Categories
Art artificial intelligence Photography

Adobe Releases Photoshop’s AI Generative Fill

Adobe has announced a new Beta version of Photoshop that comes with generative AI technology.

Pam Clark, writing for the Adobe Blog:

We are thrilled to announce that the Photoshop (beta) app has released Generative Fill, the world’s first co-pilot in creative and design workflows, giving users a magical new way to work. Generative Fill is powered by Adobe Firefly, Adobe’s family of creative generative AI models. Starting today, Photoshop subscribers can create extraordinary imagery from a simple text prompt.

This brings two imaging powerhouses together — Photoshop and generative AI, enabling you to generate content from inside Photoshop with a text prompt and edit it with Photoshop’s comprehensive range of tools to create extraordinary results.

I’ve been having a lot of fun playing around with DiffusionBee running Stable Diffusion and yet this one from Adobe blows me away by how fast and awesome it is. Even if working on Photoshop itself is not your thing, don’t miss the examples.

Categories
artificial intelligence backmasking

ChatGPT on Backmasking

GPT-3 (short for “Generative Pre-training Transformer 3”) is a state-of-the-art language generation model developed by OpenAI. It is designed to generate human-like text by predicting the next word in a sequence based on the words that come before it. I’ve been excited about the capabilities of ChatGPT and the ease at which one is able to generate human sounding responses to questions on obscure topics. I was also intrigued when I found out what it could do to explain how a specific section of code works.

Here is its quick response when I asked it to tell me about backmasking:

Backmasking is a recording technique in which a message is recorded backwards onto a track that is meant to be played forwards. This technique has been used by artists in various genres of music, from rock to hip hop, to add hidden meanings or subliminal messages to their music.

Some people believe that backmasking can have a powerful and even dangerous influence on listeners, particularly when it comes to subliminal messages that are meant to manipulate the listener’s thoughts or behavior.

However, the effectiveness of backmasking as a means of influencing the listener is a topic of much debate. Many scientists and experts have conducted research on the topic and have found that, while it is possible for subliminal messages to be processed by the brain, they do not have the power to significantly change someone’s thoughts or actions.

In addition, the use of backmasking as a means of subliminal messaging has been largely debunked by the scientific community. While it may be possible to hear a backwards message when a track is played in reverse, the brain is not able to comprehend and process the message in the same way it would a message that is played forwards.

Despite the lack of scientific evidence supporting the power of backmasking, the technique remains a controversial and interesting topic in the world of music and popular culture. Some people continue to believe in the power of backmasking, while others see it as a harmless and even humorous gimmick.

Regardless of where you stand on the issue, it is important to approach the topic of backmasking with an open mind and to be aware of the limitations and limitations of this recording technique.

While it’s still early days as far as GPT-3 taking my job as a teacher, I also checked out what it could write up for lesson plans for grade five health class and it was impressive. It does play fast and loose with the facts however — I was asking about world record information and it remixed some real world divers and their records.

I would warn users that just like the complacency that can overcome drivers in a “full self-driving car”, the facts gleaned from GPT-3’s responses should be considered with the same dose of skepticism one would apply to secret backward messages. Not everything you hear, read, or see, is as legitimate as it appears.

Categories
artificial intelligence Photography video

The Follower

Using open source cameras and AI, Dries Depoorter created a website that creates clips of the exact moment instagram photos were taken.

From Dries’ site:

How does this work?

  1. Recorded a selection of open cameras for weeks.
  2. Scraped all Instagram photos tagged with the locations of the open cameras.
  3. Software compares the Instagram with the recorded footage.

It makes me wonder what the world will be like as AI begins to enable more and more of these types of projects. It’s exciting and terrifying.

(via Six Colors)

Categories
Art artificial intelligence design

Diffusion Bee

I’ve been watching the emergence of AI image creation tools like Midjourney, Dall-E, Craiyon and Stable Diffusion and I think it’s pretty neat. But not everyone has been happy about our new AI overlords. Art communities and freelance artists have been particularly vocal about their displeasure.

As much as I see there are serious philosophical issues, in the meantime I wanted to try it out and have been delighted at the news that users of Apple silicon Macs can now access the Stable Diffusion engine on their own computers using Diffusion Bee.

Divam Gupta:

Diffusion Bee is the easiest way to run Stable Diffusion locally on your M1 Mac. Comes with a one-click installer. No dependencies or technical knowledge needed.

  • Runs locally on your computer no data is sent to the cloud (other than request to download the weights and checking for software updates).

It works just fine on my M2 MacBook Air. I’m glad I splurged and got the 24GB of RAM. Because the results can be a bit random, I popped into Photoshop to improve the results of the above “Homer Simpson as a Muppet” prompt. (I’m not sure what the floating head beside him is all about, but I left it because it’s amazing.)

(via Daring Fireball)

Update: I’ve just discovered CHARL-E, another app that utilizes Stable Diffusion to create images on Apple silicon Macs.