UPDATE 💥 The FACTS ABOUT GENERATIVE FILL in PHOTOSHOP 2024

Generative Fill and Generative Expand in Photoshop ... Low Resolution or High Resolution? Will that change? Will it cost me to have High Resolution results? Is it 1024px square or 2000px square? Does using a smaller selection work better? What about the future? What about credits?

𝐀𝐝𝐨𝐛𝐞 𝐏𝐫𝐞𝐬𝐬 𝐑𝐞𝐥𝐞𝐚𝐬𝐞:

𝐀𝐝𝐨𝐛𝐞 𝐆𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐯𝐞 𝐅𝐢𝐥𝐥 𝐅𝐫𝐞𝐪𝐮𝐞𝐧𝐭𝐥𝐲 𝐚𝐬𝐤𝐞𝐝 𝐐𝐮𝐞𝐬𝐭𝐢𝐨𝐧𝐬:

😊 𝐉𝐨𝐢𝐧 𝐦𝐲 𝐄𝐦𝐚𝐢𝐥 𝐂𝐨𝐦𝐦𝐮𝐧𝐢𝐭𝐲 𝐚𝐧𝐝 𝐠𝐫𝐚𝐛 𝐦𝐲 𝐁𝐥𝐚𝐜𝐤 & 𝐖𝐡𝐢𝐭𝐞 𝐏𝐫𝐞𝐬𝐞𝐭 + 𝐒𝐤𝐢𝐧 𝐅𝐢𝐱 𝐀𝐜𝐭𝐢𝐨𝐧 𝐟𝐨𝐫 𝐏𝐡𝐨𝐭𝐨𝐬𝐡𝐨𝐩

🖥 𝐓𝐡𝐞 𝐁𝐞𝐧𝐐 𝐃𝐢𝐬𝐩𝐥𝐚𝐲𝐬 𝐈 𝐮𝐬𝐞 𝐟𝐨𝐫 𝐞𝐝𝐢𝐭𝐢𝐧𝐠 𝐩𝐡𝐨𝐭𝐨𝐠𝐫𝐚𝐩𝐡𝐬 ... 𝐒𝐖𝟑𝟐𝟏𝐂
🖥 𝐓𝐡𝐞 𝐁𝐞𝐧𝐐 𝐃𝐢𝐬𝐩𝐥𝐚𝐲𝐬 𝐈 𝐮𝐬𝐞 𝐟𝐨𝐫 𝐞𝐝𝐢𝐭𝐢𝐧𝐠 𝐯𝐢𝐝𝐞𝐨 ... 𝐏𝐃𝟑𝟒𝟐𝟎𝐐

𝐀𝐬 𝐚𝐧 𝐀𝐐𝐂𝐨𝐥𝐨𝐫 𝐀𝐦𝐛𝐚𝐬𝐬𝐚𝐝𝐨𝐫 𝐈 𝐝𝐨 𝐡𝐚𝐯𝐞 𝐭𝐡𝐞 𝐨𝐩𝐩𝐨𝐫𝐭𝐮𝐧𝐢𝐭𝐲 𝐭𝐨 𝐡𝐞𝐥𝐩 𝐰𝐢𝐭𝐡 𝐭𝐡𝐞 𝐩𝐮𝐫𝐜𝐡𝐚𝐬𝐞 𝐨𝐟 𝐚𝐧 𝐒𝐖 𝐨𝐫 𝐏𝐃 𝐝𝐢𝐬𝐩𝐥𝐚𝐲. 𝐅𝐞𝐞𝐥 𝐟𝐫𝐞𝐞 𝐭𝐨 𝐬𝐞𝐧𝐝 𝐦𝐞 𝐚 𝐝𝐢𝐫𝐞𝐜𝐭 𝐦𝐞𝐬𝐬𝐚𝐠𝐞 𝐨𝐫 𝐜𝐨𝐧𝐭𝐚𝐜𝐭 𝐦𝐞 𝐭𝐡𝐫𝐨𝐮𝐠𝐡 𝐦𝐲 𝐰𝐞𝐛𝐬𝐢𝐭𝐞 𝐢𝐟 𝐲𝐨𝐮❜𝐝 𝐥𝐢𝐤𝐞 𝐭𝐨 𝐤𝐧𝐨𝐰 𝐦𝐨𝐫𝐞.

•𝐌𝐨𝐭𝐢𝐨𝐧 𝐆𝐫𝐚𝐩𝐡𝐢𝐜𝐬, 𝐓𝐢𝐭𝐥𝐞𝐬, 𝐋𝐨𝐰𝐞𝐫 𝟑𝐫𝐝𝐬, 𝐈𝐧𝐭𝐫𝐨𝐬, 𝐓𝐢𝐭𝐥𝐞𝐬 𝐞𝐭𝐜 ... (𝐌𝐨𝐭𝐢𝐨𝐧𝐀𝐫𝐫𝐚𝐲):

𝐉𝐨𝐢𝐧 𝐦𝐞 𝐨𝐧 𝐈𝐧𝐬𝐭𝐚𝐠𝐫𝐚𝐦 / 𝐅𝐚𝐜𝐞𝐛𝐨𝐨𝐤 𝐚𝐧𝐝 𝐓𝐰𝐢𝐭𝐭𝐞𝐫: @𝐠𝐥𝐲𝐧𝐝𝐞𝐰𝐢𝐬

𝐏𝐮𝐛𝐥𝐢𝐬𝐡𝐞𝐝 𝐁𝐨𝐨𝐤𝐬 📘📘📙📗

[mikedixonphoto]

4:00 This is the beginning of the end of affordable and hassle free Photoshop. Once Adobe gets a taste of being able to limit/charge more for individual features, they will start doing more of it. Mark my words.

[ralphpayne]

Keep chasing the facts on this issue - others seem to be ignoring it. I hope they won't price it away from us!

[alphat]

Seems to me that this AI isn't ready for prime time. Should of stayed in beta. Adobe has rushed the release because they are playing catch up.

[mipevo]

I always felt it was going to be a way to generate additional revenue...

[Ian_Walsh]

Thanks for clearing thing up....I've had a dabble and tried using different sized selections and like you i found them soft looking, So, I merged the generated layers duplicated the result and used a high pass filter with hard light to sharpen it and it improved it quite a bit.

[StephanBollinger]

Good video, as always, and "cheers from Australia" btw, it has been quite some time. As for those who get "amazing results", I would assume they start with low-res images. In my workflow, especially since a lot ends up being printed quite large, the quality simply doesn't hold up. To me, it looks like taking a 300 dpi image, downsizing it to 72 for web, then blowing it up again, something like that. In any case, it's not ready for primetime, or any commercial use, unless someone exclusively works for social media.

Here's my theory: As everything is rendered in the cloud, the original image must be sent up as reference. Considering the large file sizes we use, it would take quite some time to actually upload. Instead, I feel as if a low-res version is uploaded, then analysed for light direction, colors etc, and the returning results match the uploaded low-res file.

Consider this - even though we think about the actual selection and the overlap, the returning results take into account the entire photograph, lights, shadows, colors etc, so it must upload the entire image for analysis. This upload alone takes time, aside from the processing. Anyway, just a theory.

[SpikeJackson]

Great video again Glyn. I know it still has a long way to go, but I am still blown away with what it can do and where it is going to go.

[banditalley]

Adobe need to be careful here - Other software developers are upping their game with AI, and if they push for money at every turn it will be tempting to migrate to another system altogether.

[trevorpsy]

My specialty is photographing cats, which implies a lot of time working with blurry fur. I have found that increasing the contrast in blurry areas often results in what looks like (i.e., can pass for) increased resolution. This is often a good solution for blurry fur in generative-fill areas. I need to experiment with hair and see what happens. Perhaps it would also work with shrubbery, rock faces, and the like.

[JennaCee]

A ton of stuff I did not know - thank you. Especially about the credits and 'resolution' of the AI fill

[gianmariozucconi]

You spoke clearly and unequivocally about how things are.
Many people only look at things on the surface, but you go deeper and this is positive for me.
Let's stay at the window and wait for November 1st..., then we can return to the topic with a few more elements;-)

[stellanewman]

Thanks for keeping us up to date and explaining it  :)

[fielding]

Once again, a top tutorial, so professional and straight out of the top drawer. Thank you.

[rdbimages]

Thanks Glyn for being the voice of reason and truth. Something we need more of.

[JohnKorvell]

At this point, I see these as teases and not functional tools in the printed art I do. Maybe we can start a points market where those who are not using the fill tools can sell off our unused points? Might as well get some good out of it at least.

[PhreddCrintt]

Again, well done Glyn - and thank you for this update.

[DavesEpicPhoneShots]

I appreciate this update, thanks Glyn👍

[vincealcazar]

If the Adobe bean counters are left unchecked, I see an outcome where Adobe's Generative AI monetization strategy drives users/subscribers away...to the inevitable competitors who will *not* monetize mimicking Adobe's potential approach.

[Photobandits-gljy]

Glyn, I think the original image source has not to be larger than 2000px. Not the selected area. Probably Adobe scans this image with its informations and used this for generating the fill. If you try it, the outcome is more usable and seamless.

[guybrennan]

thanks glyn. only used it a couple of times for fun. will be great once Ai gets better! looking forward to the full class of Dai portrait that you took in the back of your car.. take care. Guy