Loss of Intent & Midjourney Sued! - AIxC: 86
+ ChatGPT > Colab > Kontext for Paid Subscribers
Today we’re talking about the realities of utilizing Ai, lawsuits involving Ai, and last but not least, how you don’t always need to reach for Ai or image outputs to streamline your workflows.
Hector & AIxC team
01 / TIPS
Loss of Intent
Here’s the thing with Ai tools - every generation has a loss of intent. Image, video, 3D, all of them.
How much loss every step gets you depends on the type of generation, but they all have some loss.
Some times straying from the target is frustrating.
Perhaps you’re looking for a precise detail to come through when you go from sketch-to-render, but it doesn’t make the cut. So close, but yet so far. You now have to go into photoshop to clean things up.
Other times it’s an opportunity for discovery.
Happy accidents. You click on Vary Subtle in Midjourney and you get to see variations that may an improvement of what you started with.
In the video above you can see the following:
Sketch to more sketch views with Flux Kontext. Good results, but different widths, details changing - loss of intent.
Sketch to render in Vizcom. Definitely leaving out some details.
Render to 3D in Vizcom. Again, losing detail.
I say all this to say that loss of intent is par for the course with Ai. You have to plan for it. And if you know it’s coming, you can better gauge where Ai can plug into your workflow.
The silver lining with all of this? The loss of intent has steadily been lessening. So I think we can reasonably expect for that loss to keep smaller and smaller.
02 / NEWS
Midjourney in the News
Midjourney is in the news as of late. One reason is because they are getting ready to jump into the video realm.
The other is because they are being sued by Disney and Universal. Read about it all here.
We’ll be keeping an eye on this and the implications.
03 / FOR PAID SUBSCRIBERS
Not Every Tool and Workflow Has to Be Visual
We focus a lot on outputting images, videos and 3D models with Ai. Well, not everything has to be Ai or focused on visuals to be of use. In today’s R&D Lab, we’ll highlight how Manu utilized a mixture of ChatGPT and Google Colab to creation a shortcut solution to a problem we were facing.
Sign up below for access to the AIxC R&D Lab board and this workflow documentation and the detailed breakdown with prompts. Link and password behind the paywall. Please note the new password for the month!