Anyone spend time today putting the Photoshop beta through it’s paces? I easily wasted a couple hours* and as a v1 of the tech it’s bananas. Is it perfect? No, but it represents the first commercially available generative ai use-case I’ve seen that’s even remotely close to useable.
Anyone else have a play?
*I fully acknowledge this may be a function of being currently “Job Insecure”
Saw the demos. Looks rad. I’ve “Bob Ross”d my fair share of things with flame paint. But I can’t see a reason not to just be popping stuff into photoshop now and get me some nice looking clean plates
Saw the demo. Downloaded a beta (24.6). Totally confused as to where the text to image bit is supposed to be
Maybe the correct beta hasn’t made it all the way to NZ yet. Too busy to hunt for it
Checking it out now. It is quite exciting! It did a reasonable job on a frame in Log.
I think it has got to be something that we embrace. Is it flippant/ignorant to compare it to a calculator or spell check? I know that ChatGPT will write your essay not just check for spelling errors but it is coming, it is here, so we should get busy using and mastering it.
I tried it for 5 mins and it spit out some real garbage. But I think it’s going to threaten matte painter’s careers.
I’m going to change my answer. Played with it another 5 mins and… it’s not perfect, but it’s surprising. Matches lighting, got the shadows and color cast right…
Ok, played with it even more and… set extension matte painting? Wow. Obvs clients are picky, but some of that nuts and bolts everyday cleanup…? Done. Fin. The End.
while not perfect, I’m certainly going to use it for some quick clean jobs.
Lets hope AI doesnt learn how to make the Logo bigger or the packshot more wow or I’ll be out of a job.
Downloaded new photoshop and looking forward to trying.
Spent a bunch of time w MLOPs integration in Houdini which embraces the tech with mind-blowing/unexpected possibilites.
AI solves are an inevitable tool going forward.
We quickly/unexpectedly get great reults of our preferences at a blindingly fast rate.
The flipside is that it quickly becomes a narcisstic, navel-gazing videogame.
What’s done with that initial result is the where the artist’s path lies.
The craft of creative disciplines is largely problem-solving.
The “problem” dealt to us always evolves.
This will replace tedious work within creative disciplines.
BUT lessons earned thru that tedious work are not experience gained by the AI result.
Embrace tools that gets to the “creative” part quicker.
Any clarity from clients within the creative process is welcome.
Admittedly only a sketch, it’s a map to what’s on their mind.
No longer words but pictures, which they’ve helped decide.
Those who are more talk than walk,
will indeed need to up their game.
Those bringing years of problem-solving to the table,
should not discount it’s value.
I’m reminded of a professor in art school who would paint the inside of his mouth in exchange for donations to his Las Vegas fund.
The beautiful/toxic colors required more green, as you’d expect.
Dwight’s methodology has stuck with me and boils down to this:
Lost interest playing around as 50% of the time “the images cannot be shown as they are against the guidelines”. For pretty normal prompts like mountain range or forest…