Hey friends,
I hope you had a wonderful chrismas season.
As some of you might know, i have started a music carrer in my old days before quite a while, besides maintaining Bforartists. The first two singles Darkest Nights and Here comes the morning and the first album This is the place already appeared back in 2021 already. However, there was some trouble involved. So we decided to rebrand the musician artist name to Tom Goodnoise. Which took unfortunately longer than we wanted and expected to.
I am very excited to tell you that after the rebranding we have now released three more albums at once to the public. Wings of Fate, Listen to the sound and Dystropia. And the first album and the two singles are also re-released. The music style is alternative with dream pop and easy listening elements.
The Tom Goodnoise homepage is located here: www.tomgoodnoise.de
You can follow me at Facebook www.facebook.com/TomGoodnoise/
The music is published by David Wide Music & distributed by Danmark Music Group
www.facebook.com/davidwidemusic/
www.facebook.com/davidwidemusiclicensing/
www.facebook.com/dwsoundtracks/
www.danmarkmusicgroup.com/
www.danmarkmedia.com/
The for me also very interesting part is that i can now finally present the videos to the public. They are all made offline with AI tools. I could and can not effort online tools like Midjourney or RunwayML. I was an early AI adopter. And toy around with this new technique since the first public release of Stable Diffusion. So you can imagine that it nagged me quite a bit that i was not allowed to put the videos into the public.
First let me add that historically the newest videos are made for one of the singles and the album This is the place. Made with AnimateDiff in ComfyUI. Silent or Here comes the morning for example. The oldest videos are in album two, Wings of Fate.
The first video that i have ever created with AI content was the song In your Eyes from the second album Wings of Fate, where i simply used AI generated stil images, and have put them together in DaVinci Resolve.
Still images were fine for a while. But then i discovered AI animation. The first animation tool that i have used was the Deforum extension for Automatic1111, which uses a seed traveling technique. The first and cruelest video in this regards that i have ever done with it is the song She is dancing at the river. Proof that you can do it much better is the song Wings of fate from the second album. This was back in the days when Greg Rutkowsky was still a thing in the Stable Diffusion prompt to achieve his style. I hope he don’t mind. But i have also used it in later songs like Pitchblack. Big parts of the videos of album two and three were made with Deforum.
Then AnimateDiff became a thing. I have mainly used it in album four. In all its three incarnations. AnimateDiff 2, AnimateDiff 3 and AnimateDiff LCM, which is imho currently the best and most flexible AnimateDiff model. The quality differs, dependant of the used model and my knowledge at this time. And i was also limited with my graphics card. It just had 8 gb vram available. Which made me more and more trouble. And i have run into so many OOM’s that i cannot count it anymore.
All is One and Gone from the album Dystropia are two of my favourite videos.
One of the biggest drawbacks with most of my used methods is that the result was more or less coincindent. Well Deforum allows to use a starting image. But AnimateDIFF not. You can workaround with IP Adapter. But it is not one to one. And there is not much control over the animation itself. Wel, yeah, Controlnet with OpenPose would allow at least some posing. But my graphics card was too weak for that. And even the video for Dance with me, which was made with video to video technique, shows a very varying result. Far away from the input videos.
This now slowly changes. We will get more control over the result with the next AI video generation tools which just has hit the market shortly. So my next music videos will most probably be done with other tools than Deforum and AnimateDiff. I toy currently around with CogvideoX. Hunyuan looks very promising too. And i have finally a rig that allows me to produce a better video quality. The 8 gb card was really limiting here. Now it’s 16 Gb. Still not enough, sigh …
I admit i am AI addict. At least when it comes to images and the videos. The music is of course, and will remain, hand crafted. I use AI for the videos only. I see no point to let AI create a song.
Let’s see where this journey will lead to 🙂
Laters
Reiner / Tom Goodnoise