Notable additions in this release include Kdenlive the leading video editor on Linux
[...]
Kdenlive is the most versatile non-linear video editing software available on Linux. It recently finished its incubation process to become an official KDE project and was ported to KDE Frameworks 5. The team behind this masterpiece decided that Kdenlive should be released together with KDE Applications 15.04.

(Source: https://dot.kde.org/2015/04/15/kde-applications-1504-adds-kde-telepathy-chat-and-kdenlive-video-editing)

in order to built[sic] the best video tools.

(Source: https://kdenlive.org/)

Kdenlive: Superior Video Editor ... not just for pros
[...]

Kdenlive is KDE's industrial-strength non-linear video editing application. Like the rest of KDE software, Kdenlive is feature-rich, well-integrated with the rest of the user environment. It's flexible, intuitive and a pleasure to use, even if only for YouTube shorts.

(Source: https://dot.kde.org/2012/03/20/kdenlive-superior-video-editor-not-just-pros)

I've just wasted the entire afternoon trying to put together a 20 second video. kdenlive manages to hang while opening files in about 50% of the cases, and is extremely crashy all-around. Drag a track around in a way it doesn't like? Crash! Remove effect in the wrong magical convoluted state? Crash! Adding a title "animation"? Of course it would crash. Pleasure to use? I tend to disagree.

Now, what really bugged me though is the way it simply has no method of outputting yuv444p video. Even with the lossless profiles, it uses 4:2:2 (holy moley even ffmpeg automatically selects pix_fmt=yuv444p with -qp 0 when using x264). After some googling, I came across mlt_image_format=rgb24, and adding that to a profile did not change anything, and neither did combining it with pix_fmt. Flexible? No. Intuitive? No. Most versatile? Definitely not.

And while I'm already having a go, the complete lack of a unified animation system (some transitions/effects have animations, some don't) and the lack of anything but linear interpolation between keyframes also make it not feature rich, not intuitive, and certainly not the leading video editor. (Blender does video editing too, and in this regard, it does it better.)

I could forgive all of this. I work on open-source software myself, and sometimes code bases tend to get messy and nobody has the time or motivation to work on them. But I don't claim that my application is the best or the leading, and I certainly try to eliminate some low-hanging fruit before releases.

I don't mean to be insulting, but kdenlive is a disgrace for the KDE brand. And while it's true that I could submit patches, these quotes show that the biggest problems aren't in the code, but the developers.

Actually, you do mean to be insulting.

For your first post, you sure did a good job of introducing yourself.

I'd take the time to help if not for my aversion to ingrates and Layer-8 dullards.

BTW....I seriously doubt you could submit patches. It seems all you can do is complain.

Steve Guilford wrote:Actually, you do mean to be insulting.

You're right, I actually do. But not towards any developer. I'm insulting the software, because the various KDE-hosted texts praising it to high heaven are simply wrong. Bunching it together with actually good KDE software such as Krita, Okular or Konversation is insulting to those projects.

Steve Guilford wrote:For your first post, you sure did a good job of introducing yourself.

I don't come here to socialise, I'm here to talk about KDE software.

Steve Guilford wrote:I'd take the time to help if not for my aversion to ingrates and Layer-8 dullards.

You seem to have misunderstood this as a support request, which it isn't. I'm simply sharing my experience with the software and how it differed from the "experiences" others claim to have had.

Steve Guilford wrote:BTW....I seriously doubt you could submit patches. It seems all you can do is complain.

You're awfully quick to judge a person. Only a few lines above this one, you noted how this was my first post. Yet you're jumping to conclusions about the kind of person I am, despite only having a sample size of one forum post.

Not being perfect or even not being good at all is no sin, but acting like it despite not being so is. I'm merely suggesting that people who write about kdenlive should be less heavy on the marketing BS, because there's no product being sold here.

All the nice things written about Kdenlive were about the kdelibs4 version.
15.04 is the first port to Qt5/KF5, and due to the rare availability of KF5 platform at the time of releasing 15.04, this much changing version did receive almost no testing at all.
I think we did a mistake to release Kdenlive at that time (port not mature enough).

Moreover since then we haven't had much time to fix the bugs and are spending our little energy on further refactoring the code and bringing long waited features (GPU effects, track effects, etc).
I think for the past year we are losing our good reputation, and we may deserve it. Hope we will rapidly be in good shape to help people forget that bad period.

So I am sorry for the bad moments we made you experience, and am not satisfied with the situation either.

vpinon wrote:So I am sorry for the bad moments we made you experience, and am not satisfied with the situation either.

Thank you for your honesty and transparency.

I figured KFrameworks5 had some role in some of this; especially the "hang on open" bug. I've seen this one before in kate on ArchLinux and the bug report went upstream to kate which then fixed it shortly later.

Generally though, quite a few of my gripes are caused by the software not properly abstracting some concepts. The animation is one of them (and fortunately I see this is a goal for the coming GSoC), but other concepts simply don't lend themselves well to certain workflows.

For example, having to choose video presets for projects is a hassle when all you're trying to do is edit some arbitrary clip and then render it out. All information - framerate, colour space, pixel format, resolution, etc., can be derived from said clip. The "Adjust profile to current clip" functionality appears to simply try to choose the closest matching profile. Why have profiles at all? The entire profile could be derived from the clip information if I'm not mistaken.

Another example would be project rendering. For the most part, the user has to figure out the interface to correctly map what they want ffmpeg to do. It's not like the user interface adds anything. Yet again it has some concept of profiles, but in the end you'll still be throwing command line options, but simply in a slightly different format and without the knowledge which process actually receives those options. For me, debugging why mlt_image_format=rgb24 didn't work was impossible, because it gave me no way to do so whatsoever. Generating a script instead is the closest I could get, but said script ends up just launching the opaque kdenlive_render process.

There were earlier attempts at using kdenlive by me where I gave up and simply used ffmpeg with a filter graph instead because doing multimedia editing through a text interface ended up being simpler than doing the same operation through kdenlive, a graphical application.

The general lack of a concept for objects also seems to make kdenlive scale badly for cases where more than one visible element should be doing something. If you want to overlay five different images (e.g. arrows pointing at something) in the video, you'll end up with 6 video timelines, which you'll then have sticking around throughout the project. Text being seen as "Title Clips" is another shortcut kdenlive seems to have taken where the intended use for the kdenlive software is simply slapping together some videoclips and having a title fade in at the beginning.

The abstractions are simply not general enough to make them greater than the sum of their parts. I could do title clips with a generic text object and a generic shape object by putting text onto a rectangle. But I can't do the reverse with a title clip function.

This is why statements such as "industrial-strength", "flexible" and "not just for pros" are simply wrong: The concepts the software exposes to the user are quite the opposite of all of those things. They're very much aimed at the Windows Movie Maker niche.

I'd love for KDE to have a proper video editor which provides simple but reusable and solid functions which can be applied for a variety of use-cases. Sadly, kdenlive in its current (and past) states does not provide this at all.

hello fratti

You are the only one with such difficulties ... try to go to training about computer for skill progress ... and let serious people work and use this software ..

Thnaks

fratti wrote:I'd love for KDE to have a proper video editor which provides simple but reusable and solid functions which can be applied for a variety of use-cases. Sadly, kdenlive in its current (and past) states does not provide this at all.

Which FOSS video editor would you recommend we move to?

Which past versions of Kdenlive have you used?

What version are you trying to run now?

fratti wrote: Why have profiles at all? The entire profile could be derived from the clip information if I'm not mistaken.

Because you might want to render your project into a format different from the source format. You use the project profile to let kdenlive know what frame rate and image size you are planning on rendering to.

fratti wrote: doing multimedia editing through a text interface ended up being simpler than doing the same operation through kdenlive, a graphical application.

That my friend is total BS. I challenge you to a time trial - you can use a text editor and ffmpeg to edit a video and I will use Kdenlive. Lets say 10 min video, 50 edit points. 10 with transitions. 3 titles and 30 seconds of picture in picture. I will do it 10-50 times faster than you.

While the OP was a bit rude, he was clearly venting his frustration at using current Kdenlive. And I have to agree with him. This went from a fairly solid and intuitive program to one that crashes constantly, half the effects don't work, and most common features seem to be broken in some way.

Lets just say that I'm glad my day job doesn't depend on kdenlive. I suppose i'll let this project stew a little bit longer before I come back to trying to use it on a regular basis.

best thing to do is to use Kdenlive 0.9.10 which is best stable version, less bugs, more complete features working

the new versions 15.x are full of bugs, missing features, because the software is in a transition moment

I believe the developing team will correct and improve the new versions, so it is time to wait and keep using the 0.9.10 version

I'm running a couple week old version from git and it is very stable at least as far as crashes and other problems related to adding/moving clips in the timeline.

From an argumentative standpoint the OP didn't suggest alternative software. I started editing video on Linux with Cinelerra... the only thing I miss is background rendering of effects. Kdenlive is superior in my mind in every other aspect.

Did several projects in Kdenlive 15.04.x and 15.08.x. In the past weeks Kdenlive has come around nicely and works quite satisfying and stable for me. So the next public releases will be ready for production, especially 15.08.3 and 15.12.0.

zekthedeadcow wrote:I started editing video on Linux with Cinelerra... the only thing I miss is background rendering of effects.

http://heroinewarrior.com/cinelerra/cin ... -RENDERING
Wow, that sounds like quite a feature.

As for the topic: Kdenlive is great My main issue is that it doesn't run natively on Windows and it's relatively hard to set up on Mac, which unfortunately slows down its popularity growth massively.