• Welcome to the Community Forums at HiveWire 3D! Please note that the user name you choose for our forum will be displayed to the public. Our store was closed as January 4, 2021. You can find HiveWire 3D and Lisa's Botanicals products, as well as many of our Contributing Artists, at Renderosity. This thread lists where many are now selling their products. Renderosity is generously putting products which were purchased at HiveWire 3D and are now sold at their store into customer accounts by gifting them. This is not an overnight process so please be patient, if you have already emailed them about this. If you have NOT emailed them, please see the 2nd post in this thread for instructions on what you need to do

I Just Wanted to Post an Image Thread

Saphirewild

Brilliant


I finally got around to using the Eastern Dragons!!
 

Ken1171

Esteemed
Contributing Artist
And now for something completely different... this is an experiment to conceptualize what an alien settlement would look like in Mars if designed my H R Giger. Image generated from a text description using a Python VQGAN neural network trained over my RTX 3090 GPU, creating this image in about 2 minutes.

Ladies and gentlemen, the task of creating concept art will never be the same now that Generative Adversarial Networks (GANs) have become a reality. Could this threaten concept artists who do this for living?

GigerInMars_1440.jpg
 

Ken1171

Esteemed
Contributing Artist
Here's another example using a homemade Python VQGAN with a community pre-trained model. I asked it to make me a "biomechanical woman in Giger style", and I think I've got exactly what I asked for. I had to keep the image size small (256x256) because GANs require absurd amounts of VRAM to process this, but I can code another neural network just to upscale this to a larger size.

GigerWoman_1.jpg


The image generation starts from a random cloud of noise, and if we use the same randomization seed, we can reproduce this exact image at will. Conversely, by changing the initial noise seed, we can create a virtually infinite amount of variations from the same input text description. In the example below, I have just changed the initial random noise seed to another number a couple of times, and these are what I've got from the same text input. Is this cool or what?

GigerWoman_2.jpg
GigerWoman_3.jpg
 

Ken1171

Esteemed
Contributing Artist
Thanks, Pen. It's exciting being able to create new and original paintings from our favorite artists who have already passed. The examples above are like all the H R Giger paintings he never made. As amazing as it may sound, all these images came out of Python code. To what category such images would belong to?
 

Pendraia

Sage
Contributing Artist
Good question...are you playing with ai? Copyright is something that needs to be resolved for ai images.
 

Ken1171

Esteemed
Contributing Artist
Yes, I have been playing with AI in Python lately. Do you mean to whom AI-generated images legally belong to? Perhaps the dilemma here is not to whom it belongs to, but instead who is the rightful "author"? It takes a lot of time and effort to develop the skills to create paintings and sculptures, be them digitally or in the real world, but AI makes images out of text descriptions anybody can type.

It's a new kind of art creation that requires no skills that need to be developed. Since AI is trained with a large quantity of art created by real people, it basically steals the skills from the actual authors when creating new pieces based on their work. For example, the images above required training a GAN model using thousands of H R Giger paintings. If someone wanted AI to create "Ken1171" renders, it would need thousands of my renders to train yet another GAN model to feed the AI with.

In a related subject, some Hollywood actors, like Mark Hamill, has sold the rights to his likeness to allow Disney to create 3D versions of him in movies where he will not even participate. They have his permission to do it. When it comes to AI images, similar permissions might be required if the source material artist is still alive.
 

Miss B

Drawing Life 1 Pixel at a Time
CV-BEE
You know, there's a thread in the Poser forum at R'osity that I think deals with a similar AI setup. Some of the renders folks have been posting are cute but not serious renders, but some look nice, and all that's needed is a few keywords or phrase to get it done. Check it out as it may be very similar to what you're using.

Turn Words To Art
 

Ken1171

Esteemed
Contributing Artist
That is a good example where poorly chosen words produce poorly composed AI generated images. :)
 
Top