Well, Adobe has gone and done it. Sky replacement is now a featured part of the just-released Photoshop 2021. For about a year, Skylum has been the leader in sky replacement with Luminar 4, but the story isn't that simple anymore.
Adobe first showed off sky replacement at the Adobe Max 2016 conclave. It got, as I remember, much applause and interest. But then, nothing. Now, four years later, Adobe is offering this feature, seemingly playing catch-up with Skylum.
While sky replacement is still a contentious issue, it can mean a lot to many real estate photographers, wedding photographers who do outside events, and some landscape photographers who find it useful at times. The controversy will always be with us, and that's a healthy thing, but how does the Adobe offering stand up to Luminar's popular sky controls?
That's just what I wanted to find out, so I took an image that has routinely tripped up Luminar because it had trees with many small branches and leaves. It would be very difficult to do a sky replacement by manually masking in a new sky with all this vegetation, so automation offers hope for an image like this.
How Do the Two Programs Compare?
Here's the original image of the location:
And here is a closeup of the trees that can be problematic for auto-masking software:
Luminar has some issues here, and while one can use the Close Gaps control to help, it still has problems filling in between the leaves and the branches. You can see the areas where it is less than perfect:
I tried the same challenge with the new sky replacement feature of Photoshop 2021:
This was much smoother, although again, I had to play with the edge controls in Photoshop to improve things. If I went too far with the edge controls, the sky image itself was altered, which is not a good outcome.
Here's a look at the sky controls for Luminar, followed by the Photoshop controls.
In the Adobe controls, note the options to reposition the sky image at the left of the panel, and the scale control that's missing from Luminar. Photoshop lets you move the sky image vertically and horizontally. Luminar only allows you to move the sky vertically. It's just not as flexible.
Who Is the Best?
Normally, I'd call it a day and declare Photoshop the winner, but it's not that simple. Both software programs have sliders to spread the new sky color on the landscape, making it a better integration between the original image and the new sky. I found Luminar had a much better option there. I got a far more realistic outcome using the Luminar Relight Scene slider than I did with the Photoshop color adjustment slider. With Photoshop, the effect was subtle or not visible at all.
On the majority of sky replacement tests I did, both Luminar and Photoshop looked about the same. Both let you import skies from your own library, and both let you flip the skies horizontally. Luminar lets you add atmospheric haze and lets you defocus the sky. Photoshop doesn't, but because the sky is a separate layer, you can do that with the existing Photoshop tools. The results can look quite good with both programs, but Photoshop has an advantage where the mask has to integrate with a complicated, busy foreground.
Still, I've had excellent results with Luminar when I've needed it, like this photo in an Arizona ghost town. It's not perfect around some of the detailed metalwork, but it beats manual masking.
Neither Photoshop nor Luminar do water reflections yet. In both cases, you'll have to manually insert them into bodies of water by making a new layer and creating a mask. However, Luminar has announced its new Luminar AI update shipping late this year will do sky reflections in water, and that's a pretty big deal when you need it.
Photoshop sky replacements are done in layers, making it easy to readjust image values. Luminar drops in the sky, and you are left with one layer. If you don't like what you wind up with, you have to start again with adjustments, although Luminar does allow you to go back and insert a different sky before you save the image. Photoshop does the same. A new sky becomes a new layer, which can be kept or removed. Overall, Photoshop offers more flexibility, while Luminar is striving for simplicity.
Of course, Adobe could add that feature as well, just as Skylum could ship Luminar with more options to reposition skies.
It's pretty clear many Photoshop users were clamoring for sky replacement, and Adobe has answered that need. Luminar isn't standing still either, and Luminar AI, when it ships, could leapfrog Adobe. And both programs will surely evolve from here.
Since Photoshop is subscription-based, it was great to see this feature pop up as part of that plan. Luminar is something you will have to buy, and Luminar AI, which will include a more sophisticated sky replacement feature with water reflections will be another expense for Luminar 4 owners. It's a justified sore point, and I wish Skylum would have simply updated Luminar 4 with the water reflection feature.
Both programs beat manually masking a new sky in by a long way. I'm impressed with both applications, and I'm hoping Luminar will offer better sky placement tools, and I'd like to see Adobe offer better color-matching and water reflection options.
Sky replacement. Why not drop photography and use an algorithm replacing sky, people, surroundings to peoples calculated liking. Fake everything! I think there should be a clear distinction bethween fake smear and photo. This is not painting with light, it is just computational lies posing as photography.
I think there should soon be a limit for what could be presented as photography. All this fake junk serves no other purpose that to con ppl. All this colorful sunsets, the perfect skies, the people without any marks on their skins; it’s about ethics.
I haven’t actually set any limits/borders, other than pointing out that to have a «sky replacement» tool is way out of what I consider limits. It must be done through discussions. I think the disasters of fake news could be a climate change in this discussion. Stalin edited out ppl from photos in order to get the story right, and that is another limit for me.
Personally I change exposure, ISO and contrast without exceeding my selimposed limits. I never retouch/edit out pimples or molds etc, and never use extra light, though extra light is not a problem for me.
Even straight out of camera is a digital interpretation; it may not see what the eye sees. This idea of authentic vs. fake is strictly speaking, fake :)
When you are selling your house, do you wait for a day with beautiful skies so the real estate photographer can take good photos? Or do you just advertise your house with photos taken on a rainy day? When I'm selling my house, please come photograph it ASAP, don't delay, and then please replace background with something that helps to sell it.
Oh. And if there's a dog shit on my driveway, please clone it out, don't be a purist...
Yes. I do not leave it up to the real estate ppl to photograph it. Photographed and sold my houses several times. But I am able to plan ahead of selling, so please do not worry about me. I adjusted contrast and exposure afterwards. Nothing more. All of them sold for higher prices than estimate from real estate broker (I photographed, broker sold)
I have a feeling that if there was a Venn diagram of people who comment about the destruction of the purity of photography on this site, with people who don't have any of their work posted to their profile, it would just be a circle. Every single time. If you're achieving top quality work with zero post processing, you should be sharing that work with the rest of the world!
Maybe. Maybe not. I think this is a question about defining photography, not about sales. I would also think many curators would disagree with you regarding the correlation about quality and posting of photos posted here.
I never said anything about sales or curators. Simply that people who feel the need to gatekeep what is & isn't photography on every editing software article never have their own photography posted to their Fstoppers profile. Or a link to their work.
You would also know that most of the academics and curators discussing such topics are mediocre photographers, or not even photograph much themselves. To define this is not a matter of photography skills, but rather analysis, ethics, and theory. Roland Barthes is an example.
As a technology geek I find this fascinating but as a photographer, I’m quite appalled and insulted. This current obsession with sky replacement reminds me of the over-the-top, garish HDR fad years ago. I’m not at all interested in using it but if it makes you happy, have at it.
I agree that this has the makings of just another fad, but I'm also keen to see how many images I've got where the subject was great, but the sky less than ideal. The deciding factor will be the application in the end.
If it'll allow me to save a so-so image, I'm all for it. Just like other tech such as denoise and sharpening software has allowed me to save older images.
Guys. It’s not blasphemy. It’s not the death of photography. It’s just a piece of software for image enhancement. Like any other tool, in the hands of someone without talent or taste it’ll produce some unfortunate results. In the hands of a pro who understands what a client might be looking for and can use some restraint with the sliders, it’ll produce perfectly usable and attractive results.
No problem for me. Just call it illustration, because people believe a photo represents more truth than a drawing. Photography it aint.
Photography isn't any more pure; it's all a mechanical facsimile of what the eye can see. If you only think, "straight out of camera," is real, you're just trusting Canon's decisions for the illustration.
Great article.
I recently wrote a brief blog post comparing the two applications and I align with your thoughts. I agree that Photoshop 2021 does a better job with masking. Luminar 4 also struggles when there are prominent clouds and highlights in the original sky which can cause unsightly results. http://mostly.photos/blog/2020/10/21/skylum-luminar-4-vs-adobe-photoshop...
It costs thousands of dollars to fly to a location and shoot. We dont care if there are bluebird skies or dark non-descript clouds,,we shoot and hope for the best. However, with sky replacement I am afforded the opportunity to revisit without spending another $1000 or even $2000. This, in my mind, justifies the result. Money and Time.
Sure. «If you can’t make it, fake it». Goes for many areas of life. Still I think it needs to be some boundaries, demands, to call work photography.
And Steiglitz expounded that only straight photography (i.e. pure rendition of reality, sans exposure tricks and perspective distortion) was the only acceptable photography. Else, it was just painting. Fortunately, the f-64 group broke out of the mold and propelled photography into an artistic medium. I believe that even Ansel Adams was not above compositing of photographs in the darkroom when it met his artistic vision.
I think we still call their work "photography". In fact, it is what the purists pine for.
It may not be your taste, but is not the purpose of art to push boundaries? I don't particularly like cubism, but had Picasso not explored it and made it an artistic force, we probably wouldn't have Dali-esque surrealism, of which I am certainly a fan.
I had an artist friend say to me when I was quite young and complaining about the lack of classical artistic technique in abstract art, "well, you can only paint a bowl of fruit so many times". Changed my view completely. Metaphorically, is it not the same with photography? Had we no advances in cameras, lighting, optics and processing techniques from the days of Daguerre, would photography even be of interest to anyone anymore?
No one ever started a revolution by standing behind the barricades.
I have no objections re artistic, creative use. But at one point «painting with light» becomes «painting with Photoshop». You can fake a whole photo with AI. So my point is that given all this tech, it’s a need to draw a defining line somwhere between the puristic view of a photo, and sky replacing software and other blatantly fake alterations. The line between photography and illustration.
I guess I see no argument for a "need" to draw artificial lines to define an activity.
Should we draw lines between what is rock music and what is jazz? If so, we wouldn't have had the Mahavishnu Orchestra, or Return to Forever, or Larry Carlton, or Steely Dan. Or a line between what is "real" music and what is synthesized music? If so, we wouldn't have Yes, and Emerson Lake and Palmer, and Brian Eno, or even Baba O'Riley for that matter.
What a boring world it would be with artificial lines.
So you would accept a picture as a photo, when there only has been a computer involved in the process? No camera, only an algorithm?
I would think it is a picture. If I like it, so be it. If not, I wouldn't look at it. Photos that are too over processed are of no interest to me, and I tend to pass them by. It doesn't bother me that photos can be synthesized because I tend to not like composited photos if they create an altered reality. But if someone composited a photo of Half Dome from a dozen other photos in such a way that it looks completely natural, I would not be able to tell that it was not "real" and have no reason to dislike it, nor reject it. It is only those photos that are obviously faked that offend our sensibilities. If a picture looks fake, then it holds no interest for me, but I don't get angered that someone created it, nor do I think that there should be some community proscription against the act of trying to create it.
Should we deem "Fantasia" to not be a movie because it was illustrated, not photographed using real actors?
I just find that those who try to define boundaries ALWAYS end up advocating some form of censorship or censure. I have no intellectual interest in that sort of regulation because it is ultimately an act of arrogance on the part of those advocating it, in that their view is somehow more "correct" and should be accepted as an inviolate norm.
In the end, EVERY photograph is an interpretation of reality - it is never 100% realistic because the world is not time invariant and a photograph captures an image as time invariant. So, I prefer not to set an arbitrary cutoff line as to how much interpretation is acceptable.
As I have tried to say in earlier posts, I’m not interested in regulations, but distinctions: This is a photo (filling certain criterias). This is an illustration (filling other criterias). Like news: This is verifiable. This is not verifiable. So what constitute a photopraphy should be defined. I have worked with 3D, alterations of photographs, mix photo/computer graphics, and plain photo. Given the development of all this smart sw and algorithms, AI we need some definitions in order to be able to evaluate the value of a photo presented in a news article, a commercial aso is.
The photographic equivalent of lip-syncing. And if you didn't even shoot that sky yourself you're lip-syncing someone else's track.
"Someone else's track" - What if it's your sky too, but from another shoot? Sure that'd be just lip-synching, no? :-)
As another person noted, you may have spent months of planning and thousands of dollars travelling to the photo location. The weather doesn't always work out and while sometimes you can "work with the weather" (been there, done that), that isn't always true (been there, done that too).
What is missing is a wider viewpoint - What is the market or intent? Lip-syncing for an animal portrait, or a non-repeatable landscape image surely is only dishonest where the sky is a main feature instead of incidental, don't you think?
I'd say it's closer to a songwriter vs. a singer -- and that's SUPER common for a singer to perform something with a different songwriter. Do you think Whitney Houston's "I will always love you" is less real than Dolly Parton's?
I appreciate that photoshop creates a mask for their sky replacement - that has usefulness in and of itself for me, even though I don't do sky replacements.
Regarding the larger philosophical and in some case, pointless, discussion happening; no matter where you stand on sky replacements and computational editing, you would have to be dishonest with yourself to think that photoshop and other tools haven't already been used for years to do this, things are simply becoming easier and faster to do so. With that said, we'll have to see but I think its possible that sky replacements could become more saturated on people's IG accounts for sure, and maybe in other spheres of photography. At some point it will become apparent when people are using skies that have been already used elsewhere - no doubt this will cheapen people's photos, and for those that don't do mind doing that to themselves, to each their own.
But i also have an alternative perspective that i haven't seen hardly anyone talk about. WHY is replacing the sky so important to people? I think photography, particularly in landscape, has been shaped by a generation of purist and poshy people who complain about blue skies and wanting more 'drama' * eye roll*. Yes of course clouds and light adds an aesthetic dimension to a photograph, no argument there. However, the attitude that an otherwise amazing photograph might not be portfolio worthy, or that you need to revisit and retry in order to find said drama, is perhaps reason for this as well. In fact many of you complaining share the same perspective, you want a rare-looking hard to get sky, and you value that more than otherwise, so perhaps there is a larger cultural/aesthetic that you 'purists' are accountable for as well.
That’s a good question you ask, and I think it’s because, for landscapes especially, the sky is part of the uniqueness. Like, Half Dome will always be there. But if you happen to catch a rainbow or the “perfect” sunset, that’s what really makes your photo more special than some visitor who went with his family at noon on a cloudless day. So the ability to cheat there is, to some degree, sacrilege.
Personally, I think it’s fine for commercial work. Anything is if you’re being paid to create a specific look. Beyond that, yeah it’s kinda tacky but I’m glad the tools exist to do it all the same.
I’m not particulary obsessed by sunsets and skies, but more in what the photo communicate to me. I have seen so many utterly grotesque fake sunsets and skies, and that ruins any communication. But it’s also a question about taste; if you scroll through Fb the sunsets and skies are from other planetary systems. The more colorful and grotesque, the more likes.
Maybe they get more likes because that is what people on FB like. Does that somehow diminish your appreciation of a good photograph? People also seem to like icky-sweet donuts more than savory donuts. Should one be offended by that?
FB is not the Tate Museum. No one ever claimed it was the repository of the world's great photography and art.
It’s an example of different taste, and also a warning that «what people like» should not be a parameter when we discuss photography vs illustration
The great thing about today's post-processing innovations is that you can achieve your targeted results both quickly and accurately. The other 'great thing' is that with increasing competition, it keeps software designers on their toes...leading to badly needed innovations that are far simpler to execute than ever before.
The greatest ideas don't always come from the market leaders in any segment. Fresh competition broadens the idea pool, forcing all parties to constantly innovate. Innovative companies also do the best job of actively seeking user feedback, realizing that such information provides an opening for competitive advantage.
You mention Photoshop, and the common theme is 'complexity'. Which of course led to the design of Lightroom. There are still numerous opportunities to bring over the tools in PS to LR in a way that greatly simplifies the process...think selection tools for example...and composites without layers, etc, etc. When you read "but now I have to go to PS..", that signals, to me, that's there's an opportunity for simplification. And with PS/LR now offered as a subscription package, LR should cannibalize PS as 'the simple solution' as often as possible. Adobe really has a lot of opportunities with this. It's all about the results...regardless of how you get there.
We all want our personal results. And we want them in the shortest time possible. Time reduction equals simplicity. And the developers who embrace these needs will always come out ahead. You can't fall asleep in his industry...and that's a good thing. For us.