I think some of these are ok for kids but be aware that many are using the data to refine LLMs and use human intelligence to refine larger models used for image recognition and text based descriptions that will feed into commercial models. The issue with this is that it's skewed to the populations who use it, making the way images are described and generated by AI skewed to particular brackets where kids have access to what's needed to get to and use these tools.
I have a favorite addition that's not AI based - Pixicade, which lets you draw or design all kinds of games that come alive in an app. It's been expanded since the edition I found on clearance at Walmart and had so much depth. It's a really fun way for kids who draw levels for games to make them come alive. It can require adjustment, experimentation, and trying different things to see what works best.
I also really like the KiwiCo STEM crates which teach electronics, circuits, etc. Lego robot kits have come a long way and I like combining these tools with basic logic and pre-coding activities that teach logic and loops through play.
Hey Moira, thanks for sharing Pixicade. I haven't heard of the app before, and it sounds promising. And yes, the KiwiCo STEM crates were on my "to get" list for a while, but Denmark has a very prohibitive import policy, so we never ended up ordering one as it would likely cost just twice the price to receive them here.
As I am developing the artwork to go along with my rock opera, I'm hoping to model it after punk rock show flyers from the 1980s. How does MJ do with ragged looking photocopy type art? Or is it too "pretty" to use for that?
I think it can handle almost anything with the right description. Here's what I got by simply throwing some of the keywords from your comment as a test run: https://i.imgur.com/69Sr5kR.png
If you really want to mimic a very specific look, you could even use an external image as a "style reference" for Midjourney to work with.
To begin with, you can simply add "graphic novel" to your prompt. Here's my very first try with simply "elephant in the jungle, graphic novel --ar 3:2": https://i.imgur.com/VV3ZiKP.png
[subject] is where you put the description of your subject or scene. In my demo case, my three subjects were "elephant," "sofa," and "cityscape." So the entire prompt would look like "elephant, kawaii." That's it!
Yeah exactly. Also, using names of specific artists will obviously have a strong impact as well, since it captures more than just the brushtroke, or color palette, etc. But I try to steer clear of recommending specific artist names, because that always feels iffy to me. Going for broader styles like this feels a bit more "fair game" though.
And yeah, graphic novels by Midjourney tend to be pretty great as well. Comic books, too. (Actually still have that one for a potential future list.)
I think some of these are ok for kids but be aware that many are using the data to refine LLMs and use human intelligence to refine larger models used for image recognition and text based descriptions that will feed into commercial models. The issue with this is that it's skewed to the populations who use it, making the way images are described and generated by AI skewed to particular brackets where kids have access to what's needed to get to and use these tools.
I have a favorite addition that's not AI based - Pixicade, which lets you draw or design all kinds of games that come alive in an app. It's been expanded since the edition I found on clearance at Walmart and had so much depth. It's a really fun way for kids who draw levels for games to make them come alive. It can require adjustment, experimentation, and trying different things to see what works best.
I also really like the KiwiCo STEM crates which teach electronics, circuits, etc. Lego robot kits have come a long way and I like combining these tools with basic logic and pre-coding activities that teach logic and loops through play.
Hey Moira, thanks for sharing Pixicade. I haven't heard of the app before, and it sounds promising. And yes, the KiwiCo STEM crates were on my "to get" list for a while, but Denmark has a very prohibitive import policy, so we never ended up ordering one as it would likely cost just twice the price to receive them here.
I'm wondering if your comment was meant to go under another post of mine that was aimed at AI for kids? https://www.whytryai.com/p/ai-for-kids
China makes oodles of similar stem kits. Amazon
Yes it was and I think substack is absurdly difficult and overcomplicated and I want Usenet back heh
Thanks! That is very helpful.
Thanks! That’s close but not quite right. I’ll play around.
Here's what happens when I use this cover image as a Style Reference: https://www.pinterest.com/pin/222365300322756307/
Midjourney mimics it rather well (I made both color and black and white versions - and tried both guitar and drums):
https://i.imgur.com/oU4vy1s.png
So I think it's a matter of finding the right reference image and seeing if Midjourney can apply that style. Good luck!
As I am developing the artwork to go along with my rock opera, I'm hoping to model it after punk rock show flyers from the 1980s. How does MJ do with ragged looking photocopy type art? Or is it too "pretty" to use for that?
I think it can handle almost anything with the right description. Here's what I got by simply throwing some of the keywords from your comment as a test run: https://i.imgur.com/69Sr5kR.png
If you really want to mimic a very specific look, you could even use an external image as a "style reference" for Midjourney to work with.
Pls do a how to on a graphic novel w MJ
To begin with, you can simply add "graphic novel" to your prompt. Here's my very first try with simply "elephant in the jungle, graphic novel --ar 3:2": https://i.imgur.com/VV3ZiKP.png
When you say [subject] what am I putting in there. What’s the full prompt. That would have helped me more to see in full. Thank you.
[subject] is where you put the description of your subject or scene. In my demo case, my three subjects were "elephant," "sofa," and "cityscape." So the entire prompt would look like "elephant, kawaii." That's it!
I like the simplicity of this approach, and I bet it will work just as well (or close) for GPT4/Dall-E. Nice!
I've been asking a lot more lately for graphic novel style stuff. I love those old styles and they represent a tiny piece of me in my work.
Yeah exactly. Also, using names of specific artists will obviously have a strong impact as well, since it captures more than just the brushtroke, or color palette, etc. But I try to steer clear of recommending specific artist names, because that always feels iffy to me. Going for broader styles like this feels a bit more "fair game" though.
And yeah, graphic novels by Midjourney tend to be pretty great as well. Comic books, too. (Actually still have that one for a potential future list.)
I think Dall-E restricts you from using individual artist styles now (probably wisely as the legal issues are navigated).
Yeah it restricts the use of living artists I believe, or those within the last 100 years.
Ooh, I know some really good artists from more than 100 years ago.