I generated some AI images using AI Comic Factory, Midjourney and Stability Matrix (mixing them with some of my drawings, and also with photographs), and then animated the AI images in PIKA LABS -- to create another AI ANIMATION "COLLAGE":
Contemplating the new Contemporary Art
EbSynth
The original animation
generated in Pika Labs
I followed the YouTube tutorial for EbSynth (note that the "key" style image has to be the same pixel dimensions as the video frames):
I generated several different style variations with EbSynth, and made a video of all the variations:
This might be the most interesting variation:
AI FRAME INTERPOLATION
I used AI "Frame Interpolation" in Runway ML on one of my old animated GIFs:
AI MOVIE TOOLS
These are short AI animations pieced together like a collage to make a "movie." The next step is to make a longer "movie," perhaps by using these AI tools below:
- Deforum (inside Stability Matrix, installed on the Quelab computer)
- Pika Labs (free, but on Discord)
- Runway ML (free and paid, online, I used "Frame Interpolation" on Runway ML)
- EB Synth (free to download, style transfer on video)
- With a smartphone we could film with AR objects floating around in the real world -- easily by uploading a GLB into <model-viewer>. I have plenty of GLB files to upload.
- We might incorporate Gaussian Splattering into video using PolyCam
- Polycam -- Gaussian Splattering trial
GAUSSIAN SPLATTER SCENE
I experimented with the free Polycam 3D Gaussian Splattering trial -- the process worked, but my resulting scene was one big blur:
I created a 3D figure (210504__A.glb) and animation in Paint 3D, and then broke that down into individual frames using EzGif. Polycam accepted those frames and created a Gaussian Splatter image (in "ply" format), but without the 3D figure I uploaded. I think this is because the figure had no background for the software to reference.
The "ply" file did open in Blender:
Gaussian Splatter file in "ply" format
did open in Blender
QUESTION
Is there anyway of altering a whole "movie" with AI, in the same way we can upload an image and alter it with text:
Google's new AI feature (SGE) answered the above question:
AI is developing so quickly that I doubt that anyone, even Hollywood, has a full grasp about what can be done. This has to be an opportunity for an outsider to slide in and make something significant... or not.
IDEA: What if we constructed a scene in the real world, a scene that was first created in PIKA LABS, and then filmed that short scene. Then we might collage the animation with the video to make something bigger -- life imitates AI -- albeit still only a few seconds long.
No comments:
Post a Comment
Note: only a member of this blog may post a comment.