Originally published at: The history of using AI to make images | Boing Boing
…
I’ve been messing around with this stuff, and I think we’re getting closer to the singularity.
I’ve been playing with text-to-image using my own computer (as opposed to using remote computers). My graphics card is just barely powerful enough to do this kind of work, so this took about 2 weeks of hearing the fan spin at top speed. Each image was pretty small, then upscaled to HD, then upscaled again to 4K (make sure to choose 4K resolution or it looks pretty pixelated).
Edit: The VOX video is really good. Who is the artist, indeed.
This is another of my experimentations with text-to-image, and I think this an area that hasn’t really been explored much yet: using the technology to effect existing images.
Here I took video of my self singing, and used each frame as a seed image, then used a text prompt to change the way I looked. If the flickering can be reduced, I can see this technique being used in movies and music videos (once again, choose 4K or image is really pixelated).
now you just have to speech to text your lyrics to fully close the loop
My beta test of Midjourney ran out. But it was fun. I would have to quit some other monthly charge to justify it now. Very cool to see all the imagery generated out of the written word. Mine: Skull sucker.
As an Illustrator I’m ambivalent. It looks like an amazing tool to brainstorm new takes on ideas, but it’s taking work from photographers and artists too (as you said, you don’t commission images to illustrate your articles anymore, you just throw prompts at it)
I love new technology and experimental techniques and I love seeing what people achieve with them. I just worry about artists struggling to make ends meet and how these tools will be ethically used.
I did wonder WTF was going on with images on the RSS feed, it’s been producing nightmare fuel like this:
This topic was automatically closed after 5 days. New replies are no longer allowed.