47 karma · joined December 24, 2020
It's painful (for me, that is :D), but I know it's the right thing to do.
Hehe, I remember those days :D Yes, code has gotten A LOT worse, and right now, whenever I pick a new library, I expect it to have bugs.
A lot of people will start coding and bring in a TON of open-source libs, each with their dependencies. And then when something goes wrong, spend an insane amount of time just figuring out who's to blame. Or, just fix the symptom and continue.
Clearly, the above won't work in the long run. But who cares about that, at the current pace of business, it'll be someone else's problem in a few months? They'll get promoted or switch jobs :D
I'm on the opposite side of the spectrum: I choose VERY FEW libraries, go for mature ones, and do quite a bit of testing. For pretty much each library, I simply add an extra layer on top of it, so I don't need to deal directly with it, just in case I may want to later switch to another implementation, or perhaps do some workaround until the issue gets fixed.
And this brings me to Microsoft: the code since Windows 8 (2012) has been beyond horrible, and their direction has been, lets say, "flexible" (to read: as the wind blows). I've discovered so many CRITICAL bugs, I've had to struggle with depression for months.
> I want to rely on other code, but more and more I find that it’s safer to just assume it’ll be broken from the start.
Yes, that is definitely a safe bet. The more you rely on 3rd party libs, the more you increase the risk of (critical) bugs. The more libs you use, the higher the risk. It's a really dim picture -- but something we unfortunately need to deal with. It's the new reality, and we just need to adapt :D
As a side note, I'm working on a video editor, and what I'm working on now, is the ability to add shapes (amongst other things :D) and have an insanely easy way to animate them.
It's really hard for a new player to come into this crowded market, but I'm sure it will happen sometimes in the coming years. Zucky has brought nothing but toxicity, and hopefully, that will cost him in the long run.
What I'd like is something simple that would allow me to stay in touch with my friends, I don't want any reminders about "X posted Y", or "You might know Z" or whatever.
Thanks again!
Here's what I have: you can hover over any video track and I instantly show you what's there. So you can decide what's in and what's out.
With this, assuming there's enough interest in something like this, I can optimize it further and do it like in Vegas with a few quirks. Like, you would select several tracks and create a multi-track. When hovering the multi-track, you can choose to view all at once, or a single video and have a hotkey to switch between videos, and another hotkey to select that video as "current video".
I have crop, merge, insert transitions/texts.
I will integrate blending. That's pretty much on top of my todo list (after some stability fixes :D) It will take some time, since I will do this as part of a larger module that will allow advanced masking and such :)
The module I'm talking about will also allow some pretty cool text effects with close to no work on your part at all :) But it's quite a bit of work on my part :P
I remember looking into this a while ago in quite a bit of detail: https://www.engadget.com/2010-05-04-know-your-rights-h-264-p... (there's plenty of other links to this).
Long story short, MPEG LA owns some patents that relate to h264/h265, and you need to pay them A LOT of $ in order to use these codecs. Using ffmpeg will not devoid you of that.
Also, even somewhere on the ffmpeg page, when I last read it, it will say something more in the lines of "we're not lawyers, so take this with a grain of salt".
> a number of programs will externally fetch FFmpeg and just interop with the libraries or FFmpeg app.
As far as I know unless they are actually free, they could end up being sued by MPEG LA. I remember at some point that's what I wanted to do - just download and use ffmpeg as an exe. But if I read stuff correctly, I do think I'm not allowed to do that - in other words you (as an application) are allowed to open a web browser to the link, but it's the user who needs to do the actual download and install.
----------
So all in all, after doing quite a bit of research, seeing how vague the licensing from MPEGLA is and all that, I started looking for alternatives. I found one (from Microsoft), and that's what I'm using. At this point, it satisfies most of my needs (even though I had to do lot of optimizations on top of it) -- in case this won't hold true in the future, I may revisit the whole ffmpeg mambo-jambo :D
So basically where it counts (x264 and x265), I can't.
LATER EDIT:
I remember looking into this a while ago in quite a bit of detail: https://www.engadget.com/2010-05-04-know-your-rights-h-264-p... (there's plenty of other links to this).
Long story short, MPEG LA owns some patents that relate to h264/h265, and you need to pay them A LOT of $ in order to use these codecs. Using ffmpeg will not devoid you of that.
Also, even somewhere on the ffmpeg page, when I last read it, it will say something more in the lines of "we're not lawyers, so take this with a grain of salt".
----------
So all in all, after doing quite a bit of research, seeing how vague the licensing from MPEGLA is and all that, I started looking for alternatives. I found one (from Microsoft), and that's what I'm using. At this point, it satisfies most of my needs (even though I had to do lot of optimizations on top of it) -- in case this won't hold true in the future, I may revisit the whole ffmpeg mambo-jambo :D
Sorry for my poor choice of words. That's what I meant. Only the ratio is fixed (clearly, you can change that too, but very likely you would need to readjust everything).
> Too bad about ffmpeg, I love it so much :)
Trust me, I know :( Everything would have been orders of magnitude easier for me, had I been able to use ffmpeg.
Got it, so if I understand correctly, that means reasonable defaults.
> Maybe you can check out what is easy to re-mux with ffmpeg if you use that. Often it's just the audio that needs to be re-encoded.
Unfortunately, I can't use ffmpeg in a commercial app. But yeah, I'll figure out something :)
By output format, do you mean, .mp4/.avi and the like? If so, clearly on export is where it makes sense (which is where I have it)
> I use found-footage a lot and web video comes in formats that make e.g. premiere struggle
I see. If you download it as .mp4, it's easy. Other formats (.mkv, for instance) are much harder to interpret. I will try to strike a balance between what people want and how easy I can support them ;)
> Since you are aiming at more casual users, consider intuitive intelligent audio controls.
Agreed. I have a friend that is pretty much an expert in Logic, and I will ask him for more input. As a side-note, I think the UI will be the "easy part" of the deal :D The implementation - that may be insanely hard :D So I will have to strike the right balance :D
> Beginner and often intermediate users often do not understand the importance of good audio in a video. When they do it still takes a steep learning curve to understand the complexity and nuances of even the basics.
100% agree :)
> Most video editors don’t help users with audio
Agreed here as well :D One thing I'm kinda' proud of is you can go anywhere in the project , go to an audio track and space will only play that. Or shift-space will play all tracks (no video). And one can create music keyframes (that move along with the audio if you move it), and you can snap video to them.
> Templated signal chains for individual audio tracks that include simple things such as gate, subtractive eq, compression, additive-eq
I know a bit of the theory here, but I'll need to did lot deeper into this. At some point, I believe I'll definitely have this.
> Level suggestions for tracks, based upon if they are dialog, music-bed, room tone etc
This sounds complicated, but will delve into it at a later time.
> Not specifically audio, but being able to output multiple different renders depending on target website with a single click. (Ideal container, codec, resolutions, ratios, LUFS, etc).
Agreed, this will definitely be there :)
> This. Please optimize for casual users and due to Covid-19 social distancing more people are forced to making and editing their own video.
Agreed! I've done my best to do just that - but clearly, you will be the judge of that :)
> 1)Try to enable targeted video size feature for the generated/exported video. Also design a simple UI menu to automatically transcode and resize the generated video according to the chosen video format.
Agreed! I would say I've done achieved this - basically, when you click 'Export', you can set the FPS, and the video format (low, HD, Full HD, 4k - coming soon).
There's also an option for video ratio (such as 16:9, 4:3, and so on) - available via the project's settings. Hopefully people won't think it's buried too deep in the settings menu :)
> 2) Make it easy for the authoring of the sub-title by including dead simple and easy to use vector graphic editor built-in and integrated inside the video editor (not third party full blown vector editor).
Added to my todo list. I want to do it similar to what Adobe has - which will be a bit of work :D But yeah, I'll have it ;)
I may take another stab at it in the future. One of the (many) problems I've had with it is that quite a few mkv files have improper headers (thus, you can't know, for instance, the frame size and/or video length in seconds). Some simply can't be read at all - I just get black frames.
Note that I'm using Windows API to parse the .mkv files (I can't use ffmpeg from a commercial app). Some may be bugs in how Windows handles .mkv files (and very unlikely to be solved in the near future) and some may be bugs in how .mkv files have been written (and ffmpeg, since it's so mature, may very well ignore "bad" frames and such).
At the time I analyzed this, I considered this more trouble than it's worth. Perhaps I will reconsider this in the future, with the addendum that if I get an error or so, I will disregard the file.
> In my view, there is scope for innovation here by making it easier over how most software applications do this.
Agreed!
> Here's an example of a very different workflow: https://www.descript.com/video-editing
Yeah, that is kinda' very very cool :D
I know they used to be audio only a few months back. Apparently they moved into video as well. Interesting, I will certainly take another look at them.
I would say I have the trimming should be working quite nicely ;) I do it similar to Final Cut. I will take a look at handbrake just to make sure. I assume you mean handbrake.fr, yes?
About conversion - right now, I save videos as .mp4 h264 format, and can easily add h265 in the future. This should cover 99% of usecases (IMHO :D) -- and will allow presets for youtube/fb/insta/twitter and the like.
That would put the responsibility on the user to remember all that. If I decide to do something like that, I'll need to save the original myself, and allow the user to go back to it.
But yeah, now I get the appeal of having something like this. I've added it to my todo list -- if enough people want this, I can certainly implement it (it would disrupt quite a bit of my internal workflow, since I view "videos" as constant, and optimize them as such. so I would need to find a workaround for editing in place, but it's certainly doable).
> GUI's. Plain and simple. It's gotta be as absurdly easy to use like iMovie but at least have the basic capabilities of a Sony Vegas Home version (the $30 kind).
Agreed! I do hope I've done this, but yeah, you would need to try it and let me know :D I'm sure it can always be made easier, and I will certainly constantly improve on it!
> Cutting, scrubbing, aligning, have multiple tracks, and for gods sake a simple and easy way to understand what I'm encoding/rendering (see Shotcut).
Got it! Cutting, aligning, multiple tracks - it's there. Scrubbing - it's on my todo list (this is really difficult to implement! :)) I do think I have a simple way to show what you're rendering (once you press Export), but I'll take another look at Shotcut.
> KDENLive and Shotcut are two OSS video editors I've used simply because they're open source, but by god do they still have a long ways to go in terms of stability and ease of use.
Hehe, I just took a look at KDENLive - it's decent, but the UI is pretty horrible :D So I guess I'm in good shape :D
> For me, 99/100 times all I want to do is trim, split, have multiple tracks, have some fades, add text (animation or not), and lastly read and output various codecs.
Yup, makes sense! I do have a bit to go on this front, but yeah - all that I don't yet have, added to my todo list.
> Update: I had a thought right after posting. Basically if you could take Shotcut, but make it look and feel more like Sony Vegas, I think you hit the nail on the casual/semi-pro you're looking for.
Got it! Not a huge fan of Sony Vegas, but I'll take another look. My biggest inspiration so far has been Final Cut.
Here's my answer to another user:
I do want to go multi-platform ASAP. However, because of the sheer complexity of video editing - and since I focused on speed + ease of use from the get go, I chose to go with a single platform (Windows) for starters.
Once I am close to feature complete (as in, be able to properly launch and be rock stable), I can think of porting to Mac (first), and who knows, maybe Linux.
I did think about this from the get-go, so I did isolate the complicated code in a specific module. So theoretically, it should not be very easy (read = 2-3 months :D) to port. But when reality will hit, we'll see :D
> choosing output format at the start, not just at export
Got it, added to my todo list.
> file formats and conversion. Maybe auto-conversion could happen in the background while I cut with low-res?
I do support the popular formats for now. What do you mean by "conversion"? In my mind, that clearly happens when you import the file into the project.
At export, I do go through the original videos, to get the highest res for each video frame. If this is what you mean by auto-conversion, that is possible, but not sure how advantageous it would be - it takes a lot of CPU + GPU (even if in the background) and may disrupt your editing. I will dig deeper into this.
At this time, I have some pretty cool text effects. But yes, the text support is not as good as I'd like. It's something I'm working on fixing. So yeah, I'll definitely have that :)
Hehe ;) I've actually had to create a PRORES-lookalike for Windows, since normal play would be waaay too slow. I can work on some improvements at a later time, but at this time, I'm more than happy with the results. You can play a video at any speed, or backwards with no lag :P
Interesting, I'm actually curious of some numbers. This probably makes sense for insanely simple editing, but very likely if you do any edit, you'll always need to reencode (since you do some processing).
> Typical video software has it like a build process. So there are these project files maintaining the list of all resources, edits performed, settings, etc. Editing a Word document needs none of it. For many casual edits, it would be great to just open a video file, make edits (e.g., rotate the video, remove some segments, increase brightness, etc.) and click save (which overwrites the original file), and all done. There's no "project", no auto-save, etc.
Interesting, I do understand the allegory, but I'm not sure this is such a good idea. Just imagine this - you put some text on top of your video. In 5 minutes you want to move the text a bit lower. Bam, you wouldn't be able to do that.