Future is NOW. AI video with Deforum stable diffusion. #ai #trending
This is what my experiments look like for now.
I shot a video with Iphone 13 pro in cinematic mode, then I passed it through AI using Deforum_Stable_Diffusion_0.7, this is a notebook in google collab.
After some research and time spent with stable diffusion model I was not happy that I can’t really control any results much and I stumbled into Analog_Diffusion model. I added that model as a custom to Deforum notebook and since then I am pleased with results.
For the deforum settings you can ask me in comments, they are different for video and animation, also las image of video I’ve used as an init image for animation sequence. I create content with 12 frames per second, I noticed that 24 and 25 fps i can’t really make smooth. Smoothness is big deal here.)
I render everything with euler sampler, 120-180 steps with strength , means I am applying around 60 steps to the pictures, as I noticed 60 steps of euler is giving you nice looking picture already.
Then I wrote an animation sequence about 8 precise prompts long and added camera movement with a formula. Then I edited them together and made a sound, I had original voice in the video but I changed it with AI voice from and then enhanced it through
(I don’t really know how to post url’s here, but I will learn).
I can really recommend Adobe Enhancer, it makes nice voice even if you record it with mobile phone as I did.
Then I use Davinci resolve and Adobe Audition to make sound design, I make render of 12 frames per second and then interpolate it through Davinci with Optical Flow, Enhanced better settings. I’ve tried time warp but I didn’t notice any drastic changes in the picture and decided that it is not worth it.
It took me some time to make sound design and match it with voice, and I decided that a little psychedelic and glitchy sound will do the best, I want to create atmosphere of organic digital world.
Please like, share and subscribe. I need some conversations that will motivate me to continue working on it, but now I am passionate about it as never.
#shorts #stablediffusion #deforum #midjourney #analogdiffusion #animation #fashion #technology
1 view
4
0
4 weeks ago 00:08:10 1
AI Agents Will Create MILLIONAIRES in 2025 – Are You Ready
4 weeks ago 00:03:02 1
SPX Options Trading : Strategies for Big Gains!
1 month ago 00:19:33 1
Славуцкий и Крихели. Еженедельный прогноз движения котировок аграрных фьючерсов на 46 неделю.
1 month ago 00:11:43 1
AliExpress Is DEAD!? (FASTEST Shipping option)
1 month ago 00:02:10 1
Glockenspiel - “Corpse Bride Theme“ Lesson by Project Musica 🎶
1 month ago 01:12:37 1
НАШЫ ИГРЫ 2024: Первое в мире шоу русских игр
1 month ago 00:00:00 1
Live: Blast in Pakistan | 24 people killed and 25 injured in bomb blast at railway station in Quetta
1 month ago 00:08:45 2
TXT Answer The Web’s Most Searched Questions | WIRED