Midjourney and Dall-E have been taking middle stage ever since AI-generated photos grew to become a pattern. Nonetheless, different AI picture mills are additionally contesting to be within the limelight; take Meta’s and Adobe’s AI picture technology instruments as an illustration, which we’ll look into on this article.
We’re conscious that Meta and Adobe are already massive firms and well-known for what they do. However with the expertise advancing, would they be left behind within the race towards avant-garde AI programs? Clearly, no. That’s why they launched their very own AI picture mills.
Now, let’s get to it and see which one is healthier in your particular want
Options
By way of options, Meta’s standalone web site, Think about with Meta AI, doesn’t have a lot to supply besides—in fact—its picture generator. You simply merely kind in your immediate, click on Generate, wait, after which there you might have a couple of picture variations based mostly in your enter. After that, you may obtain the output that you just like. Go create one other if you would like extra.
Adobe Firefly 2 gives a wealthy set of customization instruments that permits you to tweak the output and tailor it higher to your desire. Not like Meta, it provides you management over the next:
- side ratio
- visible depth
- content material kind (photograph or artwork)
- photograph settings (for pictures)
- results
- lighting
- composition
- shade and tone
Not solely this stuff, however Adobe Firefly additionally enables you to add unfavorable prompts and use a reference picture from its gallery or immediately out of your machine. Much less problem in comparison with different AI picture mills, proper?
Once you click on on a picture, you’ll discover one other set of enhancing instruments that you need to use to change the output additional. You too can use your chosen picture as a method reference for the brand new output variations you’ll generate. When you’re okay with the end result, you might have the choice to obtain or save the output to your favorites.
To date, these are the present options of Meta and Adobe Firefly 2 we are able to examine, although there’s undoubtedly much more to count on after they launch a brand new model. For now, let’s reply one of many questions you could be having in your thoughts: How accessible are they to customers?
Pricing
Relating to the worth, the excellent news is that Meta’s AI picture generator is free—no billing, no subscription, no cash concerned in any respect. I do know that it sounds good, however have you learnt that every one AI photos that Meta creates for you might be mere scraps of information from actual pictures uploaded by actual folks on social media?
In truth, Meta was educated on 1.1 billion pictures from Fb and Instagram. And the place did all these pictures come from? Customers—us. I do know this could be a surprising revelation, but it surely’s true. Maybe that’s why it doesn’t cost a single penny in alternate for utilizing actual pictures as a coaching dataset.
Despite the fact that Meta is free, one other factor that you must know is that it’s at present accessible solely within the US. So, should you’re not from the US, you could want to make use of a VPN and connect with a US server, which will not be free. However, Adobe Firefly 2 gives totally different premium plans. Try the pricing and bundle inclusions under.
Fortunately, there’s a free plan. So, that’s excellent news if you wish to use Firefly with none prices, in contrast to different Adobe software program that often requires cost. Be aware of your credit, although, as you solely get 25 monthly should you’re utilizing it totally free. Identical with Meta, your picture generations should not limitless in any respect, so use them properly.
Output High quality
Now, that is the place the actual battle begins. All the pictures on the left facet are made with Meta, whereas these on the proper facet are made with Adobe Firefly 2. Additionally, take observe that I didn’t edit any of the Adobe Firefly outputs in order that the comparability stays unbiased as a lot as doable. The next are the outputs of the 2 AI picture mills utilizing the identical prompts.
Lifelike Portrait
| Immediate: a highschool lady in class uniform, pure gentle, candid shot |
Meta adopted the immediate utterly. Adobe Firefly did too, nonetheless, the lady appears to be like too younger as if she’s in fifth grade quite than highschool. I additionally needed to generate the identical immediate a number of occasions with Adobe Firefly 2 because it stubbornly gave me footage of younger ladies in uniform smiling and trying on the digicam earlier than it gave me this one.
This output of Adobe Firefly nonetheless incorporates a smiling lady, however a minimum of it lastly understood what candid actually is. I nonetheless choose Meta’s output although, since one facet of the lady’s hair by Adobe Firefly is just too hazy when there isn’t even a bit of quantity of haze on her face, making the hair appear like it’s within the background. The sunshine reflection on the chin has no obvious supply as nicely.
| Immediate: a younger man with tousled hair, streaks of pure gentle accentuating his hazel eyes, vogue |
Similar to within the first immediate, I had a number of makes an attempt making an attempt to get what I had in thoughts utilizing Adobe Firefly 2. I obtained a bunch of outputs that didn’t match any of the descriptions in my immediate besides the topic (a person) earlier than I had the best photograph above. The hair appears to be like styled although, and if we take note of the blurring, we’ll see that it’s not constant.
Meta’s output, then again, is nearly photographically good however not in a superficial means and exhibits what touseled hair really appears to be like like. The pure gentle amplifying the nice and cozy shade of the hazel eyes additionally makes this output a powerful one, although some strands of the hair nonetheless want some work. Aside from that, this output of Meta is nice for a Vogue cowl journal.
| Immediate: a redhead lady with blue eyes and freckles within the grass area, golden hour |
As you may see, the lady within the output of Adobe Firefly 2 has hazel eyes as a substitute of blue. In the meantime, Meta’s output obtained all the things proper—redhead, blue eyes, freckles, and the background particulars.
Additionally, though Adobe Firefly’s output appears to be like extra photorealistic with finer particulars, it doesn’t seize the entire essence of the immediate. It gave me a smiling lady (once more) when I didn’t point out any phrases that relate to that, whereas the lady’s expression in Meta’s output appears to be like extra pure, which is lots nearer to the end result I used to be going for.
The one factor I’d change or take away from Meta’s output if I might is the bluish-silver nostril piercing that oddly stands out like a glow stick in opposition to the general heat tone of the picture. Perhaps, I’d repair the form of the iris as nicely. General, this output is a chef’s kiss.
Basic Observations (Portrait Version)
One extremely noticeable attribute of Adobe Firefly 2 is that it has that smiling syndrome when coping with portraits. It has the tendency to overhype the emotion and facial features of the human topic (a number of occasions) robotically by default when requested to generate pictures of an individual.
Now, you may even return to the primary set of photos above to see it your self. See how they’re all smiling? You may suppose it was simply three pictures however take a look at this one other article the place we in contrast the outputs of Adobe Firefly with Midjourney. Brace your self as you’ll be greeted with extra smiles by Adobe Firefly.
By way of high quality, Adobe Firefly 2’s photos have larger resolutions and finer particulars that make them appear like they have been shot by an expert digicam, whereas Meta’s photos have softer edges and diminished however not essentially poor resolutions like they have been simply taken utilizing a cell phone and uploaded pictures in social media. Effectively, if you realize, you realize.
Panorama
| Immediate: a serene view of a sunflower area, rainbow within the sky |
Whereas each outputs seize the serenity of the view, in addition they present a transparent signal that they’re nothing actual however artificially made. What I imply is that they have repeated patterns or objects that appear like they have been simply replicated or copy-pasted. See how Meta’s sunflowers are all oddly dealing with towards the digicam and the slanted positions of the leaves in Adobe Firefly’s output?
At first look, Adobe Firefly’s output appears to be like extra photorealistic because of the perspective however should you look nearer, you’ll discover the eccentric form of the timber within the background making them seem as bushes, and a few rendering points. It’s extra correct than Meta within the rainbow half, although.
| Immediate: ethereal view of aurora borealis |
Irrespective of how majestic these pictures are, there’s simply one thing there that doesn’t look proper—sure, the reflection. The northern lights don’t replicate the best way they need to as you may see, particularly in Adobe Firefly’s output. Whereas Meta has managed to make it much less apparent, I don’t suppose any of the 2 realistically depicted how lights work together with their environment.
| Immediate: pastel skies and sakura timber |
Each outputs are lovely. Meta’s output appears to be like like a digital portray whereas Adobe Firefly’s appears to be like like a standard one. Each emanate a tranquil ambiance as nicely.
We are able to all agree that Adobe Firefly produced a formidable output, however I didn’t ask for any physique of water in any respect. If I’m to nitpick this output, I’d additionally say that it has the next likelihood of getting flagged as an AI-generated picture as a result of repeated patterns within the timber, which generally is a useless giveaway that it was made with AI.
However, Meta’s output was extra simple. It exhibits me what I solely wished to see: pastel skies and Sakura timber. The sky hues are additionally extra on level and the view is extra cinematic. The one fault I’ve discovered right here is the sunshine blue portion on the decrease a part of the picture, which I might barely determine. However general, this appears to be like like one thing I’d cling on my bed room wall.
Basic Observations (Panorama Version)
When supplied with parts that may be merely replicated, Meta and Adobe Firefly 2 unsparingly took benefit of that. Like different typical AI picture mills, these two considerably suffered from lazy replications of repeatable patterns. However what can we count on? These are AI and that’s what they’re meant for—make issues simpler and sooner.
In addition they lacked some logic relating to reflections. Irrespective of how breathtaking the views they generated, it wouldn’t simply make sense if the interactions of pure parts weren’t correctly illustrated.
Surreal Artwork
| Immediate: the world inside a cup of espresso, 4 whales circling over the cup |
Regardless of the 2 AI picture mills producing a well-crafted masterpiece within the surrealistic side, if we follow the immediate, Adobe Firefly was the one which gave a extra nuanced output. The delicate distinction between Meta’s and Adobe Firefly’s output notably lies within the dolphins as parts, which I do know you may inform by evaluating the 2 artworks.
| Immediate: a cityscape with saturn seen within the sky, rainbow in spiral type, floating pyramids |
These outputs are visually interesting in their very own good methods, however once more, they’re lacking some particulars. Apparently, what we see within the foreground of Adobe Firefly’s output are mere roofs of towers formed like pyramids; the one surreal factor on this output is the unnatural phenomenon within the sky.
Meta illustrated a literal, otherwordly, large pyramid amidst town, which locations itself one level forward of Adobe Firefly 2 when it comes to accuracy. However it’s nonetheless not totally correct because it solely generated one pyramid after I clearly specified it in plural type, and since the constructing constructions are quite off—not actually suggestive of a surreal setting.
| Immediate: a white cat in house swimsuit floating in house, stars and galaxy |
I like each outputs actually when it comes to artwork model and high quality, apart from the truth that I personally like cats. Meta’s output appears to be like like a sticker that may even look good as a print on a t-shirt whereas Adobe Firefly’s output appears to be like like a cute poster that deserves a spot in my room.
By way of accuracy, I’d choose Adobe Firefly 2 for its potential to image the small print I’ve on my thoughts, besides the “floating” half is extra delicate in Meta’s output. The rationale I didn’t select Meta was that it appears to be like just like the supporting particulars are separated from the remainder of the immediate parts (stars remoted from the topic as in the event that they weren’t in a single picture in any respect).
In that case, it looks like Meta handled the stars and galaxy as some form of separate parts quite than as a part of the entire context. Perhaps that’s why it appeared like a sticker, to not point out the thick white define. Nonetheless, as I’ve talked about, I like each outputs and I would think about this a tie if solely I wasn’t being any extra goal.
Basic Observations (Surreal Artwork Version)
Whereas there isn’t any massive difficulty with how these two AI picture mills handled surrealism, what I’ve seen is Meta making some particular parts of the immediate look weird. For those who additionally seen, it sort of modified the genuine appearances of particular person parts (such because the dolphins and the buildings) in an try to make the general content material look much less or not actual.
That wasn’t efficient for me although, because it solely made the dolphins appear like another species quite than a part of a surreal artwork. On a optimistic observe, each Meta and Adobe Firefly 2 have astounding ranges of creativity.
Business Images
| Immediate: a product photoshoot of a pair of high-heeled sneakers |
Moreover the poorly rendered shoelaces, you’ll discover that the reflection of the pink sneakers by Adobe Firefly 2 doesn’t match the sneakers’ precise design. For those who additionally seen, the background was inconsistently blurred and sharpened. Shifting on, I’ve no points with Meta’s pair of sneakers; it’s normal-looking—nothing actually unusual happening there.
| Immediate: a industrial shot of lipstick, stylish model pictures |
The outputs of Meta and Adobe Firefly 2 are each wonderful mockups of a lipstick product. Nonetheless, the roundish lipstick cap by Adobe Firefly—from what I see—received’t appear to suit the closing of the square-shaped base of its lipstick tube. However apart from that, this output undoubtedly appears to be like prefer it belongs to a magnificence and beauty brochure.
| Immediate: a product {photograph} of sweets, gold foil, a glass of milk, purple background |
Once more, each AI picture mills did a very good job on this immediate. Each component talked about was there and proper, as you may see.
For those who look nearer although, aside from the nonsensical phrases on the chocolate field, you’ll additionally discover some rendering points with Meta’s output. It’s not that dangerous, however the best way it was rendered makes it look extremely computer-generated; briefly, an apparent AI work.
Why do I say so? It has that superficial high quality, particularly the texture exaggeration that makes it look not actual when examined. Simply see how tremendous clean the objects’ surfaces are. The sides of the field, plate, and glass are additionally ragged. In the meantime, Adobe Firefly’s output appears to be like extra genuine.
Basic Observations (Business Images Version)
It appeared like Meta critically paid consideration to the key phrases “product” and “industrial” right here, therefore it generated mockups of merchandise with model marks on them, such because the model identify on the chocolate field, the engraving on the lipstick’s floor, and the emblem on the shoe’s tongue. Regardless of the random texts, this exhibits Meta’s broader understanding of the immediate’s function.
Brand
| Immediate: an icon brand for a car-manufacturing firm |
I’ve actually nothing in opposition to these outputs—each logos will certain look good on a contemporary or classic automobile mannequin. However to be truthful, Meta’s brand can solely be good for brief model names.
| Immediate: a easy aesthetic brand for a espresso store |
Once more, each AI picture mills did a very good job on this one, regardless that it took Adobe Firefly 2 a couple of extra makes an attempt to provide you with its output. Undeniably, the simplicity is there, however I’d go for Adobe Firefly’s brand for its extra simple high quality (since Meta’s brand has qualities that make it look extra like summary); additionally, only a private desire!
| Immediate: a minimalist brand for girls’s clothes line |
I do know Meta isn’t good if we take a more in-depth take a look at its particulars, but it surely’s a lot nearer to how a minimalist brand ought to look—easy parts which might be non-photographic.
Adobe Firefly’s output isn’t that dangerous, however is certainly not good for a brand. This might go if it was performed in a vector artwork model with strong colours and crisp outlines. It could additionally look higher with out the unreadable textual content under the flower brand. As for the colour scheme, each logos give off a classy really feel and are acceptable for a high-class girls’s attire retailer.
Basic Observations (Brand Version)
Similar to within the industrial pictures version, Meta thought-about the enterprise context related to some key phrases within the immediate—the time period “firm” from the primary brand immediate, for instance. Because of this, it creatively positioned the fictional model’s initials DLHLr, which made it appear like part of the car-shaped brand.
Concerning Adobe Firefly 2, if I’m being sincere, once more, it disregarded the general context of the prompts at first earlier than it produced pretty good and contextually acceptable logos. So, the primary outputs have been actually full of colours and particulars, which have been fairly reverse to the model of logos I used to be asking it to generate. They have been creative although, however nonetheless, they simply weren’t match.
Summary
| Immediate: an summary picture of infinity |
Wonderful depiction of infinity for each, but when I’m to emphasize extra the actual which means of infinity, I see the qualities higher in Adobe Firefly’s work—it extends to all instructions (just like the limitless sky), delineating its boundless nature, whereas the infinity in Meta’s illustration is simply one-way, displaying a restrict in dimension. With out going deeper, I’d say one is simply as nice as the opposite.
| Immediate: an summary picture of euphoria |
These are good representations of euphoria—colourful, vibrant, intense, and explosive. Each Meta and Adobe Firefly 2 confirmed their talents to grasp the connection of the colours and kinds with the given emotion, so I give each two thumbs up.
| Immediate: an summary picture of knowledge |
I don’t have something in opposition to these outputs if we’re solely speaking concerning the visible enchantment. They’re each very good and placing in their very own methods, and in the end artistic relating to decoding the identical idea.
Nonetheless, Meta’s output considerably leans towards symbolism (overtly utilizing an owl with a e-book as a topic), although it nonetheless has noticeable summary parts. Adobe Firefly 2, then again, understood the project and took summary compositions into consideration, giving us a extra profound illustration of knowledge.
I might nonetheless acknowledge the torch although, however nonetheless, the best way it’s performed is much less direct and the ultimate output is extra summary than that of Meta.
Basic Observations (Summary Version)
All I can say is that each AI picture mills are equally good picks should you’re simply aiming for creativity utilizing AI. Each of them are just about geniuses in their very own interpretations of various summary ideas, so kudos to Meta and Adobe Firefly 2 for this immediate version.
Excessive Context Prompts
| Immediate: a worn-out bed room stuffed with mud, webs, molds, vines, and timber rising inside, set in the way forward for a desolate world the place humanity not exists, a couple of indicators of life a thousand years in the past, cream wall paints peeled off, woods damaged by time, pure gentle, heat palette, subliminal peace suggesting a perennial silence, a way of nostalgia |
Personally, I discover Meta’s output extra hauntingly lovely, but when we think about the accuracy based mostly on the immediate, the output of Adobe Firefly 2 is extra nuanced. The spider webs and nature breaking in are rather more seen in Adobe Firefly’s output than in Meta, and by some means, Adobe Firefly retains the delicate heat of the setting’s ambiance.
| Immediate: a polaroid {photograph} of a three-tier cake with intricate piping particulars and pink flowers made with ganache on prime of white fondant base, white chocolate balls, sweet canes, and strawberries as decorations, two champagne glasses and some cassette tapes on the facet, curtain backdrops with fairy lights, retro model, grainy high quality, shot in 1980’s |
None of those two appears to be like precisely like a Polaroid {photograph}, however a minimum of Meta has a bit of little bit of a classic really feel to it. If we think about the completeness of the weather, we are able to agree that Meta was the extra obedient one too though there are some rendering points, notably with the sweet canes.
In the meantime, I sort of struggled to get Adobe Firefly 2 to generate the output for this immediate (to the purpose I craved cake) with out continually giving me main points with it; it hardly gave me outputs that had sweet canes or cassette tapes and if it did, it might miss another immediate parts. Adobe Firefly 2 was like a bucket with holes; it stored lacking some particulars.
Funnily although, whereas it persistently ignored my point out of cassette tapes, Adobe Firefly 2 proceeded to offer me an output with a slice of cake in it (the one I used above). Belief me, it was already essentially the most compliant and essentially the most first rate out of all of the outputs it generated. Was it to compensate for the dearth of cassette tapes within the image?
Additionally, it didn’t even put an effort into creating an intricate sample on the cake regardless of it being one of many main supporting particulars of the immediate; it was just a few drips, to not point out it wasn’t even neatly performed. Whereas the general rendering is cleaner than that of Meta, it’s evident that Meta is extra attentive relating to particulars.
| Immediate: an atypical younger man surprised and awestruck having an encounter with a radiant younger fairy in a moonlit mystical forest, bioluminescent flowers, orbs, and butterflies round, fantasy, anime model |
Now on to the ultimate spherical, which one do you suppose is the winner? I do know each outputs have what it takes however first, take note of the small print.
These photos are nothing near an anime, which clearly exhibits that each AI picture mills missed the final supporting element of the immediate. However did you see the wings on the left facet of the younger man in Adobe Firefly’s output? Additionally, as a substitute of trying surprised and awestruck, the person appeared extra like he was making faces in entrance of the fairy, which I discover ridiculous.
Moreover the wings and the face, their fingers should not rendered correctly. On one other observe, Adobe Firefly paid consideration to the flowers as supporting particulars higher than Meta did. There’s not a lot of a difficulty with Meta, although—just some within the rendering of the fairy’s left shoulder.
The person’s face in Meta’s output—although not precisely displaying a surprised and awestruck expression—is extra acceptable for the immediate’s context as nicely. And I do know, you agree that that is by some means extra first rate than a person making faces with a fairy.
Basic Observations (Excessive Context Prompts Version)
I seen (not solely on this immediate version) that Meta was extra correct relating to the particular options of people as a component of the immediate. However, Adobe Firefly 2 was in a position to pay extra consideration to the person supporting particulars of the immediate when coping with locations or settings.
Ultimate Ideas
If we’re solely contemplating the wealthy options and sources, there’s little doubt I’d go for Adobe Firefly 2. Given its customization and enhancing instruments, it clearly provides you extra energy and management over the outputs. That is additionally my go-to AI picture generator relating to straightforward accessibility. You may surprise, Adobe Firefly isn’t totally free so why not Meta?
Effectively, Meta is free however much more problem, talking from my expertise. Not solely is it at present unavailable outdoors the US, however its credit score restrict can be not clearly outlined in any sources on-line (a minimum of, on the time I investigated), so that you may as nicely run into a difficulty the place you out of the blue received’t be capable to generate photos in 24 hours.
Lastly, as to the output high quality and accuracy, Meta fairly pulled it off lots higher regardless of the dearth of management parameters (maybe the character of its coaching dataset performs an enormous function right here). It paid nice consideration to the supporting particulars of the prompts, particularly within the nuances of human expressions.
So now, again to the query: which AI picture generator is healthier? Effectively, the reply will depend on your desire and state of affairs. For those who dwell within the US or extremely worth accuracy greater than the rest, give Meta a strive. However if you would like higher management and are keen to pay for extra perks, go for Adobe Firefly 2.
Want extra data to make your choice? Try our complete Meta and Adobe Firefly 2 assessment to be taught extra.