be put into any of the built in channels, then I can shuffle them how I want. Its not as good as nuke, but I could use one loader instead of 5 or more. Even if the node could allow me to shuffle any AOV into RGB channels, I could use those the same way as I have to now except, again, using one loader instead of any number of them.


The LD now can put any AOV into any of the predefined channels and any other tool can input them. If you didn't use the predefined channels, though, then no other tool could input them to make use of them.

One option would be to have the LD offer a user-defined number out outputs. So you could have a button in the LD that added a second, third, fourth, whatever output. Each time you added an output you would get a new tab where you could assign the AOV to an image channel.

EDIT: Another thing I'd like to see is a means of outputting more than just RGBA to Resolve. Some way to send any number of image channels out as RGBA "clips" that are locked together in Resolve. Would it be clunkier than having Resolve manage arbitrary channels in each clip? Sure, but I don't see the other pages of Resolve each add support for than any time soon.
Chad Capeland
Indicated, LLC
www.floweffects.com
Offline

Theodor Groeneboom

  • Posts: 291
  • Joined: Fri Jun 26, 2015 12:51 pm

Re: Fusion 18 feature requests

PostSat Dec 18, 2021 7:44 am

Chad Capeland wrote:
One option would be to have the LD offer a user-defined number out outputs. So you could have a button in the LD that added a second, third, fourth, whatever output. Each time you added an output you would get a new tab where you could assign the AOV to an image channel.



This would also be an acceptable way of pulling out arbitrary channels out of a single exr.

I just hate having to manually update 15+ loaders for new renders...
Offline

Marco Quaglia Faccio

  • Posts: 69
  • Joined: Sat Feb 18, 2017 6:49 am

Re: Fusion 18 feature requests

PostSat Dec 18, 2021 4:32 pm

Add Markers to the time line (like in Davinci resolve).
Useful for synchronizing motion graphics element's with background music.
Resolve Studio 17 | Fusion Studio 17 | Win 10 Enterprise (64-bit) |
X99-A II | i7-6800K CPU @ 3.40GHz | Memory 32GB
NVIDIA GeForce GTX 1080 | Samsung SSD 970 EVO Plus 500GB | CT240BX500SSD1 | WDC WD30EFRX-68EUZN0
Offline

Sander de Regt

  • Posts: 2456
  • Joined: Thu Nov 13, 2014 10:09 pm

Re: Fusion 18 feature requests

PostSat Dec 18, 2021 8:08 pm

Marco Quaglia Faccio wrote:Add Markers to the time line (like in Davinci resolve).


You can do this now. They're called guides and at least in the timeline and spline view you can add them to your comp.
Sander de Regt

ShadowMaker SdR
The Netherlands
Offline
User avatar

Chad Capeland

  • Posts: 2742
  • Joined: Mon Nov 10, 2014 9:40 pm

Re: Fusion 18 feature requests

PostTue Dec 21, 2021 10:45 pm

Theodor Groeneboom wrote:
I just hate having to manually update 15+ loaders for new renders...


Would also reduce network traffic and improve caching.

As long as the requests to the outputs were processed independently, I don't see much downside. You could always use multiple LD's if that's what you would prefer, but having one LD with multiple outputs would solve most of the workflow issues without requiring any other changes.
Chad Capeland
Indicated, LLC
www.floweffects.com
Offline

Eyn-0519

  • Posts: 4
  • Joined: Thu Jun 10, 2021 4:00 pm
  • Real Name: Hui Eyn

Re: Fusion 18 feature requests

PostMon Jan 03, 2022 4:33 am

Any news recently of Fusion 18? thanks!
Offline
User avatar

Bryan Ray

  • Posts: 2080
  • Joined: Mon Nov 28, 2016 5:32 am
  • Location: Los Angeles, CA, USA

Re: Fusion 18 feature requests

PostMon Jan 03, 2022 2:25 pm

Per BMD's usual policies, there will be no news until they actually release it. And they typically do that at one of the conferences, either NAB in the spring or SIGGRAPH in the summer.
Bryan Ray
http://www.bryanray.name
http://www.musevfx.com
Offline

Rick van den Berg

  • Posts: 970
  • Joined: Tue Jun 02, 2015 7:47 am

Re: Fusion 18 feature requests

PostTue Jan 04, 2022 9:54 pm

And the beta will be released last september at IBC
resolve 17.1
windows 10
i7 6950x
2x gtx 1080 8gb
96gb DDR4 3200
Offline

Kel Philm

  • Posts: 505
  • Joined: Sat Nov 19, 2016 6:21 am

Re: Fusion 18 feature requests

PostWed Jan 05, 2022 1:51 am

Hopefully the biggest change won't be there is now an 8 where the 7 was ...
Offline

Jeff Ha

  • Posts: 157
  • Joined: Tue Apr 14, 2015 10:38 am

Re: Fusion 18 feature requests

PostWed Jan 05, 2022 12:19 pm

Kel Philm wrote:Hopefully the biggest change won't be there is now an 8 where the 7 was ...


that and the 3 weeks it'll take to change the documentation completely to reflect that change. I can't see them doing much with Fusion. They're essentially giving it away in Resolve and the Studio version is dirt cheap with no follow on maintenance or paid upgrades. It's tricky to pay developers and having them add a lot of meaningful features when little revenue is coming in from your department (software: Resolve/Fusion). This is how Pixologic got in trouble with Zbrush... free upgrades for years. It came back to bite them in the ass over the past 3 years and ended up selling out to Maxon recently. "Fusion: At least we don't cost as much as Nuke..." it's a shame Natron isn't picking up steam anymore.

I'm still looking forward to Autograph later this year; https://www.left-angle.com/?page=95-en#page=95
Offline
User avatar

TheBloke

  • Posts: 1904
  • Joined: Sat Nov 02, 2019 11:49 pm
  • Location: UK
  • Real Name: Tom Jobbins

Re: Fusion 18 feature requests

PostWed Jan 05, 2022 1:03 pm

Jeff Ha wrote:and the 3 weeks it'll take to change the documentation completely
More like three months. Or six months if you count from when the first beta of v17 landed.
Last edited by TheBloke on Wed Jan 05, 2022 8:19 pm, edited 1 time in total.
Resolve Studio 17.4.3 and Fusion Studio 17.4.3 on macOS 11.6.1

Hackintosh:: X299, Intel i9-10980XE, 128GB DDR4, AMD 6900XT 16GB
Monitors: 1 x 3840x2160 & 3 x 1920x1200
Disk: 2TB NVMe + 4TB RAID0 NVMe; NAS: 36TB RAID6
BMD Speed Editor
Next

Return to Fusion

Who is online

Users browsing this forum: No registered users and 11 guests