Adobe FireFly Destroys Dalle-3 and MidJourney! Ultimate Comparison...

a few days ago I made a video about Dolly 3 becoming a part of Chad gbt and at that time I discovered Dolby Firefly and it's seriously blew my mind can it really be the best AI image generator so let's Clash three Giants Dolly Firefly and mid Journey the first request will be to generate a transparent crystal chandelier eliminated from within casting intricate shadows and a textured wall Firefly will be the first one to Maus hopefully I will input The Prompt and wait for it to do it work Bam Bam Bam and it's done and I can instantly see how good the image looks especially the first two images Adobe seems to really follow my instructions the chandelier is really Crystal and casts intricate Shadows but following the instructions is not the best part about this AI also on the right there is a settings menu where I can customize the output I can change the model or aspect ratio I will set mine to Portrait there I also can choose the type of content that I want to get photo or art I can also ramp up visual intensity making images even closer to the prompt I'm not sure that changing the visual intensity helps in any way but at least it's there another interesting part of what Adobe Firefly is its effects this is the only AI that offers deeper customization of results done in a simple and intuitive way to change the visual look of the images all I have to do is choose the effects I like I for example want to see the bul effect in 3D Doby also allows me to choose color tone lighting and composition I want the prevalent color to be golden and lighting as golden hour in the composition section there are a ton of options to choose from such as close-ups shallow depth of field wide angle and so on and this is still not everything can do with Adobe Firefly I can manually choose the aperture of the photo shutter speed and field of view this basically allows you to emulate any camera lens available given a huge creative freedom in advanc settings I can give a negative prompt and to describe elements I don't want to see here is how the images look with all the the applied effects I think they all look absolutely gorgeous I like the style and aesthetic the colors are really good and the BOK effect looks super nice and natural using dolly is much easier in chbt I just need to switch to Dolly 3 and input my prompt in an input field then Dolly takes a few seconds to thank and GI four Images just look at them this Chandelier looks absolutely photorealistic like it it's just a photo and not a generated image this is simply incredible very unexpected results such a difference in results is caused by the way these AIS were learning Adobe used only stock images and it usually causes some parts of the image to look Supernatural and some unnatural dolly was training everything they could get their hands on and most of the time the results are super good but we still have to see how mid Journey performs right now I will ask Dolly to change images to Portrait orientation golden hour and apply boka effect just look at these results the first image looks like a photograph the wall has a ton of realistic texture and each individual piece of a Chandelier looks like a real crystal also for each photo I can copy The Prompt that AI has created for it allowing me to easily replicate the same photo over and over again or just learn how prompts are written as for the mid Journey it is far less easy to use than the other two for each image I need to write a custom prompt starting with imagine which is not really convenient and as you can see the results are not ideal they are too artistic too unrealistic and even if I ask it to be more photorealistic the images still look unnatural consequently when I add golden hour and the B effect the images look even less realistic I think M journey is more suited for artistic things and not realistic it is good at creating artwork and not photos now let's take a quick break and talk about this video sponsor Riverside Riverside is the perfect service for everyone who wants to start a podcast but doesn't want to bother with editing requesting footage and so on Riverside simply simplifies everything you create a studio send links to people and start recording from your side setting up is easy click new studio name it choose between recording audio and video and select the language for transcription among dozens of supported ones the interface looks like a typical meeting app but with a few key differences a big red record button starts a proper recording with separate tracks for video and audio of each participant you can also select your microphone and camera that you want to use which is super handy Riverside supports up to 4K webcam footage which automatically uploads to the cloud Riverside offers additional features like a teleprompter script only visible to the host a chat feature and a media tab with sound effects like clapping after recording you can edit everything in Riverside too you can enable or disable audio tracks change speaker layout Riverside also creates an accurate transcript that you can edit if needed the same transcript can be used for for creating captions directly from the editing page and if needed transcripts can be downloaded as a text file or SRT format or simply copy export options are also flexible you can export an up to 4K resolution and include features like audio normalization and background noise removal you can even export projects directly to Adobe Premiere a unique feature is generating Magic Clips which uses AI to compile the most interesting parts of a podcast this makes creating content for Tik Tok YouTube shorts or Instagram re incredibly quick in summary Riverside makes podcast and interview creation straightforward and efficient use the promo code AI Master to get Riverside with 15% discount on a personal plan and now let's get back to the video how does the dolly perform let's try the second prompt and see whether the Dynamics change now I want to see an outdoor Bazaar at Twilight where each vendor stall has unique items and all the people have detailed faces and clothing I I will also set the aspect ratio to WID screen and you know what these images from Firefly look absolutely fantastic that is exactly what I was hoping to get it is a relatively close-up shot of a small and Narrow Street with shops and each stall indeed has different items the people and their clothes look quite realistic if you don't pay enough attention and in the last photo it just nailed the face come on it's basically a photograph at this point now let's turn on the hyper realistic effect at the dramatic lighting and see what we get these results are a bit less photorealistic but I do like the style this time it wasn't able to create a nice face as it did before and so the woman looks a bit strange hope I won't get nightmares after this for the dolly I will input the same prompt and already asked it to do it in a wides screen and the results it gives are really interesting I was expecting to get something like from Firefly but these images are something different they're not really far from my request and some stalls have identical thanks but I like the artwork and unusual twist and here are the same images but hyper realistic with dramatic lighting I think these ones are probably the best versions of these images we can get but let's take it even further and ask for close-up shots now this is properly amazing you know I've been playing with AI generators for a while now and this seriously impresses me this looks almost like a photo with a filter on top of it especially this one with a lot of people in it if someone said to me that it was a photo and not the AI generated image I would have believed it straight away even the individual hairs on people's heads are visible this is absolutely mental and how about M Journey well its results are again super artistic it decided that bizaar should be Arabian and really sticks to this style however despite me not really liking the results from M Journey so far it still has a few features that I like under each generated image there are buttons and they either upscale the photo or generate the alternative version of it and this upscaling is seriously good after the image is upscaled I can create variations of it or change individual parts of the image let's say I want to change the stall assortment to spices I just select the area of the image and write a prompt then it gives me four versions to choose from all that's left to do is just upscale the one I like the most the third prompt will be artistic so mid Journey should be able to redeem itself I want to see steampunk citycape with moving gears flying airships and diverse character interacting in the foreground fireflies results are really stylish but upon closer inspection they lack a bit of clarity let's add a couple more effects I'm feeling digital art art no cinematic surreal lighting and toned image these results look a bit better but not exactly what I'm looking for how about Dolly I think these images are Miles better in terms of following the prompt I cannot say they look exactly like steampunk but they definitely are gears and airships let's ask to make everything more photo realistic at surreal lighting 4 to3 aspect ratio and wide angle and these images look even better so far Dolly shows the best results the same prompt in mid Journey yields quite unusual results mid Journey decided to make an emphasis on airships instead of the city and people with some additional tweaking the most interesting and stylish photo looks like this I cannot say that I'm a fan of these results but in terms of preserving the steampunk aesthetic M Journey definitely did the best job from what I see dolly recreates by far the most accurate images and follows instructions the closest and it is really easy to use if you are an inexperienced user all you have to do is just chat with it and tell it what you want to see mid journey is by far the worst one it can create stunning images however the usability is terrible writing a separate prompt for each image and each change is tiring and timec consuming you really have to be smart about it Doby Firefly seems like the perfect balance it does not require knowledge like Journey but operates in a different level than Dolly and its effects and presets make it more versatile than other AI generators so if you want to bit more control over the results Firefly is your guy but wait there is a hidden secret about it because generating images is not the only thing it can do Firefly can also edit your images and change parts of it to do this on the main page you have to click Genera a fill then it is as simple as uploading the picture selecting the areas you want to change and typing in what you want I can give myself a unicorn horn or headphones or sunglasses however I see that it has trouble with logos text and giving me strong muscles somehow and you know I was really hoping for the last one to work since it would be much easier than going to the gym well aside from that it is a very good and simple to ous feature it's not as advanced as the one in Photoshop but more than enough for the majority of people and simple edits and there are also text Styles AKA custom fonts creating the font is is super easy you just type in the textt that you want and Enter The Prompt described in the style then you just let AI do the job under right in the menu I can choose the shape and how it follows it font background color and choose any sample style I think it is a very interesting way to spice your posters and social media with uh interestes and font overall I'd say that adobe Firefly might not be the best AI to generate images but it is one of the easiest to use and very convenient in its results here you don't even need to type that much stuff as you have with Dolly 3 you can simply choose from the menu what you want to get and all those options are already optimized so you can be sure the results will look great great AI can't wait to see what will get added in the future updates and thank you guys for watching and see you in the next one

Channel: AI Master

Views: 6,845

Keywords: adobe firefly, adobe firefly ai, adobe firefly beta, adobe firefly tutorial, dall-e 3, dalle 3, dalle 3 ai, dalle 3 tutorial, dalle 3 chatgpt, midjourney, midjourney ai, midjourney tutorial, firefly vs dall e vs midjourney, ai tools, ai tools comparison, best ai image generator, midjourney alternative, midjourney for free, adobe firefly vs midjourney, best ai, ai master

Id: ObTeoFj9p2s

Length: 11min 47sec (707 seconds)

Published: Wed Oct 25 2023