I’ve been checking some of the latest AI Image creation tools, and spent a little time using
Midjourney which as of this date is one of the most exciting tools out there.
But today I’ve been playing Microsoft’s Co-Pilot which seems to hook into Chat-GPT4 and Dale-3.
There’s some confusion as a user as to how exactly I ask for images, but so far I’ve managed to coax a few out.
Not that everyhting has to be Muppet related but here’s the efforts for “Fozzy Bear sitting in a soccer stadium eating a hotdog”.
Not too bad, but let;s try to extend…
Now we ask for “Fozzy Bear sitting in the Ethihad soccer stadium eating a hotdog”.
Less Fozzy Bear now.. so let’s do one last attempt at our prompt.
“Fozzy Bear from The Muppets, watching a football match at the Ethihad soccer stadium eating a hotdog”
This takes some time – and the output on the screen while this runs isn’t the best yet…
.. with those 707070 just keep displaying… but if we give it a coupld of minutes we eventually get an image…
This is not bad… wonder if we can change the character??
“A photo realistic image of Kemit the frog watching a football match at the Ethihad soccer stadium eating a hotdog”
It seems that we have more data to create a more real Kermit.
I wonder if we can try a real person? Let’s try Rod Stewart.
And not exactly yet :)
And that didn’t work. Copyrighted possibly?
But we can put Darth Vader there!
And we change direction with a “Steampunk polar bear walking the Yorkshire Dales”.
All pretty impressive, but no way to save the prompts like MidJourney.
But included as part of your Microsoft account – this is pretty amazing.
I’ll play are trying to create posters next.
