Binary Odyssey by Futurist Computing Collective
___ ___ ___ ___ ___ ___ ___ ___ ___
___|\__\ |\__\ |\__\ |\__\ |\__\ |\__\ |\__\ |\__\ |\__\
| | | | | | | | | | | | | | | | | | |
| | | | | | | | | | | | | | | | | |____|
| |__|__ |____| |____| |____| |____| |____| |____| |____|____|___/
|_______|__| |____|____|____|____|____|____|____|____|____|____|____|____|____|__________|
"Binary Odyssey"
A demo by Futurist Computing Collective
-------------------------------------------------------------------------------
Greetings to the following demoscene groups and individuals who continue to inspire us:
Demoscene groups:
- Farbrausch
- ASD
- Conspiracy
- Still
- TRBL
Demosceners:
- Gargaj (Hungary)
- Saga Musix (Germany)
- Alcatraz (Sweden)
- Darya (Russia)
- Blueberry (France)
-------------------------------------------------------------------------------
___ ___ ___ ___ _____ _____ _____ _____ _____ _____ _____ _____ _____
|_ | _| |_ | \|_ _| | | __| | _ | __|_ _| __|
| | | | | | | | | | | | | |__ | | | | | | | | __|
| | | | | | < | | | | | |____|_____|__|__|__|__| |_| |_____|
|___|_| |___|_|\_\ |_| |_____|_____|
</AI generated content>
---
Introduction
---
Title: Binary Odyssey
Group: Futurist Computing Collective
Compo: Wild at Revision 2023
This is an experimental project where Artificial Intelligence creates a demoscene demo. This is a project that should ideally be revisited by anyone from time to time to see and assess the progress of artificial intelligence in making demoscene productions (demos or any other productions).
In this production the demo name and the group name has been chosen by ChatGPT. The demo's story was created by ChatGPT, and from the ChatGPT's story the visuals and music was done by various AIs. Human interaction and intervention during the demo creation process was reduced mainly to:
- Define primary technology, length of the demo, and number of scenes
- Fix the code if it did not run / be visible on screen
- Act as intermediary between AI products but follow AI's vision and prompts as close to original as possible with the limitations of AI's current coding skills
- Integrate together the outputs of various AI products to form the demo
See "Demo making process" for more details.
---
Open invitation to Futurist Computing Collective group
---
Every demoscene production where majority (let's say roughly >= 80%) of the production content is made by Artificial Intelligence is welcome to use the group name and join the group if they choose so. However, detailing what AI algorithms and/or products were used and being reasonable explicit on the amount of utilized human interaction needs to be explicit somewhere for others to see. This is to facilitate openness with use and progress of AI technologies.
P.s. You can also use your own handle but, for maintaining dehumanizing aesthetics, creators of this production chose to go with "Human <hexadecimal sequence number>" handles
---
Credits
---
- Storyline: ChatGPT (3.5)
- Code and effects: ChatGPT, GitHub Copilot
- 2D Graphics: Midjourney
- Music: Soundraw, ChatGPT
- Initial AI prompts and parameters: Human 0x00
- Additional code & AI integrations: Human 0x00
- Video Speech/presentation: Human 0x00
- Video production: Human 0x01, Human 0x00
- Video speech writing: Human 0x00, Human 0x01, ChatGPT
- Video deepfake: Faceswap, Human 0x00, Human 0x02
- Deepfake training material: "Elon Musk talks Twitter, Tesla and how his brain works — live at TED2022"
- Video intro&outro music: Robot 0x00
- Info file ASCII Graphics: ChatGPT
- Other tools and technologies: WebGL, Three.js, Tween.js, OBS, iMovie, ffmpeg, Audacity, GIMP, Visual Studio Code
---
Note about Revision competition rules
---
Quote from general rules 2023: "No purely AI generated content. While we do recognize that AI generated content can provide assistance or inspiration, the majority of the submitted entry must exist out of own/group work"
Please note that even if very significant visible and audible portion of the demo is purely AI generated content, significant effort was still needed in the demo making process. Effort was needed, for example, to combine various AI outputs, "chaining" of AIs, creating video and integrating the visuals. So this is most definitely not purely AI generated content when you take the whole creation process into account. Also the entry is part of "Wild" compo deliberately and not any regular PC demo compo even if technologies would fit to PC demo compo.
Also it might be fair to argue that with current state of AI, it does not yet produce, without much assistance, compo winners what comes to audiovisual and code superiority :-)
---
What is this crap?! I think AI can do better than this!!11
---
We do not doubt that our overlords could do better. Especially in a form where human would give more imaginitive instructions like "pretend you are in an LSD trip and you're about to write a screenplay for a demo" and when human extensively corrects/reinstructs it after initial answers. Essentially currently (2023) using AI as a tool Currently best AI use-case would be as an assistant/"copilot" in the making process, not as the lead. However, that's not the point of this particular demo. Point of this demo is to see with as little intervention and interaction as possible what kind of demo AI makes ...and whether it is enjoyable by humans.
This demo is an experiment that should ideally be repeated periodically, after AI solutions improve, to document current state of artistic expression of AI to the future generations of humans.
---
Artificial intelligence is killing the demoscene!11
---
We have survived the dawn of GPUs and demo making using commercially available game engines and other make-to-purpose demo tools. We can probably manage to "survive" AI as well?
AI is just another tool that can be used if production author wants to use it. Personally 2023 state of AI audiovisually still hurts creative person's eyes and ears - during making of this demo there were more than a few times when the Humans wanted to intervene and change effects and scenes but had to stick with AIs vision, as those were the rules of this engagement.
However, it still would not hurt to be specific with tools used to create productions. By acknowledging tools viewers&listeners can better judge and compare production's merits and also learn new things.
---
Demo making process
---
- Human got an idea to go seriously overboard with use of AI.
- Original prompt to ChatGPT by Human: "in style of demoscene write a screenplay of a demo that contains 5 scenes and demo's playtime is 2 minutes. please add also to the scene 4 greetings to known demoscene groups"
- Human collected music cues ("pulsing electronic beat", "music builds to a climax", "music becoming more ethereal and dreamlike") from ChatGPT answer and used them as input for Soundraw which created the music.
- Human prompted ChatGPT to come up with demo name and group name.
- Full disclosure: Original group name proposed was "Digital Horizon" but Human prompted a new name from ChatGPT with similar tone as there was a company/product with the same name already.
- Human prompted ChatGPT to describe graphical elements (backgrounds, overlays) of each scene.
- Human prompted ChatGPT to describe transition between each scene.
- Human prompted ChatGPT to write WebGL/GLSL code for each scene.
- Human did multiple prompts for each scene to avoid "As an AI language model, I cannot write functional WebGL code on this text platform, but I can provide you with a sample WebGL code snippet that could be used as a starting point" and shortened/cut answers from ChatGPT in order for ChatGPT to actually write the scenes described in the screenplay.
- Human used ChatGPT or GitHub Copilot when necessary to fix or amend the code to make it work. Effect was deemed as "working" if it followed the story narrative from ChatGPT, it compiled/ran and it produced visuals visible to humans.
- The effects and scenes do not time terribly well and may not work on all computer setups but that's for the future AI to fix. Humans did not make visual framerate / processor speed independent unless AI generated the code.
- Human integrated, with help of GitHub Copilot, all code from various prompts together into the demo project.
- Human also tweaked various timing parameters while integrating code together to get the visuals aligned in a sequence as described by ChatGPT in the screenplay
- Human prompted Midjourney to imagine pictures based on ChatGPT's description of demo's graphics. Human preferred first variation provided by Midjourney.
- "complete statue of a human figure, use only wireframes, black solid background color"
- "dark, starry sky background with a swirling nebula in the distance"
- "desolate, post-apocalyptic wasteland, with a rusted, metallic sky and towering, wireframe ruins"
- "dark, pulsating void background"
- "neon cityscape at night, towering skyscrapers, flickering lights in the distance"
- "bright, colorful background, abstract"
- Human combined graphics, music and code together.
- Human captured the demo to a video.
- Human shoot and created a introduction video with speech made by ChatGPT and human.
- Human trained a deepfake to replace Human's face with Elon Musk's face.
- Human created & edited the final video that contains introduction video and demo capture.
- Human submitted the demo to the competition.
---
Original screenplay from ChatGPT (all visuals, graphics and music are based on this)
---
Video shows a summary written by ChatGPT from the screenplay, here's the longer version.
FADE IN
SCENE 1:
We see a black screen with white text scrolling up, it reads: "The Power of Art". The text fades out, and a pulsing electronic beat starts.
The opening scene could feature a dark, starry sky background with a swirling nebula or galaxy in the distance.
A subtle graphical overlay, such as a flickering digital distortion effect, could add to the sci-fi feel.
SCENE 2:
A 3D cube appears, rotating in space. It splits into smaller cubes, which start to rotate independently.
Each cube is made up of colorful wireframes, and they move in synchronization with the music.
For the landscape scene, the background could feature a desolate, post-apocalyptic wasteland, with a rusted, metallic sky and towering, wireframe ruins in the distance.
A subtle graphical overlay, such as a color grading, could add to the dystopian feel.
SCENE 3:
The cubes come together again, forming a complex geometric shape that looks like a futuristic city skyline.
The camera flies through the city, showing off the intricate details of the wireframe structures. The music builds to a climax.
The music visualization scene could feature a dark, pulsating void background.
A graphical overlay, such as a VHS-style scanline effect, could add to the retro-futuristic vibe.
SCENE 4:
Suddenly, the wireframe structures start to morph and change, becoming more organic and fluid.
The camera flies through a landscape of swirling colors and shapes, with the music becoming more ethereal and dreamlike.
Greeting to groups are displayed in the screen.
The background could feature a neon cityscape at night, with towering skyscrapers and flickering lights in the distance.
A subtle graphical overlay, such as a lens flare effect, could add to the cyberpunk feel.
SCENE 5:
The landscape fades away, and we see a close-up of a single wireframe structure.
The camera zooms out, revealing that it's a statue of a human figure, made entirely out of wireframes.
The ending scene could feature a bright, colorful background, perhaps inspired by an abstract painting or a fractal pattern.
A subtle graphical overlay, such as a radial blur effect, could add to the surreal, otherworldly feel.
The music fades out, and the screen goes black.
FADE OUT
---
License
---
- Three.js: https://github.com/mrdoob/three.js/blob/5add869e1abfd77e3de35686b25d249669d18d56/LICENSE
- Tween.js: https://github.com/tweenjs/tween.js/blob/01f9dc7322d062932dbd26478854f1f675e3cb03/LICENSE
- Font file 'helvetiker': https://github.com/mrdoob/three.js/blob/5add869e1abfd77e3de35686b25d249669d18d56/examples/fonts/LICENSE
- Source code: https://unlicense.org/ (public domain)
- The demo as a whole or its audiovisual assets (images, music): Creative Commons Attribution 4.0 International (CC BY 4.0): https://creativecommons.org/licenses/by/4.0/
Commercial versions of AI services were used which allow relicensing the outputs of AI as per specific AI services' use agreements.
---
EOF
---
RIP Shadow[ back to the prod ]