New Photoshop ai BETTER Than Generative Fill?
HTML-код
- Опубликовано: 25 дек 2024
- New Generate Image in Photoshop beta. Colin Smith shows you how to generate images with new Generate Image using Firefly 3 in Photoshop.
► Free Photoshop add ons: photoshopcafe....
► I'm a BenQ Ambassador and use their monitors: www.benq.com/e...
► THE GEAR I USE: www.bhphotovid...
► www.amazon.com...
► THE MUSIC I USE: share.epidemics...
► SUBSCRIBE FOR MORE VIDS: www.youtube.co...
► PREMIUM COURSES: photoshopCAFE....
► INSTAGRAM: / photoshopcafe
► WEB: photoshopcafe.com
► FACEBOOK: / photoshopcafe
► TWITTER: / photoshopcafe
#Photoshop #adobeFirefly #generativeai #communityxadobe
When are they going to get rid of the ridiculous censorship? you can't type in 'bikini' or you get a TOS message, but you can say 'swimsuit', is a bikini THAT much worse? oh my stars!. They need to open it up to let us create ANYTHING we want to when we are paying for it.
I tried 'bikini' once and got referred to the Adobe thought police, I then tried 'beachwear' and got a bikini. Trying to second-guess the algorithm is a time consuming nonsense.
There seems to be no real thought gone in to this, I mean Adobe stock images will serve up images of women in bikinis by the dozen, but its AI has a nervous breakdown at the mere thought.
@@yannmaenden7236 Yeah, bing did that, even with innocent stuff, but then I actually found myself some amusement trying to push the boundaries, and getting it to produce something that it never would if it actually knew what was meant.
For example, I actually wanted a cartoon rabbit one time, and I'd mis-typed and left the 'T' off the end to get the unsafe warning, which inspired me to try and make some cartoons that DID include the semitiс people's preachers.
I gave up using it …. 95% of any description I typed was immediately flagged…..
Try swimsuit instead
Great vid, Colin! 👍
Thanks
Generative fill is designed to fill a selected area of an image with something you prompt or by leaving the box blank. Generative fill, if left blank will try to fill a selected area based on the content of the image
Afer the update, it's leaving harsh lines all around the selection. I didnt have this issue before.. Many people are complaining in Adobe forum about this but it still has not been fixed. :(
Thanks so much as always for keeping us updated ❤
Right now one of the best applications for this I've found is for creating custom art for the Samsung "The Frame" tv. Photoshop lets me tailor the image precisely like I want and then add an oil brushed effect to it.
cool idea
Amazing!!! Your videos are very informative. Keep them coming.
Awesome... you explain things so much better and easier than anyone else out there! Whenever new features come out, I know to check your channel to figure out how to use them!
Thanks so much for the kind words
It is interesting to see where this is going. You are right it would be hard to see where this fits in a use case for photographers but over time more and more things come to mind...backgrounds, side panels, elements, items and so on. it is good to know how this is evolving.
I've been trying to generate template backgrounds (for sports portraits) but they still look too cartoonish.
I’m loving these! Thankyou
Digital art and cartoonish images look ok but it just can’t get photo realistic images, they always look off, like air brushed
is because they are... the noise pattern gives away ...
Looks like Gen Fill is still using Firefly 2, or maybe a separate Inpainting model. Inpainting (generating an irregularly-shaped image based on the surroundings) needs different model training than straight text-to-image. With Stable Diffusion, most models don't work well with inpainting and only some dedicated ones do.
Good comment
Excellent demonstration 👍🏼👍🏼😃
We have been playing with this at the agency level for a while, and the effect on the bottom line is immense when you consider the production time savings. But it is the top line, and I mean the creative top line, that I have to question and wrap my head around the potential of Ai. Does anyone know how this affects the future of image-making and content creation? We don't and we are not even sure of the questions to ask either. The cliche in any artmaking is that it’s all been done before. Machine learning is the apperceptive mass of what’s all been done before because it regurgitates new things from large data sets, which are things that have been done before. Creative people will use AI creatively, but what concerns me is that at the end of the day, it depends on how you deploy AI. If you are going to relinquish control of your creative talents and your own abilities as a visual artist, as a creator, as a generator of ideas that are not trite and boring, and as a thinker, then the tool owns you and your work, not you.
Big topic. I made a video a while ago "willI be replaced by ai" those are my thought and they haven't really changed
You do a great job of keeping up with all this stuff and showing us how it works. Thank you.
How fun! Great video, Colin!
Nice that you dive so much into this, no matter if we like it or not, it WILL be a part of our future.....
Yes indeed, “coming, ready or not.”
Excellent video.
Always fun and interesting.. thank you!
Great video as always. In my opinion, it is nice to have this option and other similar ones together in a graphics program, but when comparing PS this option to other programs, the accuracy is pathetic. And a orange warning about violating the regulations is an overstatement not even try create anything bad.
Great job. Very good explanation. Thaks a lot.
This is a good video, pretty informative. I wonder when this will be pushed to Photoshop Proper?
Thanks
Great video. You have inspired me to do my own illustrations for my new book with the help of Photoshop. Cheers!
That was coool, I'm currently trying stylar, any videos regarding it?
Thanks for another splendid tute..
Hi Colin, Can you use these new features to swap a shirt on an image from a selected new item? As a menswear retailer, new items arrive in-store, and loading them up to a website displaying a new photo wearing the shirt or other clothing item would save us a lot of time and effort. If you have already covered this in another video, please link me to your example?
Incredible! It would be great if you could take one of your own photos and then take the editing style from a reference photo. So it would keep the original as it was except for colouring and editing style.
You can do that. See a previous vid I made recently
@@photoshopcafe I can't seem to find it. Can you point me in the right direction?
@@Fotomadsenit’s not hard fella. Click on Colin’s page. Click videos. Read the titles. It literally is RECENT as in less than two weeks.
You can create simplistic reference images using rough and ready composites. You don't need to make them super realistic. I know someone who does artwork for the movie industry, and she uses Pixel Squid 3D models to add elements to simple reference images, for instance, and Pixel Squid models are far from photorealistic.
Thanks for the video. Is all this processed on device or online?
Online
Incredible how Ai is progressing.
Can you use a photo and apply a new ai generated effect on top? Like a portrait of someone and applying a cyberpunk effect and keeping the person recognizable?
Wowze... some features here.... I am wondering ... what is the max resolution to use this... (still 1024 pixels?). Thanks for your tuts... they are awesome!
Gen image is 2048 X 2048. Gen fill is 1024 x1024
@@photoshopcafe Thanks a lot!!
I have used the "Create Image" feature many times now and the "Reference Image" never appears in the image I create. I get stuff "like" it but I want to have, for example, an image of my cat, or a person that I have an image of and I cannot get it. If I try to use a person I know with a beard, I get the same guy in all the samples who has a beard but doesn't look anything like the person I want. Or, if I put my cat on the reference image, I get cats with different colored fur.
Structure Reference in firefly online is more accurate to the reference. I hope they split structure and style in PS, like on FF
The biggest problem is that everything that happens to the image is beyond our control. These are random actions of the algorithm. You can arrive at some results by trial and error, but it's still not a tool that gives you control over the resulting image
Is this the 26.03 beta or an even newer version? I haven't had a chance to try it.
The one from Tuesday
I liked you ‘glitch in the matrix’ reference 🤪
Yes! Someone got it :) my jokes are subtle
so my PC will not support the new Beta Photoshop will this still be the case when the latest features get updated to regular Photoshop?
Without knowing your system specs, and what version you are currently running, it’s impossible to say for sure. If you are running the latest release of Photoshop, there is no reason these features shouldn’t work for you in main PS build
What's even more impressive is that they are going to implement this in Premiere..
great tuturial
WARNING!!!! You are limited to a certain amount of AI credits on your account depending on which Adobe plan you have. I have the photography plan. Since I had the plan before November 2023, I am allowed 250 credits per month. People who got on the plan after November 2023 are only given 100 credits per month. Each time you generate an image or a generative fill you lose one credit. So here I was playing around and seeing what I could create, and when I needed to use the generative fill for a work project, I only had one credit left. That frosts me! That should have been made very clear when I started using the fill.
Thanks for sharing.
anytime
Wondering if a person uses his own photo of a model he shot as a reference in Generative Image will it use the face and then add elements from the prompt like background and change outfit. That way you could do cosplay characters (or fantasy outfits like from D and D for example) but keep the same face and general body shape the same as the original photo. Would this work? This would have practical use to dress up actual model photo poses if it actually works. Will it work?
I don't think it works that way, but firefly is closer and you could composite the face in after getting the same pose
I appreciate that you always let us know what the lated AI update is doing - Adobe does not usually let you know. This was very entertaining (I love cats)!
This seems to be quite stable. Do you think it will debut in regular PS soon? Thanks
could do
Do you need a specific version of Adobe Photoshop to have access to the AI features?
ruclips.net/video/t0vU4arjlJk/видео.htmlsi=HgpZgeINYT8dk0fN
Couldn't you use it as a background generator for photos to salvage shots with a poor or distracting background by overlaying a subtracted subject?
Yup, sure
This is great thanks for sharing the news and demoing the process - one thing I use photoshop a lot for in my fashion and glamour photography work is to swap and add outfits on models and I use generative fill for this - is there a way to use firefly 3 instead somehow as I cant seem to figure a way to trigger firefly 3 with a selection? perhaps though there isnt much difference in the output and eventually generative fill will be using firefly 3 anyway!
Gen fill is supposed to be using FF3 right now.
Thanks
Thank you for the tip re the better quality of generate image and generative fill; good to know but odd - it seems to me - that there would be a disparity since one would think that the same AI engines are "generating" for both.
Does anyone have an explanation for this anomaly?
That's, why its beta, they are fine tuning the knobs
I think enough Beta. We need a blog, or informative page when these features will be out of beta and into the public release.. Too many updates and no need for beta apps, just have a enable beta features inside the primary application like you did with Neural filters!
I love Photoshop!
Thank you for sharing. We run CC at tiny locally owned print shop / plans exchange in bumfk Montana USA. AI making many things easier but damn is it funny to show some of the old folks how fast the adobe fire flies into nearly limitless directions.
Don't forget about using it to generate backgrounds
So much fun.
Very Exciting!
😀
Wowzers!
I have never liked Generative Fill. It gives some of the weirdest stuff I have ever seen. Hopefully Generate will be better.
This looks like the web based Adobe Firefly, integrated into photoshop.
As I mentioned on the video, it basically is
Yes, it is great fun. However I'm not sure how many of us will find it useful. It's OK for generic images and illustrative purposes, the sort of stuff people look for by trawling the search engines, but I would have thought that most people ( not just photographers ) use Photoshop for editing existing photos .
We do the mundane stuff. We add text and logos, we crop and resize, we remove that weird guy, that nobody invited, from wedding photographs etc.
The concept that we are all disciples of Andre Breton, itching to create wild and wonderful graphic designs, is probably misplaced.
True, it’s not for everyone all the time. But if you’re doing compositing, it’s great for backgrounds or even making things like, for illustration purposes.. a turtle, a rotary phone or something you might not have a photo of. Not everyone will need this feature. But not everyone will need the type tool or the slice tool. But those are indispensable to those who need them.
As a freelance graphic designer for more than 5 years.
I honestly dont see any future for me in the years to come… 😅😅
Very interesting
Whoa, how can THIS company be lagging so far behind ????
I can think of half a dozen things that could be improved in Photoshop that would be VASTLY more useful for us that THIS!!!!!! However, improving those little points won't make Adobe as much money as this will... So I guess improvements are OUT!!!!!!!
Powerful stuff. No need to continue with your sketching I guess?
I still enjoy sketching and will continue to do so
Not so good yet But it's comming and will get better afcourse.
Definitely a WIP, but I think it’s moving pretty fast
Hi
You are an artist, what are you doing?
Showing the new features Adobe has added, like always
What is the use case for this fakery?
Adobe are now pissing on photographers - that's not to be celebrated.
lol no they’re not. Stop scaremongering. This addition is for digital artist. You clearly didn’t watch the entire video.
No theiri mor. Remember when digital photography appears? Every photographer or at least the most, o no this is the end of photography.
And remember, Firefly needs photos to generate what you ask ,and who's taking photos.
Photographers will be around for a long long time,those who refuse to adapt will vanish
@@jmgcg I remember when digital images were finally allowed entry into my camera club competitions, and some of the old timers hated it. They thought them synonymous with Photoshop, which gave people an unfair advantage. In actual fact, a surprising number of people cheat by using Photoshop on Natural History competition entries, which is totally against the rules. In non-NH competitions and Exhibitions, an entry can be 30% computer generated here in NZ, which sounds a lot to me, but those are the rules, and they apply to everyone.
its a complicated subject u chose what about a real photo
Looks like crass junk