Bad Apple but AI Generated (used MMD as reference for it)

635,227
0
Published 2023-03-08
!! Go watch the Original MMD and show support !!
YT:    • MMD6 - Bad Apple (3D animation version)  
Nicovideo: www.nicovideo.jp/watch/sm13595028

Hopefully I can make a better version using the black and white original at some point. This version used the MMD which Is a large part of why its consistent.
Never expected this to get more then 5 views, just thought it looked interesting. Thank you for watching

Made using Visions of Chaos Stable diffusion Web UI ebsynth extension and controlNet extension
SD Model: Anything v4.5
Extra program: ebsynth

Other versions:
Reconstructed from the music:    • Bad Apple but Reconstructed from the ...  
Papercut:    • Bad Apple but Paper Cutout (AI Genera...  
Fruit:    • Bad Apple but it's Sliced Fruit  (AI ...  
Chibi:    • Bad Apple but CHIBI  

Unreal engine thing I made:    • I'm making a Deep Sea Horror Game wit...  

AI Visualization of Everywhere At The End Of Time:    • AI Visualization of Everywhere At The...  


Once I get monetization, all profits will go directly into helping my girlfriend get out of where she lives and into a safer environment. It’s not much but I hope it can help her.
Thank you so much 💕


Twitter: twitter.com/manshoety


VODS channel:    / @manshoetyvods7086  

My Discord tag is manshoety


I DO NOT OWN ANY MUSIC IN
THIS VIDEO IN ANY WAY, SHAPE,
OR FORM!

#stablediffusion #stablediffusionanimation #aivideo #aianimation #badapple #touhou

All Comments (21)
  • @manshoety
    I just want to mention that this was made purely to see if stable Diffusion could make it look more like anime instead of CGI. In no way is this an attempt to make it "better". I love both the original black and white, and the 3d animated version (which this used as a base when generating) and in no way am I trying to be better or replace them. I thought this was a fun and interesting thing to try. I kinda feel like the attention and views this video got should’ve gone to the original MMD (top of description) so please go check that out and like it if you see this. Personally I feel like this video is a little too close to the original MMD and would’ve preferred if it was more unique. Also, I am making no money whatsoever on this video. Ads are because the song is copyrighted, so all revenue goes to the copyright holders. I completely understand the issues people have with AI. I personally don't feel I am doing anything unethical with this video, but you're free to have your own opinions on that. I expected this to get like 20 views like my other stuff. There was no ill intent in the creation of this video. Thank you for watching and I hope you have a great day! Also, regarding a tutorial on how I made this, it's been so long that honestly I've forgotten half of how I made it. And as of now there are so many new tools to use so the tutorial would be very outdated. Maybe in the near future I can make videos showing the process of experimenting with stuff like this and show the full process of how I make it, using new tools and strategies. Might soon try an AI version from the black and white version. I think the technology exists now to do a decent (probably still janky) version. But it will be more of a full video with me experimenting and explaining what I’m doing. Just need to make sure I’m doing my college work first. - edit from 9/8/2023 The rest of this comment is from a while ago. Thank you so much for getting this to nearly 100k views! That’s insane! Bro what now 1 month later it’s at nearly 200k Thank you so much! I get like 25 views on my other stuff lmao
  • @TamTroll
    as a non-tohou player, this is the first time i have any idea what any of these characters are intended to look like outside silhouette form.
  • @luxion8376
    The fact that the AI generation (which I believe makes the animation frame by frame without trying to create a link with the previous frame) clicks the character’s design during the movements is surprisingly very smooth, I love it!
  • @BierBart12
    This feels like a "Bad Apple but I dreamed it after spending too much time looking at Touhou fan art"
  • The remaster we deserved. But seriously, this AI coloured animation is finally looks somewhat closer to original fluid movements and transitions!
  • Imagine if a competent artist did this but more so used it as an animation tool and touched up every frame to make it make sense and flow better, it would be so awesome.
  • @Mars-nx7lx
    This is gonna take awhile but here we go 0:01 Reimu Hakurei (I like how sometimes she Change expression’s and of course not too sexy side view ,since the 3D/MMD ver has her side boob exposed luckily the Ai actually censored it in a way) 0:15 Marisa Kirisame ( she has a cute happy expression , but i think she’s missing her leg / invisible leg) 0:29 Patchouli Knowledge ( this is not bad but of course the ai still messed up the hands In some frames) 0:35 Remilia Scarlet ( Lego hands lol 😂 but In a way her hands still in place just the ai having trouble with the object is hilarious) 0:43 Sakuya Izayoi ( I don’t find any problems on her so far , as expected from the “perfect” maid) 0:50 Flandre Scarlet ( she’s actually gives you a cute look instead of the whole psychopath look which make sense now since she’s no longer the psychopath people think she is , just a bit destructive and lonely ) 0:58 Youmu Konpaku (her clothes keeps changing colors and she did 360neck turn..dam that’s creepy) 1:04 Yuyuko Saigyouji (she actually almost hard to see on the youmu sword scene until the camera focus on her and in some frames she changes her eye color , overall not to messy ) 1:11 Komachi Onozuka(she only gets messed up if the camera isn’t too focused on her , overall nice shading by the AI) 1:18 Shiki Eiki ( just like komachi she gets the special shading ) 1:24 fujiwara no mokou(her hands in some frame is funny and also her head at the end looks weird but overall it’s insanely good) 1:32 keine and her hakutaku form ( they exchanged glasses lol) 1:38 Eirin (i think she stared at my soul) 1:46 Kaguya Houraisan ( her objects got screwed up a bit but overall it’s good) 1:53 primsriver sisters ( in some frames they change colors but the object doesn’t get messed up that much ,good) 2:00 - 2:06 Chen , Ran , Tewi , Reisen , Momiji (all of them looking good ) 2:08 Sanae Kochiya ( she’s actually got the special shading which is cool I guess) 2:11 Hina ( not much of a change really) 2:13- 2:19 Kanako and Suwako ( holy this is good I like how kanako doesn’t express that much and suwako happy face is cute) 2:20 - 2:35 yukari yakumo , tenshi hinanawi(they actually looks so good woah) 2:37- 3:34 aya , suika , alice , nitori , yuuka , Elly , pc-98 reimu and pc98 Marisa , windows marisa , windows reimu( they all look so good in their own way , maybe suika having mouth problems and pc-98 reimu and pc-98 marisa turning into windows in some frames) Man that took like 20 minutes
  • the fact the ai picked on the the designs so well is crazy also 1:01 the ai did have a problem and switch between making her head look like a shrine cause of her moving around so far back
  • the speed at which ai technology is progressing is scary
  • The way the colors of the character's eyes shift slightly is mesmerizing to look at. Not 100% in the touhou rabbit hole but if I'm not mistaken for some of those shifts was cool cause the character is sometimes drawn with either of the 2 eye color options. If am wrong on that is still very well done as most of their designs are pretty on point throughout the frames despite it being ai generated.
  • @what_homework
    3:32 I absolutely love that the ai made the dark side the moon and the light side for the apple, it just adds more to the themes of the animation. Also I love the fact that every character had facial expressions that matched their personalities (the original literally had no eyes to base anything off of). The transitions were surprisingly nice as well, it seems ai has a weird way of effectively doing it better than most people when using color.