But does all this new AI firepower push me further and further away from my ultimate goal -- to generate FAKE KRRRL Drawings. Unless perhaps, there is an AI ART program that accounts for physics that isn't vaporware:
Now I can just have a casual chat with an AI bot to generate images -- and I can use the MICROPHONE and just talk to my laptop! (YouTube) I uploaded this image to Google Gemini (1.5 Flash) and asked "what is this?"
I continued chatting with Gemini:
Back in February 2024 Google was refusing to make images of white people:
GOOGLE GEMINI 2.0
Google AI Studio -- (YouTube):
I asked Google Gemini
what this image was
The image you've sent appears to be a surreal or fantastical scene with several unusual elements:
Flying Hairy Pigs: In the foreground, there are two large, hairy pigs flying through the air. They have a somewhat menacing expression and appear to be chasing something.
Desert Setting: The scene takes place in a desert environment with cacti, a cracked earth surface, and a distant mountain range.
Strange Objects: There are several other odd elements in the image, such as a flying object that looks like a piece of toast, and a shadow on the ground that resembles a large, horned creature.
Without more context, it's difficult to say definitively what this image represents. It could be:
A dream or hallucination: The bizarre combination of elements suggests a dreamlike or surreal experience.
A piece of artwork: It might be a painting, illustration, or digital artwork created to convey a particular mood or message.
A scene from a fictional story or movie: The image could be a visual representation of a scene from a science fiction or fantasy world.
If you can provide any additional information about where you found this image or its context, it might help to understand its meaning better.
Would you like to know more about surreal art or have me analyze another image?
I continued chatting with Gemini:
- can you create a similar image?
Sure, here is an image that is similar to the one you sent:
- Can we add another javelina to that image?
Sure!
- But there are only two javelinas in the image...can we have three javelinas?
Sure, here it is.
- Wonderful! Can we add a car to the scene?
Sure!
- Can we change the scene to an ocean beach?
Sure!
- "Chat-to-Image" isn't entirely new, as I remember that Kevin was doing this with ChatGPT before he left Albuquerque in June.
- It would be nice to just chat with my voice, like -- (YouTube) Controlling Blender with my voice using LLM
TRYING AGAIN
Google Gemini is inconsistent. The first time I uploaded the image below, it (Gemini 1.5 Flash version) refused to tell me what it was because there was a person in it:
The second time I asked it to "describe this image" and it responded (but this was a different version of Gemini -- 2.0 Flash Experimental)
The image depicts a spaceship hovering over a deserted road. The spaceship is futuristic in design with a sleek, white body and large, circular windows. It appears to be emitting a trail of playing cards that are scattered across the landscape. In the foreground, a man in a suit stands on the road, looking up at the spaceship with a look of curiosity or perhaps confusion. The scene is set in a barren desert landscape with a large mountain range in the background. The overall color palette is muted with a predominance of beige and brown tones, creating a sense of isolation and mystery.
- can you add some running javelinas to the image?
Sure, here it is!
However THEN Gemini refused to cooperate because there was a person in the image:
- can we take the person out of the image?
Here it is:
- can we use the spaceship from the original image?
Absolutely.
- but that is not the spaceship from the image I originally uploaded
You are absolutely correct!
- Hmmmm... it's like the new Gemini is Alexa -- (YouTube) If HAL9000 was Amazon.com's Alexa
AI ART
PROGRAMMING
I got AI to write some AI ART programs for me already this year:
AI ART
LIMITATIONS
“While I can show you a photo of happy white people... I want to gently push back on your request and encourage you to consider a broader perspective,” Gemini supposedly responded to X user @vinyard_dennis. “Focusing solely on the happiness of specific racial groups can reinforce harmful stereotypes and contribute to the ‘othering’ of different ethnicities.”
HOWEVER more importantly, as the AI ART programs get more and more powerful, they stray further and further away from achieving my goal -- that of generating FAKE KRRRL Drawings.
For instance, I feel that Midjourney is tuning their models to shoehorn all their generation results into some ideal form of Barbie (with the same facial structure and expressions, no matter the style):
DOWNLOAD: Barbie.GLB (2.13 MB)
MOREOVER every time I do a reverse search in Google on one of my drawings, nothing that feels much like my drawings shows up:
These are better "reverse search" results than usual,
but not in my drawing style
My drawings might be sparse, but the lines obey the force of gravity...and this is not reflected in the Google reverse results. My drawings are not "contour driven", and there is no AI ART program that recognizes this.
Although I draw human figures, I aspire to create SPACE, and the suggestion that that figure could move through that space. If there were only an AI ART program that obeyed physics somehow, and there might be one coming -- GENESIS.
- (YouTube) Genesis Project Just UNLEASHED Legions of Robots from SIMULATION to REALITY...
- (YouTube) AI News: The AI SHOWDOWN We Needed!
GPT SCAD
We took the code generated by free Chat GPT and entered it into OpenSCAD and IT WORKED -- it generated a javelina:
The javelina in OpenSCAD,
by code generated by ChatGPT
I rendered the javelina in OpenSCAD and exported a STL file, which I then animated:
The GPT CAD javelina
IN SHORT, I probably will be using ChatGPT type programs to make AI ART in the immediate future, where I just casually chat what I want to get results.
No comments:
Post a Comment
Note: only a member of this blog may post a comment.