Has Photoshop Gone Too Far with Generative Fill?

preview_player
Показать описание


🔥 🎓 15% off our Professional Photography Tutorials: Use Code YOUTUBE

🦸‍♂️15% off our photo course The Well-Rounded Photographer featuring 8 different professional photographers: Use Code YOUTUBE

Subscribe to the Fstoppers Youtube Channel:

Our Gear: 📷 and 🎥Workflow Recommendations:

🥰Our Favorite Gear

🎸Music in our videos🎶

Epidemic Sound

💻Software📀
Adobe Creative Cloud

Boris FX Optics:

Luminar Neo

Capture One

🛒🏪🛍 Support Fstoppers by shopping at:

B&H Photo and Video

Amazon

📸Follow Fstoppers on Instagram:
Follow Lee and Patrick's Puerto Rico Instagram:

In this video Patrick Hall takes a look at Adobe's newest and craziest Photoshop tool, Generative Fill. Generative Fill is a new way of editing photos that uses AI or Artificial Intelligence. By downloading Adobe Photoshop and using the Creative Cloud to gain access to Photoshop Beta, all users of Photoshop can now access this synthetic fill wizardry. With Generative Fill you can now easily expand your background, remove your background, remove people, add clouds, change or replace the sky, add clothes (you can't remove clothes), expand your canvas, turn a vertical photo into a horizontal photo, clean up your seamless background, remove tattoos or add tattoos, and just about anything else you can imagine. In this reaction video, Patrick uses several of his own photos to explore just the tip of the iceberg on what generative fill can do prior to launching to the public.
Рекомендации по теме
Комментарии
Автор




🔥 🎓 15% off our Professional Photography Tutorials: Use Code YOUTUBE

🦸‍♂15% off our photo course The Well-Rounded Photographer featuring 8 different professional photographers: Use Code YOUTUBE

Subscribe to the Fstoppers Youtube Channel:

FStoppers
Автор

I’m a real estate photographer and I generally outsource the removal of things like cars, dumpsters etc. in exterior shots and pass the cost off to my clients. I removed a car from some drone photos this morning using this amazing tool in about 5 seconds with outstanding results. I think it’s great!

crshenise
Автор

When you think of the hours we’ve all spent retouching things in and out of images, spending years learning the craft then this comes along. Very impressive

techtipsuk
Автор

I just experimented with this a bit and the results were absolutely insane. I used a photo I took of a friend's car, and then used the generative fill to change the background to make it look like it was in a forest, on a mountain, in a city, whatever I could think of. The cityscape ones were a little funky looking, but the more nature based ones were completely convincing. I am speechless. I think I need to go lay down for a bit.

KevinBein
Автор

I went down this rabbit hole yesterday. The generative fills work really well, but the generated objects are very hit and miss. When it gets mid-journey level of generating objects, it will be stunning. I assume that will be in a couple of weeks!

acamilop
Автор

The image is produced in a resolution of 1024 pixels on the longer side. In order to avoid the blur problem on big images you have to do the expansion in small chunks. Or work on a low resolution image.

paolovolante
Автор

As an ex prepress finished artist, working from 91-2006, on titles such as Time magazine, Cleo, cosmopolitan, Harper's bazaar, Dolly, People, Picture, Ralph, Elle, Home Beautiful, Men's Health, Marie Claire, new idea and many others, I can seriously confirm this is nothing short of astonishing!
I remember spending an entire shift changing Cameron Diaz' dress, giving her a shorter haircut (older Getty image had longer hair) and pinning a rosette on the dress.
There isn't any wonder why the Sydney school of graphics arts building at the side of the UTS has long been completely bulldozed.
Once I showed Australian consolidated press that it was possible to create an entire cover exclusively on Photoshop (when they introduced vector graphics, first major mag publication in Australia)
They no longer need the million dollar DaVinci prepress systems any more.
Dinosaurs of the past. Scitex prepress systems etc.
I recall being amazed at seeing the new scitex blaze at our workplace.
Paintbox like, but prepress specific.

MNCPMSteve
Автор

As a wedding photographer, this is going to be incredibly helpful.

JamieChorley
Автор

I used it today to add more toys to A child's bedroom, I shoot real estate photography. I was also able to add a few more pictures to the wall and plants around the living room to stage the house a little better. super excited for this tool

pixelprovideo
Автор

As Ansel Adams said "you don't take a photograph, you make it". That quote has taken on a whole new meaning.

zoojarlakdawala
Автор

I was messing around with it for the first time yesterday. Had an eerily lit underground parking garage scene with overhead fluorescent lighting. Told it to add a "parking garage toll booth" while expanding the width. It dropped in a pretty convincing kiosk, complete with matching eerie overhead lighting. Mind blown.

norwegianblue
Автор

The workaround to get better detail and clarity in sections, such as generated grass, is to only generate smaller sections such as left or right side one at a time (or quarter sections) and the refined detail will be less fuzzy and more detailed. Other reviewers have shown that works.

adrianvanleeuwen
Автор

The reason it is struggling to generate enough resolution is that it works on max. 1024 px on the longest side. So, if you have a high resolution photo, that just doesn't fit. You have to do that in small junks of max. 1024px length. Overlapping a few pixels is giving you a better result, so consider that. You have to do it in small pieces till you have filled the empty space. A bit more work, but the results are awesome.

Bee-KL
Автор

I don’t think we should call it photography any more. Basically just digital creation. This could push creatives back to film.

PappaMike-vcqv
Автор

As a photographer, I'm not concerned, but rather encouraged because the thing that this allows me to do is focus on more high quality people interaction and worry about editing the photo later. Also, I'm not the greatest at photoshop, so having something be able to do something that I wasn't passionate about in the first place is really exciting.

TheMichaelKuo
Автор

As a company You can never go to far when it comes to giving your clients more than they expect. Somehow we all knew this was coming in one way or the other, following development of imaging software. I think it is awesome, and I love it. However, using an image for evidence of anything becomes more and more impossible, the old days where we said a picture never lies, will never come back.

carlkristensen
Автор

I’ve been a professional commercial photographer for over 50 years. Some time ago my experience with 2D and 3D CGI radically changed my practice as I was enabled to create photorealistic imagery without a camera. For all intents and purposes, I’ve moved my physical studio into a virtual one which provides spectacular advantages over traditional camera work. Now with AI that world has become even more vast. I see all these tools as providing we artists incredible potential. AI is no more a threat to creativity than Postscript, clipart or all the other tech-based tools we’ve come to accept and employ. AI is just another way to make a mark.

timforcade
Автор

I think that this will only effect the parts of photography that have to do with art/advertising. People will still always want to hire photographers to capture their real life moments like the photographers at Disneyland or at a wedding. They will always be needed but it might effect the way that companies can create corporate media

j_akae_
Автор

When you select the crop tool, use the Generative Expand in the drop down menu top right of the screen. Using Alt+Shift you can drag a corner out which exposes all sides or just drag any side without using any keys to drag that part out left, right, down or up to make a selection.

adejones
Автор

On a video by Phlearn he said that it uses Adobe Stock images to fill the gaps, so as Adobe Stock images library grows, so will the AI's ability to find missing parts of images. For example I removed a person from a photo I took at Times Square and it even recreated the STOP letters on the asphalt. I verified another image with the model in another position and the letters were the same as on the second image, so it looks like it took an Adobe Stock image of the same street in Times Square to fill in the gap.

nmelcam