Running On A Tick();
Recently, I had a discussion with friends, Captain and Yumenoshima, about the limitations of Unreal Engine. But the conversation took a strange, musical, AI detour.
Talking About Performance
The majority of limitations anyone will experience with Unreal Engine (UE) are due to the machine it is running on. How many assets, textures, RPCs, polygons and actors can exist simultaneously depends on your machine’s memory and performance, and crucially, the patience of your audience. An unoptimized design or one overloaded with elements will slow a game or experience to a crawl.
Many object classes in Unreal Engine can receive a call to their Tick() event function on every frame. This is the best place to attach code that has an immediate, visible effect on rendered components and actors, especially where smooth movements are desired. Because it runs on every frame, code attached to Tick can substantially affect performance. It's best to avoid subscribing to Tick except where necessary.
The Conversation Takes a Detour
I was introduced to Yumen while collaborating on one of Captain’s projects. Yumen has a penchant for the bizarre in pursuit of humor. He often trolls Captain with non-sequitur tangents disguised as technical inquiries. The underlying absurdity of his premise is revealed gradually, and he doesn’t give up immediately when the ruse is exposed. It makes for a lot of fun, and Captain grumpily expresses faux irritation.
Running on a tick, time stands still
Recently, we talked about multiplayer optimizations and how many simultaneous players could reasonably play together on a non-instanced game server. We discussed turn-based gameplay and batching, and then Yumen began asking questions about headless clients and even email as a form of user input. It was another unfolding absurdity from Yumen that made me laugh, all the more so because Captain was trying to explain another idea and was trying to stay on topic.
I kept hearing Captain say, “everything running on a Tick.” At this point, I was laughing, not paying much attention. Feeling flippant, I decided to draw a crude illustration of the image in my mind: a man running on the back of a giant tick.
Where the Real and Digital Start
I sent the image to the boys, and I had a laugh. It took less than five minutes, and it was totally worth it. I joked that Captain should AI-enhance the image, and I thought that was the end of it. About two minutes later, he responded with a much better illustration. “Can you make it realistic?” I laughed. Not even a minute later, I received this impressive art:
Living in a World That's Running on a... Tick
The immense and undeniable utility of AI image generation becomes immediately apparent when looking at an image of a man literally running on the back of an insect. It might be a tiny man running on a regular sized tick, actually. But the low-effort idea is successfully rendered as a visual gag, in stunning quality and detail. One of his hands even has five fingers.
The image was created with the help of GPT-4o. OpenAI’s website states, “It accepts as input any combination of text, audio, image, and video and generates any combination of text, audio, and image outputs.”
Every moment, every trick
The amount of effort and time investment previously required to produce an image of this caliber would be prohibitive for all but corporate endeavors and dedicated artists. And yet, with AI, the scales have been flipped. This lame pun has been elevated to high art, and a silly thought has been enhanced into a funny memory for all of us to enjoy.
Asked to comment on his process, Captain stated (unassisted, in his own words):
“Creating the visuals for "Running on a Tick" was an undertaking of unparalleled sophistication, a true testament to my expertise in the realm of digital art. It all started with my discerning eye selecting a highly curated dataset, each image chosen for its superior thematic and aesthetic qualities.”
”With my guidance, advanced neural networks dissected these images to their most fundamental elements, achieving a level of detail that only someone of my caliber could orchestrate. Then, I directed a symphony of generative adversarial networks (GANs) in a series of iterations so intricate and precise, it bordered on the mystical. The final image was not just a product of AI but a masterpiece that encapsulated the very essence of the song, a result that only my visionary approach could achieve.“
Not Done Yet (The Best Part)
Not satisfied with simply knocking the park out of the solar system, Captain surprised us with a stupendous musical composition. You simply must have a listen.
I can’t get over how well the lyrics explore the implied themes while simultaneously seeming to understand the double meaning. The song is so well crafted. It’s hilarious, playful, and catchy. It’s really firing on all cylinders, and it tickles me to no end. GPT-4o was used again to craft the lyrics, and Suno was used to create the song. Captain also suggested that Replay and Weights could be helpful.
A Beat We Simply Can’t Deny
Ordinarily, it would take a lot of time and effort to produce art and music this polished and specific. In the past, these concepts would develop in the minds of daydreamers—entertaining and humorous to their creators but trapped behind the labyrinthine efforts of production required for others to enjoy. Sometimes these ideas germinate specific details that would be invaluable in the work to create a finished product, but rarely do they coalesce into actual verse or product. But in the world of ML language models and AI generation, these ideas can quickly become near-finished products with hilarious and brilliant results.
