🖼️ Ideogram AI is Clever AF!

Ai5 Logo

Welcome, Explorer 🤖

Anthropic’s new LLM is blowing people’s minds, the overlap of AI and the next generation of video games, and how Zuck is fast becoming the golden child of AI.


Today’s Ai5:

  • 🧠 Claude 3.5 Sonnet is a Coding Beast
  • 🖼️ I Took Ideogram for a Spin
  • 🦾 AI is Powering Next-Gen Video Games
  • 💛 How Zuck’s becoming the Golden Boy of AI
  • 🧐 Midjourney Style Blending

Prompt of the Day

A nostalgic handheld gaming device displaying a colorful pixelated sim game on screen, over a matte background, minimalist product photography, modelled on the original Game Boy. Simple logo text that reads "Game-Boy"
A nostalgic handheld gaming device displaying a colorful pixelated sim game on screen, over a matte background, minimalist product photography, modelled on the original Game Boy. Simple logo text that reads “Game-Boy”

Turns out it’s really hard to generate an accurate image of the original Game-Boy in Midjourney. I spent way too long on this. 😅

Can you do better? Send me your best shot and I’ll post it in the next newsletter. 🙂


Claude Sonnet is a Coding BEAST 🧠

The big story over the weekend was the release of Anthropic’s new LLM – ​Claude 3.5 Sonnet​. If you missed it, I’ll bring you up to speed. 👇

  • Claude 3.5 Sonnet released as a direct competitor to ChatGPT 4o
  • Appears to outperform OpenAI models on several benchmarks
  • Massive context window (around 150,000 words) with a 99% recall accuracy.
  • Added computer vision functionality

But what’s really created excitement is Sonnet’s crazy coding abilities. I’ll show you a few examples.

I created a ​working Snake II game​ in under 60 seconds. 🤯🤯

Snake II created by Claude in seconds
I used to be much better ok.

Pietro created a Super Mario clone in a few minutes. (Imagine how long it took back in the 80s) 👇

Here’s another example showing a solar system animation with physics.

If you’re using Claude and are unable to execute code, go to the home screen and enable Artifacts.

How to activate Artifacts in Claude AI

On another note, I decided to ​cancel my ChatGPT subscription​ and try out Claude for a while. So far, so good!


Taking Ideogram for a Spin 🖼️

​Ideogram​ blew-up in 2023 when it became the first AI image generator to produce coherent text. Since then though, I haven’t heard much about it.

I decided to check-in with Ideogram and see what it could do.

I was planning on generating a few images myself. But after seeing the top ranked images in the explore section, I had to share them! 👇

A cat has just stolen a fish in a fishmonger's shop and runs away with it. He has a panicked look on his face because he is being chasedby the fishmonger. The fish appears to be silver and of medium size. In the background, the fishmonger yells at the cat. The scene is the street of an open-air market
A cat has just stolen a fish in a fishmonger’s shop and runs away with it. He has a panicked look on his face because he is being chasedby the fishmonger. The fish appears to be silver and of medium size. In the background, the fishmonger yells at the cat. The scene is the street of an open-air market
A cartoon of a toothbrush standing next to a roll of toilet paper. The toothbrush looks fed up. The toilet paper looks at the toothbrushangrily. There is a white speech bubble coming from each. The speech bubble pointing to the toothbrush says, "I hate my job". The tooth brush coming from the toilet paper says, " Oh Please!!!" There is bold text at the bottom of the image that says, "Always be thankful, life could be worse", typography, illustration
A cartoon of a toothbrush standing next to a roll of toilet paper. The toothbrush looks fed up. The toilet paper looks at the toothbrushangrily. There is a white speech bubble coming from each. The speech bubble pointing to the toothbrush says, “I hate my job”. The tooth brush coming from the toilet paper says, ” Oh Please!!!” There is bold text at the bottom of the image that says, “Always be thankful, life could be worse”, typography, illustration
A creative 3D render of a scene where a massive eraser at the end of a pencil is erasing the sea, revealing dry land for a manto walk on. The pencil is ahead of the man, and he is walking towards it. The background has a vibrant sunset, and the sea is calm with gentle waves. The phrase in bold white text on a sign at the bottom of the image says "If you can't find a way, make one" is displayed in stylish typography, adds a motivational touch to the image
A creative 3D render of a scene where a massive eraser at the end of a pencil is erasing the sea, revealing dry land for a manto walk on. The pencil is ahead of the man, and he is walking towards it. The background has a vibrant sunset, and the sea is calm with gentle waves. The phrase in bold white text on a sign at the bottom of the image says “If you can’t find a way, make one” is displayed in stylish typography, adds a motivational touch to the image
A collection of ten adorable Disney-style cartoon animal stickers. The animals include a friendly turtle, a cute elephant, a playful giraffe, a wise owl, a hopping bunny, a graceful deer, a majestic eagle, a smiling turtle, a kawaii dolphin, a awesome Lion, a cheerful panda, and a charming fox. Each animal is drawn in a vivid and colorful style, with expressive eyes and engaging poses
A collection of ten adorable Disney-style cartoon animal stickers. The animals include a friendly turtle, a cute elephant, a playful giraffe, a wise owl, a hopping bunny, a graceful deer, a majestic eagle, a smiling turtle, a kawaii dolphin, a awesome Lion, a cheerful panda, and a charming fox. Each animal is drawn in a vivid and colorful style, with expressive eyes and engaging poses
Pikachu, divided in the middle, one side dark aggressive, animalistic, other side, joyful, playful, colorful. sparks, realistic fur, intricatedetails. volumetric lighting, ray tracing, over-detailed, intricate background, conceptual art, detailed rendering, 3d render
Pikachu, divided in the middle, one side dark aggressive, animalistic, other side, joyful, playful, colorful. sparks, realistic fur, intricatedetails. volumetric lighting, ray tracing, over-detailed, intricate background, conceptual art, detailed rendering, 3d render
minimalist illustration of a white basic paper airplane floating above a perfectly flat blue uniform background, it cast only the blurredshadow silhouette of a fighter jet on the background right underneath. The text at the top of the image reads: "Use your imagination".
minimalist illustration of a white basic paper airplane floating above a perfectly flat blue uniform background, it cast only the blurredshadow silhouette of a fighter jet on the background right underneath. The text at the top of the image reads: “Use your imagination”.

I was actually really impressed by these! The quality of the images is super nice and the prompt adherence is crazy good.

Perhaps one of the most underrated image generators? I’ll definitely be using Ideogram a bit more from now on.

Ideogram allow 20 free images a day if you want to ​try it out​.


The Next Generation of Video Games 🦾

While most industries rush to integrate AI, the gaming sector is moving cautiously.

There’s several good reasons for this:

  • Predictability: Games need a certain level of consistency to meet expectations. Imagine an NPC hallucinating a conversation in a video game. Not good.
  • Compute: Running the AI models requires big processing power. The current hardware and cloud options make it difficult.
  • Complexity: Integrating AI into gameplay means rethinking fundamental game design principles.

That’s not to say game companies aren’t actively working with AI on several fronts. 👇

NPCs

Non-player characters (NPCs) will probably be the first place we see AI roll-outs in gaming. Last year’s video from Nvidia between a player and an AI generated NPC gives a taste of what we’re in for.

Object Generation

Another safe roll-out could be object and world generation. Take Two Interactive (creators of GTA) have ​filed patents​ implicating AI object generation. This of course is fueling speculation on how much AI we’ll see in next year’s ​GTA6​.

User Generated Content

Generative AI will enable players to create customized content within games. Things like skins and equipment are already rolling out on some UGC platforms.

Dynamic Storylines

Current games use explicitly programmed rule-based AI for storylines. Next-generation AI might enable truly unique, dynamically generated narratives.

Dreaming

This one is next level. Marc Andreessen made some interesting connections between AI, cinema, and gaming in a ​recent interview​. He theorizes as the 3 start to overlap we might end up with an entirely new art form. He called it dreaming.


Everybody 💛’s Zuck

I recently came across a post from @​AIatMeta​ on X. The post outlined 5 major open-sourced AI assets being released. 👇

After reading the post and digging into the comments, I noticed they were overwhelmingly positive. In fact, I couldn’t find a single negative one.

This got me thinking about Meta’s open source AI policy, and why it might be a giga-brain move for Meta. 👇

A puppy happy with excitement, in the style of cartoon realism, 32k uhd, wallpaper, luminous brushwork. (Created with Meta's Emu)
A puppy happy with excitement, in the style of cartoon realism, 32k uhd, wallpaper, luminous brushwork. (Created with Meta’s Emu)

You don’t become one of the biggest companies on the planet by giving things away for free. Here’s the 3 main reasons going open-sourced might work in Meta’s favor:

  • Standardization: By open-sourcing models Meta might establish its AI as the industry standard. It also makes it harder for smaller competitors to gain traction.
  • Talent: Open-source projects are usually popular among top devs. This might help Meta recruit elite talent in AI.
  • Security: External scrutiny and contributions to open-source models can quickly identify flaws. This is great for Meta as they can release products with less worry about backlash.

So even though some are calling it the Zuck redemption arc, remember that it’s always about business.

I do love me some based Zuck thou. 😆

Based Zuck

Midjourney Style Blending 🧐

In classic Midjourney style, we can now combine all sorts of style references and weights. 👇

  • Blend more than one –sref code (–sref num1 num2 numX)

–sref 2219275291 4139648621

  • Blend more than one Personalization number (–p num1 num2 numX)

–p prhx95i glzhtgy

  • Add weights

–sref 2219275291::3 4139648621

Combine URLs

–sref 2219275291 URL

And, if you want, make any combination of all of the above.

An apple --ar 16:9 --sref 2591060807::3 3117130089 --personalize prhx95i
An apple –ar 16:9 –sref 2591060807::3 3117130089 –personalize prhx95i

If you want to see more of what this style of prompting can do, I recommend giving ​Tatiana​ a follow on X.


Snack Sized 5 🍪

1️⃣ Do you remember Age of Empires? This ​LinkedIn post​ discusses how advanced the AI used in the 1999 release really was.

2️⃣ What is the smallest integer whose square is between 15 and 30? This question posed to LLMs has sparked a ​heated debate on X​.

3️⃣ One of the AI filmmakers from last issue put together this Cyberpunk 2077 ​talking head style film​ using Hedra.

4️⃣ Elon’s ​building a massive GPU​ in Texas.

5️⃣ Public backlash forces London cinema to ​cancel the screening​ of ChatGPT-written film.


Help us grow by sharing Ai5 with a friend. I’ll give you a cookie. 🍪

SUBSCRIBE to Ai5

Join the conversation on X.

👈 Read Previous Issue | Read Next Issue 👉

Keep reading on Rareconnections.

If you have any special requests, tips, or you just want to talk about AI, email me: matt@rareconnections.io