Modern Primitivism

 

What i love about this whole art strategy is that you can see how if you develop it a little bit more with very strategic design of the MSG effects to put together harmonious patterning you could really start to control it.

Here i've used the paint synthesizer and an image folder background texture to build something similar without using gallery show.

You can use the paint synthesizer preset in a PASeq to build out more of what gallery show was doing.

Turning on an EBM (energy based model) digital paint feature lets you tighten it up if you want that.

MSG Stack Filtering 5

 

Stack filtering MSG imagery in gallery show using temporal image operation effects in Studio Artist.  Run through a transition context.  The original MSG key-frame animation is shown below.  The MSG keyframes were generated via live mutation exploration in the Evolution Editor.

So every frame in this movie is an image on the stack.  Stack filters take in multiple images to generate a single output image.  Gallery show can randomize the positioning within the stack being filtered.

MSG Stack Filtering 4

 

MSG stack filtering 3

 

MSG stack filtering 2

 

MSG stack filtering animation experiment generated in Gallery Show.

MSG stack filtering

 

murKy water

 

Paia Lurker

 

Working with temporal Color Simplify for stack filtering in gallery show in Studio Artist V5.5.5.

Gallery Show Runner

 

AI Artist Paints What It Thinks a Deep Learning Neural Net is Seeing in It's Internal Feature Space Representation

 

Continuing today's riff on faking the output of deep learning neural net feature space visualizations.  The AI Generative Artist did all of the work. I just setup a generative art strategy for it.

I guess we should work on making it generative for building generative strategies.  Meta generative system.  Sure, why not.

Faking Feature Space Representations

 

So i was reading this multi-modal neuron feature representation paper (neural net stuff), and they were generating some wild output images. So i wanted to know how they were doing their back-prop optimization from a feature vector in their neural net to the actual images they were showing.

And then i looked at the PyTorch code for that part. And i realized you can get there way, way simpler than what they are doing.  Spent 5 minutes in Studio Artist V5.5 dialing in something close. Their facets are better, but i think you could get very close just using existing Studio Artist features.

So this would be visualizing what a clown neuron represents. Here's another slightly different approach.

My Plague Year - a68

I like the aggregate imagery that starts to pop up as you look at this. The news cycle regurgitated and spat back at you. In my plague year.

My Plague Year, a67

Lots of crazy news. Portrayed using stack filtering techniques here.  In my plague year.

My Plague Year, a66

More dual stage stack filtering experiments. In my plague year.

My Plague Year, a65

Another experiment where an effect is inserted into the input chain for a temporal ip op effect processing a folder of curated topical images. So steroid induced stack filtering. In my plague year.

My Plague Year, a63

Expanding the concept of stack filtering to include a effect processing step on the individual randomly selected frame images prior to inserting them into the temporal ip op effect.  Prototyped it as a 2 stage process, but will probably add it as a new Gallery Show feature. Working with curated topical imagery from current events. In my plague year.

My Plague Year, a62

When things were so hectic that we missed one. Did you notice. We did. 

Sorry. Very busy in the code dungeon.   Details to follow. Pretty cool ones. In my plague year.

My Plague Year, a61

Scraping the news barrel. For outrageous current events. Details to follow. In my plague year.

The coin toss

So don't get to excited about antibody testing just yet.   You might just want to toss a coin instead. From Ed Yong, but matches other articles i have read. Pay attention to the statistics. 

"False positives are a problem. Many pinned their hopes on antibody tests, which purportedly show whether someone has been infected by the coronavirus. One such test claims to correctly identify people with those antibodies 93.8 percent of the time. By contrast, it identifies phantom antibodies in 4.4 percent of people who don’t have them. That false-positive rate sounds acceptably low. It’s not.

Let’s assume 5 percent of the U.S. has been infected so far. Among 1,000 people, the test would correctly identify antibodies in 47 of the 50 people who had them. But it would also wrongly spot antibodies in 42 of the 950 people without them. The number of true positives and false positives would be almost equal. In this scenario, if you were told you had coronavirus antibodies, your odds of actually having them would be little better than a coin toss."

My Plague Year, a47

Stay safe. In my plague year.