graphics
graphics copied to clipboard
cormullion
ai
Oh man, cool! Thanks for starting the issue.
I think the vibe of the org is about making it easy and fluid to work with language, text, and other forms of media. I'd love there to be some kinda expression of "flow" of different forms of content, not sure what that would look like yet.
I think of these I tilt towards the graph looking thing, though it may be too graph-y for the org.
"fluid" suggests a few curves perhaps?
(As a bonus, if you squint it spells "AI" :) )
lmao good bonus -- I wonder if there's something we can do where we have some kind of text on the left or bottom that "explodes" into the Julia colors on the right?
First of all, thank you for helping us!
I like the first robot as it's obviously some "AI", but I also think that nice logos tend to capture the concepts.
I like the idea behind the swirls in the last, but I would struggle to recognize anything generative/AI in it.
I wonder if there is some way to combine the various concepts. For me generative AI is about:
- creating something out of nothing // the magic of it (It still amazes me) -- not sure how to capture? something being half done / or coming out of a box? maybe the gradients and size changing?
- it covers many modalities (text, image, audio) -- some pen or brush?
- it's often very iterative -- some cycle around it?
- it's led by a machine, not a human (but potentially coordinated and supervised by a human) -- maybe a robotic/machine hand?
- it relies on neural nets and chips underneath -- semiconductor-circuit-like lines?
- The benefits can range from supporting creativity/translating vague ideas into finished outputs/productivity -- maybe just some sparks?
Not sure if this helps with some ideas? Not all of the above can be combined together, but I figured it could give ideas?
Good stuff!
I would struggle to recognize anything generative/AI in it
Indeed, many logos rarely illustrate or feature examples of the products/companies they represent. Instead, they're intended to be used frequently and consistently to set up vague subconscious links in the viewer's brains. As you say - somehow "capturing" the idea, rather than showing it.
Must haves: a pleasingly-simple shape, possibly a couple of nice colors, possibly linked with a brand name, and repeated and echoed everywhere; you're much less likely to be showing any details. Here are a few examples to illustrate how rarely they show/contain concepts. If you don't recognise them, can you say what the company/product does?
I'll try to come up with more ideas soon!
That's a good reminder - I wasn't aware. It's also funny how so many of these logos look the same.
Maybe it's their influence, but could we maybe incorporate:
- the essence of infinity/loops/movement
- the process of creation, eg, change in the width/alpha from left to right
- the mechanical nature - not sure how to blend it with smooth curves, but perhaps some circuitry If it follows the letters AI (or GenAI), it would be good to probably be more obvious (triangles, not rectangles)
Or orthogonal to it, how about something like a mechanical brain with Julia dots inside it? (similar to the robot, but evokes more the cognitive element of it, not the robotics)
Not the same, but I like this blend of brain-like mechanical component (AWS):
Incorporating some of these is possible - it would be nice to simplify it though... :)
Oh hell yeah I love these as a start. I tend to prefer logos that have a higher center mass, i.e. where the lines/circuits are larger/wider towards the center. But I really love this "flow" idea where we go from digital to smooth colorful lines.
There's hints of tree growing and color mixing too... I'll think about making the lines thicker nearer the center (do circuit boards do this? :) )
I really like it as well - the transition from circuits to colors. The bolder / simpler logo is definitely better, because it feels simpler (less complicated)
Cool, I'll do a few more like this. Also, if you have some suitable text blurb, I can make some sketches for a social media preview as well.
That’s awesome!
Btw I showed it to my wife and she immediately honed in on:
- we output 4 lines but one of the colors doesn’t match Julia colors (and it’s only three dots anyway)
- it doesn’t say or evoke GenAI in any way (I know we discussed it above)
On the second point, would it make sense to:
- take the output lines to weave together in threads or to be filling the bottom of the logo with mixed colors OR having something like a generative soup of words/numbers/… that the threads are feeding into
- OR could we spell the JuliaGenAI at the bottom as if it was being generated (similar to your original design but simpler / more recognizable) (thinking almost like a 3D printer but not exactly, maybe the lines would go into G, e, n as its being generated and AI would be only empty outlines so far — since current models generate token by token)
What do you think? I’m not sure it can be done cleanly 🤔
it doesn’t say or evoke GenAI in any way
Haha, true! I don’t think I’m creative enough to make a simple and effective logo that clearly says “Generative AI” without using words… 😂 Perhaps others can come up with ideas…
Haha, true! I don’t think I’m creative enough to make a simple and effective logo that clearly says “Generative AI” without using words… 😂 Perhaps others can come up with ideas…
Sorry, didn't mean to sound too critical! It's awesome what you've done so far - It was a mere suggestion for possible directions.
Tbh, I have nothing against having JuliaGenAI at the bottom "being generated" - I think it would work well
No worries! I’ll do a few text-based ones to see how they look at 256x256 sizes. Text is “JuliaGenAI”?
That is the name of the organization. It could be just GenAI for simplicity if it looks better? But the org is called JuliaGenAI.
A couple of different ways to incorporate the name and the other things as well...
I think the black background tends to draw a lot of my eye and might make the logo seem a little dark -- I love the standard white Julia logos.
I actually really love the one on the right -- that's such a brilliant idea to connect each letter (token) to a dot. Really really cool
I agree - I really like the right-hand side one!
Would it be possible to actually change the letters to be the endpoints of those lines (with the same colours)? Making it seem like they are being generated/filled with the Julia colours.
I think it would need a different background as Cameron suggested and maybe we could enlarge the font if there were no dots above the letters? I'm not sure whether it's possible / too much effort, but it might make it more readable.
It's really looking great!
This is great! Thank you for bearing with us!
I really like the first red one, because it’s the closest to the colors in the picture already.
I wonder if we could perhaps try one with a white background as per Cameron’s comment earlier? I think the background fill distracts me / takes some attention away. Would it be possible invert it (white or transparent background with black circuits or simply black outlines whichever is easier?)
Yeah I actually really love the one with the white background. Any other comments on it?
Not from my end and haven’t seen any on Slack either! The biggest question on my mind is how to make it into a sticker 🤣
What usually happens now? Would it make sense to produce a few sizes for different use cases?
An SVG is rendered accurately at any size, or you could export a PNG version at different specific sizes.
Here's an SVG (1000x1000):
and code to export an SVG image to PNG in various sizes:
using Luxor
logo = readsvg("/tmp/ailogo.svg")
for n in [64, 128, 256, 512, 1024]
Drawing(n, n, "/tmp/logo-$(n).png")
origin()
scale(n/1000)
placeimage(logo, centered=true)
finish()
end
But we're struggling at 64 pixels though:
😀
I’m very happy you asked - increasingly people are using AI-generated graphics 😀