You would think a company like Microsoft now releasing their own AI image generator would have already caught up and hit the ground running by having more than just the generic [insert prompt here] style interface and utility.
Now I am sure they are mainly using it to harvest cheap labour with everyone testing and pushing their Microsoft rewards more.
Another newcomer to the AI race is Adobe, and it does look good the little bit they have released so far but again it is Adobe so there is no need to pay attention unless you also want to be bankrupt.
Adobe has always has some impressive tools, little features like content aware scaling but even with all that I have almost always just brushed by hand because the shit hardly ever is perfect so the time spent cleaning up has never been worth it to me.
My fave adobe tool is basically the content aware healing tool and that is it. So I hope the fanboys go berserk over their little Firefly offering but with AI now being the driving force for many tools I think Adobe will be forced to have more open systems.
Of all the tools though I think they probably have some of the highest quality legit training data so yeah it is nothing to cough at and their reach is broader than anyone; Ranging from image editing, and video to audio and web.
However they are still gated and by the looks of it although many tools seem to offer full generation easily enough ; They are also limiting.
To be fair it takes a lot to generate these images and with the use of your free quotas across the board a hobby user would be fine with what they can make given low effort. The ones who find it really useful will pay some amount just because they like something specific.
Although I assume most would only pay for Midjourney as it really leaves most in the dust.
Hence the others trying to catch up and either create perceived uniqueness, exclusivity etc etc.
But Canva I think has gone the right path, because I think full image generation is a bit of a hassle.
Personally, I will still just go find a damn photo slap it on a post and call it a day. Sure if you are feeling artsy then go ahead have fun suffering with prompts and scraping for credits.
As you can see above the girl has three different hair thingies, the original image I dragged and dropped from Canva elements and then used the Magic Edit option when editing the image to then insert a prompt for the brushed section to replace.
My first prompt was "Embroidered Pin" and the second just "Banana" , top left and right images respectively.
I think it did pretty ok, and then tried something different.
"Charred and Burning tree Stump" with the original base image being just written paper style image.
Now I guess you could do similar by finding a .png and slap it ontop of any other image; Obviously this is not better than full image generation but I think it definitely can be more useful in many cases.
What I think is really great about this is that Canva is implementing this into a already useful product, something where if people who already have a certain brand style and tend to like their designs a certain way can progressively implement AI image generation in a way that has worth to them.
Also I like the idea that Adobe got a bit snickered at since Canva released this the day after Firefly by Adobe and it already is a functional tool.