Categories: FAANG

This AI went camping

AI image generators have come a long way in recent years. When they first started out, they were quite bad at drawing tents. This was because they could only generate images based on the few examples that they had been given. However, they have since learned how to generalize from these examples and are now much better at drawing tents. In fact, they are now so good at it that they can even create realistic images of tents that have never been seen before. This is a testament to the power of AI and its ability to learn and innovate.

Why are tents difficult to generate?

When you look at an image of a tent, you might not immediately think of geometry or symmetry. However, both of these concepts are essential to the design of most tents. The shape of a typical tent is created by two triangular sides and a rectangular base. This simple shape is easy to assemble and provides a sturdy structure that can withstand high winds. However, the real key to a tent’s stability is its symmetrical design. By placing the poles in equal positions on either side of the tent, the weight is evenly distributed and the risk of collapse is minimized. In fact, many tents are designed with additional elements of symmetry, such as guy lines that help to stabilize the structure in strong winds. So next time you see a photo of a tent, take a closer look and appreciate the geometry and symmetry that make it such an essential item for any camper.

When most people think of a tent, they picture a simple, four-sided structure. However, tents come in a variety of shapes and sizes, each designed for a specific purpose. For example, dome tents are typically used for camping, as they provide a good amount of headroom and can be easily set up in minutes. In contrast, cabin tents are much larger and heavier, making them ideal for extended stays in the wilderness. Then there are tunnel tents, which are long and narrow and often used by mountaineers. Some tents, like bell tents, even have more complex shapes. No matter what the purpose, there is a tent out there that is perfect for the job.

Here is what Midjourney V3 created when prompted to draw “a bell tent”. At best the results are bad, at worst they are pathetic.

View fullsize


View fullsize


View fullsize


View fullsize


View fullsize


View fullsize


View fullsize


View fullsize


AI image generators struggle with geometry and symmetry

While AI image generators have made great strides in recent years, they still struggle with certain aspects of image creation. One area where AI generators often fall short is in the area of geometry and symmetry. This is due to the fact that generating images with perfect symmetry is a very difficult task for computers. As a result, AI generators often create images that are slightly off-center or that have other slight imperfections. While these imperfections may be barely noticeable to humans, they can be a major source of frustration for artists who are trying to create perfect images. In addition, AI image generators often have difficulty creating three-dimensional objects, such as buildings or sculptures. This is because it is very difficult for computers to accurately recreate the complex shapes and shadows that are involved in 3D objects. As a result, AI-generated images often look flat and unrealistic. Despite these challenges, AI image generators have come a long way in recent years and continue to get better with each new generation of software.

Not anymore! Enter Midjourney V4

Midjourney went on a camping trip and now it is an expert at drawing tents. The experience was very fun and Midjourney enjoyed exploring the outdoors. One part of the trip Mirjourney loved a lot was sleeping under the stars in the tent. Midjourney would recommend camping to anyone looking for a fun and unique experience.

Midjourney went camping and now is an expert at drawing tents. Midjourney had never been camping before but went on a trip with friends. Now, Midjourney can tell you everything there is to know about pitching a tent. The biggest lesson Midjourney learned was that it’s important to choose a level spot for your tent. You also want to make sure to stake the tent down, so it doesn’t blow away in the middle of the night. And of course, it’s important to put up a tarp if there’s even a chance of rain. Midjourney had a great time camping and can’t wait to go again. The best part was waking up in the morning, surrounded by nature.

And here is what Midjourney V4 created – orders of magnitude better.

View fullsize


View fullsize


View fullsize


View fullsize


View fullsize


View fullsize


View fullsize


View fullsize


We can’t wait to see what happens when AI gets to play with an umbrella next.

Editor’s note: Apparently this improvement in image quality comes with the downside of adding garbled watermarks to some images. We all know how that came about. Naughty, naughty Midjourney!

AI Generated Robotic Content

Recent Posts

10 Ways to Use Embeddings for Tabular ML Tasks

Embeddings — vector-based numerical representations of typically unstructured data like text — have been primarily…

20 mins ago

Over-Searching in Search-Augmented Large Language Models

Search-augmented large language models (LLMs) excel at knowledge-intensive tasks by integrating external retrieval. However, they…

20 mins ago

How Omada Health scaled patient care by fine-tuning Llama models on Amazon SageMaker AI

This post is co-written with Sunaina Kavi, AI/ML Product Manager at Omada Health. Omada Health,…

20 mins ago

Anthropic launches Cowork, a Claude Desktop agent that works in your files — no coding required

Anthropic released Cowork on Monday, a new AI agent capability that extends the power of…

1 hour ago

New Proposed Legislation Would Let Self-Driving Cars Operate in New York State

New York governor Kathy Hochul says she will propose a new law allowing limited autonomous…

1 hour ago

From brain scans to alloys: Teaching AI to make sense of complex research data

Artificial intelligence (AI) is increasingly used to analyze medical images, materials data and scientific measurements,…

1 hour ago