Tools and other software Media-Buddy - the MKV Buddy successor. (14 Viewers)

[Lehmden]

Hi.
It's done! Today I was able to complete a full workflow with Media-Buddy V2 for the first time without constantly having to type in the CLI. The GUI and the serial wizard are now fully functional.

The settings you make here will be applied. Everything you've configured will be saved so it's available the next time you start the program. Then simply click "Let's go," just like in the "old" Media Buddy, and the magic begins:

Now, the videos in the input folder are processed according to the specific needs of each video. This means they are either re-encoded or remuxed, the audio is handled as configured, the track selection for audio and subtitles is taken into account, and after all videos have been processed, the scraper is called, the necessary fanart is loaded, the .nfo file is generated, the files are renamed, and the season folders are created if needed. Finally, the Matroska tags are set, provided the final file is indeed an .mkv file. It's the same tried-and-tested process as in version 1. A dialog box then appears:

The dialog will close automatically after a few seconds, unless you close it yourself beforehand.

Before I implement post-processing, I'll first work on the other assistants. The foundation is now in place and only needs minor adjustments for music videos and feature films. However, the automatic generation of fan art for "other" videos, a unique feature of Media-Buddy, will be significantly more complicated to implement, as there's currently nothing available for that.

In any case, I can now use the new version in everyday life and test it extensively.

 

[Lehmden]

Hi.

The GUI has been further refined; it looks even better now and takes up less space.

In addition, it is now fully localized.

To this end, various (though not yet all) of the global settings have been implemented in the GUI:

And the movie wizard is also fully functional. For some reason, I don’t have a screenshot of it handy, but it looks virtually identical to the one for TV series. Since the TV series assistant is currently running, I can’t provide one right now. I’ll add it later when I get the chance.

Things are progressing in leaps and bounds. Much more smoothly and quickly than I had anticipated. Not only am I gradually bringing back all the features from the "old" Media-Buddy, but I’m also adding entirely new ones. Furthermore, the existing features have almost all seen technical improvements, in some cases, quite significant ones. It’s a huge help that I had already conceived, implemented, and, where necessary, refined the core concepts for V1. Because of that, I know exactly how everything is supposed to work. Next up is the music video assistant; this didn’t exist in the "old" Media-Buddy, but it’s definitely part of the "new" one.

 

[Lehmden]

Hi.

Another milestone was reached today. I actually used Media-Buddy on Linux Mint 22.3 today.

As proof:

PySide6 is truly fantastic. The GUI runs just as smoothly and effectively on Linux as it does on Windows, without requiring any adjustments whatsoever. All GUI elements function identically across both operating systems.

However—as is often the case with Linux—getting it up and running was a real struggle. To begin with, PySide6 couldn't be installed in the "standard" way; I absolutely had to set up a virtual environment (venv). Then, essential system components—which really should have been present in a Linux environment—were missing, preventing Media-Buddy from launching. Furthermore, MKVToolnix was installed as a Flatpak, which is completely useless for Media-Buddy. So, it was back to square one, involving endless, tedious typing in the shell... To top it all off, the package manager installed an ancient version of FFmpeg (v6) as the "current" release—a version that was essentially useless. Outside the Linux world, we’re already at v9... And the headaches didn't stop there. In reality, the only way to get a usable version of FFmpeg on Linux is to build and compile it yourself. Every "pre-built" version available is so heavily stripped down that it’s entirely unfit for purpose—or, at the very least, completely unsuitable for Media-Buddy. Just when I was on the verge of abandoning the whole cross-platform endeavor, I stumbled upon a working solution in the most unexpected place: Jellyfin. They provide an FFmpeg build that the Jellyfin developers have carefully optimized for specific requirements—such as support for QSV, NVENC, and AMF.

All these QSV encoders can be used within this version of Jellyfin, provided the underlying hardware supports them. On my Core i5, for instance, AV1 and VP9 are not available, as they are not implemented in the GPU. Furthermore, this version is installed in such an unconventional location that Media-Buddy’s standard tool-detection feature—which otherwise works really well—could not locate or integrate it. Nevertheless, with this specific FFmpeg version, I was successfully able to use Media-Buddy V2 to transcode videos via QSV on Linux.

The only part that went completely smoothly was the installation and integration of dovi_tool and its "sibling," hdr10plus_tool. I simply downloaded the compiled binaries from GitHub and placed them in Media-Buddy’s "tools" subfolder; the tool-detection feature found and integrated them immediately.

I reckon I’ll soon be at a point where I can bring some beta testers—or perhaps more accurately, "pre-alpha" testers—on board. I’m particularly interested (though not exclusively) in how Media-Buddy V2 performs on hardware that I don’t personally own and therefore cannot test myself. If you’re interested, feel free to get in touch!

 

[Lehmden]

Hi.

Today was a very productive, yet also exhausting day. I spent a full 12 hours working on Media-Buddy alongside the AI. The results, however, are well worth it. Since Thursday’s successful Linux test, all the video-editing modules that existed in the old Media-Buddy have been implemented and are now fully functional in the "new" version.

The first "metadata" section has also been integrated into the GUI; it no longer needs to be accessed via the CLI if one wishes to use it independently. The other three sections will be integrated into the GUI shortly—likely tomorrow.

The assistants for series, movies, music videos, and "other" videos are now fully functional, as are the features for recoding, remuxing, normalization, and video merging. As expected, the section for "other" videos proved the most demanding, as it required building something entirely new from the ground up. Just like in V2, Media-Buddy can automatically generate posters and fan art for these "other" videos—a process that now even works for portrait-oriented videos.

All in all, video editing—in all its facets—has been almost entirely carried over from V1 to V2, while also being improved and modernized. And all of this remains completely platform-independent.

 

[Lehmden]

Hi.

Just so you don't start thinking I’ve been lazy... Not only have the remaining "metadata" sections—the tools that allow you to trigger scrapers without having to edit the actual video files—been implemented and made fully functional, but a "post-processing" feature for TV series has now been added as well. This serves as a quality control step for the processed videos, and also allows for subsequent adjustments to metadata and/or artwork that were retrieved or generated during the batch run. Here are a few screenshots showing the various sections. First up is the TV series view, which lists all the series found in the target folder. From here, you can replace or modify both the metadata and the series-specific artwork.

If you click on "Edit episodes," you end up here:

Here, you can replace the season posters/banners, edit episode metadata, and change the episode thumbnail. Just like with the other graphics, you can replace it simply by clicking on it. Alternatively, you can click "Refresh thumbnails." Here is what happens then:

Anyone who knows their stuff will see that the screenshots don’t match the episode. This is simply because I used an arbitrary short sample for testing purposes. In any case, I selected the screenshot with the blue border. If you apply this, the existing thumbnail will be replaced by this screenshot.

Once applied, the selected screenshot is used as the new episode thumbnail.

The final screenshot shows Media-Buddy V2 in action, playing back an episode as a test. This allows you to verify that the video plays correctly and that the "right" audio track(s) have been preserved. Once everything looks good and all episodes have been processed properly, you can select "Delete Originals" from the "File" menu. After a confirmation prompt, the original video files will be deleted—since you now have the processed versions, the originals are no longer needed. Naturally, I would never automate a process like this; however, when triggered intentionally and accompanied by an additional safety confirmation, it becomes a very useful feature.

What remains are the "post-processing" tasks for feature films, music videos, and other types of video content—that’s next on the agenda. Once that’s done, the entire video component of the new Media-Buddy will be complete. After that, I’ll move on to the processing features for music and photos. Finally, I’ll address the many small utility tools that had "quietly" found their way into the "old" version of Media-Buddy.

 

[Lehmden]

Hi.

Post-production for feature films is now working, too.

As in V1, there are three "subpages" here that you can toggle between below the poster. The "metadata" page is shown at the top.

And here is the "fan art" page.

Finally, let’s look at the film collection page. As you can see, some graphics are still missing from this movie set. You can add them by simply clicking on the "graphic missing" indicator.

To show you that the AI certainly doesn’t get everything right on the first try—and that it requires careful supervision—here are a few "failed attempts."

A very sensible arrangement, don't you think?

One of the many attempts during which I had to explain to the AI, in detail and repeatedly, what I wanted.

That was about 10 versions later, without it actually looking any better. Regardless, we managed to get it right in the end, as you can see further up.

And despite everything, it would have taken me much longer without the help of AI. The AI is genuinely good at the actual programming logic—I stand no chance of keeping up with it there. At the very least, I’m still able to follow and understand most of the code. However, when it comes to the creativity involved in designing a GUI, the AI is still very much lacking; that’s something I’m much better at. So, even though the AI was a great help, it remains, without a doubt, my program—not an AI product.

 

[Lehmden]

Hi.

For the past few days, I’ve been working on catching up with music video processing. Since there was no existing template from the "old" Media-Buddy to build upon, everything had to be developed and researched from scratch—which is why the scraping process is taking longer than expected. Scraping music videos is a fairly time-consuming task, as it requires querying multiple databases (MusicBrainz, TheAudioDB, Last.FM, and fanart.tv, along with potential online translation services) to ensure everything works correctly. Naturally, that takes some time. In any case, it is now possible to review music videos and make corrections whenever necessary.

You can easily swap out the graphics by clicking on them. If you want—or need—to create a cover image from the video itself, you can use the "Select cover" button.

This opens a window displaying up to 16 still frames from the video, which can be used to create a square cover image. You can adjust the position and size of the crop area. Once confirmed, a full-resolution still frame is cropped according to your selection and saved as {videoname}-poster.jpg.

Additionally, there is a cool new feature currently undergoing active testing. When dealing with series like Ridley—where a German broadcaster (such as the channel "One") might split the original four 90-minute episodes into eight 45-minute segments—the "Series Wizard" can now handle this automatically. All that is required is an edit list (cut list) for each of the segments. These lists, in .csv format, can be created using tools like LosslessCut—a free, open-source, cross-platform video editing program. While Media-Buddy has long supported edit lists, a new capability in V2 is the "Series Wizard's" ability to recognize "video groups," which are then merged into a single video during processing. To achieve this, in addition to providing an edit list (which allows for the clean removal of, for example, the end credits of the first segment and the opening credits of the second to ensure a seamless transition), you simply need to ensure the filenames follow a specific convention: the segments must share the same name, with one distinct difference. For instance, if S01E01 is split across two files, the first should be named S01E01a and the second S01E01b (this also works with formats like 1x01a and 1x01b). The system will only identify and process these files as a "video group" if both the edit lists for all segments and the specific "a, b, c..." naming convention are present. Regardless of this, edit lists will continue to be taken into account, provided they are present. There’s no need to worry about this, as such edit lists don’t just appear by accident; anyone who includes one clearly intends for it to be used. Media-Buddy already offers an incredible number of configuration options—there’s no need for this to be configurable as well. If you don’t want any edits, simply don’t include an edit list with your video.

Now, I still need to implement post-processing for "miscellaneous" videos and go through the GUI with a fine-tooth comb to root out any remaining bugs, glitches, or missing localizations. Once that’s done, the video-editing component will be fully complete, and I can move on to the music section.

 

[Lehmden]

Hi.

It’s done—the complete video processing workflow, covering all areas, is now fully functional in the new Media-Buddy. Post-processing for other types of videos is also working now.

That was the AI’s first attempt, after I had previously written an entire book detailing exactly how it was supposed to look.

After several more attempts, things ended up looking like this. You can edit the "metadata" here, which now includes the option to assign a UniqueID. This allows you to define your own custom types (like "test" in this example) or even provide a "fake" TMDB or TVDB ID—useful if you’d prefer to categorize the video under "Movies" or "TV Shows." The posters and fan art displayed are generated automatically, though the design is fully configurable. You can also change them manually: either by downloading a suitable image from the web and swapping it in, or by triggering the generation of new posters/fan art while specifying exactly which screenshots should be used.

And, just for "fun"—here’s another failed attempt.

The wonderfully distorted graphics are actually in portrait orientation, as this is a mobile phone video.

Here you can see the result. For portrait-oriented videos, there are 8x2 preview images from which you can select the ones to be used for the poster.

And this is how it’s laid out for a "standard" landscape-format video—specifically, a 4x4 grid.

A new feature in V2 is the ability to manually select the images used for fan art. You can also choose to deviate from the global configuration settings, if you wish.

As you can see, the same stills used for the poster are employed here. You select which image should serve as the background. You also have the option to forgo a collage entirely; in this case, the chosen background image is used to fill the entire frame. Naturally, this only makes sense for videos in HD quality or higher, as the stills would otherwise likely appear too blurry. If, however, you do want a collage, you simply select 2 to 6 of the available images. You cannot choose just one image, nor can you select more than six. The number of design options is limited—otherwise, it would be necessary to develop an almost infinite number of algorithms.

If you select just one background, this is what it looks like. You’ll likely recognize what I’ve chosen here from the selection window shown above.

With this, all the video editing options from the old Media-Buddy have been migrated to the new version. Furthermore, there are many new features and options that simply didn’t exist in the previous program—ranging from the secure preservation of Dolby Vision videos to a complete end-to-end workflow for editing music videos. There are countless small improvements, along with several more significant additions—all of which make sense and are now technically feasible.

Since I had some extra time, I began going through the entire GUI, window by window, to fix minor bugs and address missing translations; I’m about halfway through that process. On top of that, there are two minor technical adjustments that came to light today but didn’t fit into the current workflow—though these shouldn’t take too long to resolve. Even though three of the four major modules are still pending, I’ve likely completed about half of the overall work. The remaining areas—music, photos, and utilities—are quite distinct from the video component, but they are also far less extensive.