Discussion: Future of slowmoVideo #179
Comments
|
Which video editing applications did you had in mind? May I recommend some: https://www.openshot.org/ |
|
Those are all real-time video editors...I figured a non-real-time compositing editor, like Natron, would be more appropriate. Also, it seems that slowmoVideo would benefit from an optical-flow algorithm capable of expressing that there's no prior-frame analog for a current-frame pixel. None of the optical-flow algorithms in use by slowmoVideo seem to account for that possibility. |
|
@alexmyczko I would primarily target Kdenlive because I have been contributing there as well. However, it will be located at a lower level like in MLT. See also #173. @ulatekh Is Natron still in development? I have to play around with it first in order to understand it. Current video editors use optical flow based algorithms as well, however only with static speed changes (e.g. change the clip speed to 0.5× and intermediate frames will be calculated). That would be fairly easy to integrate. With curves it is a bit more difficult as it requires a UI. The algorithms sound interesting, which ones are you talking about specifically? sV could not handle that yet but it would be a nice extension, I would expect it to give better results in many situations where area is uncovered. |
|
Hi Simon,
To answer your questions:
Do the milestones make sense for you?
At a first glance, yes, but I have one suggestion for an extra feature that
I would find very useful:
When I use slowmoVideo, I tend to make videos with at least 10x slowdown,
and on average CPU utilization is low because the interpolation tasks don't
seem to be running in parallel.
If slowmoVideo would allow running interpolation jobs in parallel, I can
imagine I would see large performance improvements (more than 4x), so I
would like to propose this as a new feature for one of the milestones.
Is there an area where you would like to help?
Yes, but currently my schedule won't permit it, unfortunately.
Best regards,
Elias
…On Fri, Apr 10, 2020 at 2:14 PM Simon A. Eugster ***@***.***> wrote:
@valgit <https://github.com/valgit> @grota <https://github.com/grota>
@ulatekh <https://github.com/ulatekh> @mgo80 <https://github.com/mgo80>
@Eliasvan <https://github.com/Eliasvan> @fishpepper
<https://github.com/fishpepper> @d-j-a-y <https://github.com/d-j-a-y>
@viktorsmari <https://github.com/viktorsmari> @pmo-19
<https://github.com/pmo-19> @new-life <https://github.com/new-life>
@metal3d <https://github.com/metal3d> @habi <https://github.com/habi>
@Ichthyostega <https://github.com/Ichthyostega> You (and others) have
been contributing to slowmoVideo, so I would like to give you a short
update.
What is this about?
I intend to revive slowmoVideo. I somehow expected other open source slow
motion tools would be around after that many years, but there are none.
(Correct me if I am wrong.) So here we go again!
From my POV, I have summarised the curent status
<https://github.com/slowmoVideo/slowmoVideo/wiki/Development-Planning> of
slowmoVideo, mainly for me to get an idea of where the project is standing
currently and to plan the next steps.
The current plan is to first clean up the most important pieces (provide
packages, split V3D away, compile less different binaries) and then work on
a few upgrades. I have created two milestones for that:
https://github.com/slowmoVideo/slowmoVideo/milestones
The most important goals are for me:
- Make slowmoVideo available for video editing applications (maybe in
a simplified way with only constant slow-down factor)
- Get more Devs aboard
- Make slowmoVideo more usable and more useful
Questions
For those of you still interested in slowmoVideo:
- Do the milestones make sense for you?
- Is there an area where you would like to help?
I have spent the last few years coding with Node.js, so my C++ and GPU
knowledge is a bit rusty. I am also happy about helping hands with
packaging for different OSes.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#179>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAXFRT4MPXFI4EBU7IWDCKTRL4ER5ANCNFSM4MFNJIOA>
.
|
|
While looking into optical-flow algorithms today, I found a number of projects similar to slowmoVideo:
|
|
Am 13.04.20 um 22:52 schrieb Simon A. Eugster:
@ulatekh <https://github.com/ulatekh> Is Natron still in development? I have
to play around with it first in order to understand it.
Yes, the Natron project is alive and kicking.
They had to face a difficult situation after the initial funding did not
continue, and one of the original authors works on different (commercial)
projects meanwhile.
Its a shame that Natron is not very well known in the opensource community.
This is probably due to the fact that Natron is clearly intended for
professional compositing workflows, where floating point and extended colour
space support is a must. It started out as a clone of The Foundry's "Nuke",
which is one of the leading commercial compositing suites for sfx. You can
basically watch tutorials for Nuke and apply them mostly 1:1 to Natron.
(As far as the 2D part is concerned; Nuke also features a 3D modeller)
As far as I can see, a Natron plugin would be a rather easy fit.
Natron relies on the OFX-Plugin-Framework, see http://openeffects.org/
Natron offers a lot of frameworks for plugins without and with GUI.
The only serious problem right now is that Natron still relies on QT4,
which is about to be phased out by larger linux distributions right now.
The upstream developers are aware of the problem and promised to work on
upgrading to QT5 during this year.
Btw, I am not affiliated with the project. I just helped with creating
debian packages lately, since I am fairly fluent with DEB packaging.
https://launchpad.net/~ichthyo/+archive/ubuntu/natron
Hermann Vosseler
(aka Ichthyo)
|
|
Nice to heard some news of the project!
Having some interest in realtime video compositing, using the venerable veejay, i'll be happy having slowmovideo as a freiOr compatible plugin! As a frei0r plugin, smv would have an interface for various linux video tools. Finally, from veejay POV, frie0r plugin is not an absolute requirement as some "video effects bridges" , respecting some conditions, could be added into veejay ("plugin-packs" - gmic , ascii art ... )
Also, I could give a hand for french translation/comunication and somehow small fix/coding. My best... |
|
Hello, i mentioned slowmoVideo also for Blender VSE new speed retiming: https://devtalk.blender.org/t/vse-strip-retiming-feedback/32581/11 as opportunity to use slowmoVideo as a library: #173 |
@valgit @grota @ulatekh @mgo80 @Eliasvan @fishpepper @d-j-a-y @viktorsmari @pmo-19 @new-life @metal3d @habi @Ichthyostega You (and others) have been contributing to slowmoVideo, so I would like to give you a short update.
What is this about?
I intend to revive slowmoVideo. I somehow expected other open source slow motion tools would be around after that many years, but there are none. (Correct me if I am wrong.) So here we go again!
From my POV, I have summarised the curent status of slowmoVideo, mainly for me to get an idea of where the project is standing currently and to plan the next steps.
The current plan is to first clean up the most important pieces (provide packages, split V3D away, compile less different binaries) and then work on a few upgrades. I have created two milestones for that: https://github.com/slowmoVideo/slowmoVideo/milestones
The most important goals are for me:
Questions
For those of you still interested in slowmoVideo:
I have spent the last few years coding with Node.js, so my C++ and GPU knowledge is a bit rusty. I am also happy about helping hands with packaging for different OSes.
The text was updated successfully, but these errors were encountered: