Action Sakuga 作画 MAD – Editing Notes –

I finally finished the Action アクション Sakuga 作画 MAD. It was a bit of a nightmare to make in places. But overall I’m very happy with it. I made this video to celebrate getting 100,000 subscribers on my YouTube channel. In this article I’ll talk about how I put it together and how make my videos in general.

I’d been planning this MAD since early last year. I’m always collecting clips for video I want to make in future. I go through Sakugabooru/Twitter regularly (daily if I can) and watch all the new clips. I download clips I like and catalogue them with Adobe Bridge on my PC to make them easy to search. For example, I add info like where I found the clip, who the animator was, tags, a description of what happens in the clip, so I can find them easily in a search. I start editing a video once I’ve found the right songs for them.

My Adobe Bridge set up, Metadata I’ve written to the file is on the right

I started making individual videos for 2nd Hand to Hand Fighting and Swordplay Sakuga MADs last year. But I didn’t feel like I had enough clips to justify making whole videos about those topics. I’ve preferred to make longer MADs for a while now. I didn’t feel like it’d be right to upload a 3 or 4 minute MAD if my uploads are usually ~20 minutes.

I thought if I made an Action MAD I could put in Hand to Hand Fighting/Swordplay/Gunplay, but also put in action clips that don’t fit neatly into those categories. I have tonnes of those less specific action clips in my collection so thought it’d be a good chance to use them.

I had an idea of what kinds of songs I would use for the middle 3 MADs. The songs follow the same theme as the 1st MAD in their series. Both Hand to Hand MADs use Cowboy Bebop music, both Gunplay MADs use brass heavy jazz fusion and both Swordplay MADs use songs by Nujabes. So there’s a connection between them.

I find a lot of music I listen to through Spotify’s Discovery playlists. But I also browse Soundcloud and get reccommendations from friends/family/places like Anthony Fantano videos

The first time I heard Sign by Roosevelt I knew i really wanted to edit to it. It feels really nostalgic and atmospheric but I also love that it gets intense as it goes on. I thought it could work well as an ending song in one of my MADs because it has that “end credits of a movie” kind of feel at the beginning (before it gets more dance hall). I was originally going to use it for the last part of a Fire MAD, but thought it worked better here.

I wanted something Future Funk for the 1st song since that’s what I’ve been usually doing. I like editing to Future Funk. Something Plastic Love related if I could. That way it would be kind of a reference to the Effect Sakuga MAD, which is a similar kind of MAD that uses a song that samples Plastic Love. Plastic Love is also the default Future Funk sample so I thought it would be appropriate for a general action MAD.

I looked through tonnes of playlists, listened to 100s of Future Funk songs but couldn’t find what I was after. So I basically said fuck it and just used a remix of Plastic Love. Tara’s bootleg of Plastic Love wasn’t as fast or long as I wanted it to be. But I liked the mix so I spent a few weeks audio editing to get a version of the song I liked. I think the end result sounds a bit muddy because of all the post processing the song’s been through. I paid a music producer to clean it up, but there wasn’t much he could do. In the end I went with the version I originally made. I enjoyed editing to it. I wish it sounded a bit cleaner and I might redo it sometime in future. But I think it works well in the video.

I pitch shifted and sped up the remix in Audacity then edited and EQ’d it in Adobe Premiere. I used Adobe Audition to test out different EQ settings.

The video editing took about 2 months. I worked with a little over 1000 clips. The intermissions were an idea that kind of came from my last video, the Liquid 液体 Sakuga 作画 MAD. I included some audio from an Utena episode in the MAD since I liked the sound design a lot. I thought it could be cool to make more intermissions like this between sections. It’s a way I could include more nice animation that didn’t fit neatly into the AMV parts. The SFX from the intermissions are from SoundSnap.

It’s hard to describe my editing process but I try to make it feel like the clips are flowing into each other. I try to have a balance between editing songs to the beat, letting them play out to see the animation and keeping the pacing and energy appropriate. There’s a lot of experimenting to see what feels right. I made around 20 different verisions of the opening for example.

I write down ideas I have for how I could edit parts of the video in notepads and test the ideas out. I usually create key sections that I really like first and edit around those. For example I knew I wanted to have the first beat drop in Plastic Love end with a close up of a character’s eye, so I edited around that. It’s then a matter of constantly rewatching the video to see what feels good or what needs work, taking notes, moving things forward and back a few frames, trying out different parts of a clip, watching my library of clips again and searching for new clips as I edit.

I really wanted to have sparks of lightning flying towards the screen after the Star Driver The Movie fight scene since I thought that would be a good follow up. I didn’t have a clip like that initially so I searched the lightning and sparks tags in Sakugabooru until I found the To Be Heroine clip that I thought worked well. I ended up using that in the final edit.

After I finished the semi-final edit it took about a week to replace the 480p Sakugabooru clips with Blu-Ray quality footage. There were 228 clips I had to replace with HQ versions.

I use WPS Office, a free Word Proccessing suite, to keep track of whether or not I’ve replaced HQ footage

Sometimes the timing of the animation in the Sakugabooru clip is different to the timing in the HQ clip. There’s some problem solving to be done when trying to fix that. If the new timing doesn’t feel right I ocassionally change the timing of the HQ clip to match the timing of the Sakugabooru clip. But more often I’ll screenshot the final or beggining frame of a clip and extend it to fill any blank spaces in the timeline left by the timing difference.

I very rarely, almost never, change the playback rate (the speed) of a clip of animation to make it fit. I’d much rather do anything else as changing the playback rate messes with with the integrity of the animation. None of the clips in the Action Sakuga MAD have had their playback speed changed.

Referencing and subtitling for the MAD took about 3-4 days. When I download clips from Sakugabooru I change the name of the file to the video ID. That way all I have to do is copy the name of the clip into the Sakugabooru URL to find the source information. I wrote subtitles using the subtitle feature in Adobe Premiere. You can add text formatting like italics in there as well. To align the subitles to the bottom left of the video and export it in a format that YouTube will recognise the formatting in (the EBU STL format is best for keeping the formatting) I used the open source program Subtitle Edit.

I write the Reference List at the same time as the subtitles so I can see the timing of the cuts

I had the idea to create a document that would let people know where they could watch the sources I used in the video since I get asked that a lot. I thought it would be nice to do in celebration of getting 100,000 subcribers. So I made the Action Sakuga MAD Companion Book in Adobe InDesign over about a week and a half. I basically had a look at some templates to get some ideas and adopted/altered design elements I liked.

My main gripe with the video is the audio quality of the first song. I didn’t see problems with it when I was editing. But by the time I finished editing the video side of things I heard its muddieness in places. It’s artifacts of compression. It was too late to redo when I heard the muddieness as I lost the project files for the song. I was able to EQ it to make parts of it pop out more at least. I think it works well, but I’ll focus on reducing audio quality loss in future videos.

I rendered roughly 32 final edits, the edit on YouTube is the 32-34th version I rendered.

I hope you enjoyed the article and the MAD. Let me know in the comments if you have any questions about my editing process.