Adobe Firefly Tutorial. BEST AI UI YET!

Key Takeaways

Hello friends in our last video we checked out the unveiling of adobe Firefly however thanks to my good friend the Billy FX who hooked me up with an adobe contact I now have access to Firefly so I'm gonna give you a walkthrough in a tutorial of how it works so let's get started with that so if you were with me in my last video you remember seeing Adobe Firefly site here and this is their generative AI tool for images and now I have access to the text to image here and the text effects and we're actually going to check the text effects first because that's their most unique feature and something that no other generative AI tool does well yet not my journey not stable Fusion not Dolly Adobe Firefly is actually the king when it comes to text you can see some of the examples here where there's just letters but you can actually generate for yourself and that's where you have the little input text box down here at the bottom so let's say we want to write something I'm going to steal this from my friend Matt who also does videos on YouTube he used the word subscribe dripping chocolate and we're going to press generate and now we have here the word subscribe and dripping chocolate and it's generating here now it gets like a letter at a time and then it pops in but I've noticed one thing when it has the same letters like the s's here for example it just clones them which is kind of well I would like a new generation of reaches but it is what it is and you can actually adapt this so first of all you have the sample effects a lot of samples that you can use flowers wires balloon bread toast Etc then you have how much of the effect here that you will see so we have a tight fit the chocolate here is going to stay closer to the letters and the loose fit is going to mean that chocolate is gonna go out a little bit which can give you much more creative effect here you have actually drips of chocolate coming from around the letters and you also have various fonts let's take the popular one here which is a good condensed Bond you also have the option to just use various background colors so our default here is on a transparent background we can for effect sake let's take this one which is a good fit to brown chocolate and if you're not happy with it you have four to choose from here it actually renders or generates each time you press this so these are not pre-rendered so you can either press one of these or just refresh and get a new one let's try a different one let's try water I'm going to type splashes of water here I'm going to change to how far for another sake what more we have now let's go with the out front we're going to generate down here to get our new input here and now we have our splashes of water and I think this calls for a different background let's take the pale blue here I think this looks pretty good now you can type whatever you don't have to type water and have it in water you can say like my family name here and it starts generating instantly so this is uh the coolest feature in my opinion with Adobe Firefly for now now there are other features that are also very good but we'll get to those in a bit and I think another great feature with Adobe Firefly is the actual user interface and you're going to see that when we look at text to image generation in a bit but just the ease of use you don't have to know anything you don't have to install anything it's super easy to get around to check some of the sample effects here I'm gonna take this jungle Vine here this is the prompt is jungle Vine and bird so you have lots of foliage coming in and birds and some kind of seeds and notes Here I think this is very cool to be honest and if you feel that oh I don't want all these leaves coming around or you have text effect bits you can change between tight it's medium and loose so let's look at how the tight one here is and as you can see we now have nothing around the letter so everything is contained inside the actual typeset so I think that is very cool I'm gonna try another one let's check this font Vortex I'm gonna take candy various candy here's another great example I'm going to choose a pretty past style that's pretty nice this Pink's nice as well I'm gonna go with this and when I feel you're finished with it you have options here you can either submit to the gallery you can copy it to your clipboard and use it to export to photoshop or whatever you can also Download Here If You are downloading they're applying a watermark as you can see down here in the corner Adobe Firefly it's a beta as of right now so the image is not for commercial use but this is a pretty big image this is four thousand by a thousand pixels roughly so you can zoom in here and here we are at 100 percent and it's fairly detailed I mean if you go close you can see the stuff here is kind of mushed together but for a beta I think this is very good and I mean looking at this size it looks sharp and crisp moving on to the text image which is what most of you are probably already familiar with from mid-journey stable fusion and Dolly and that is text generation a text input prompt that will output an image and here are some examples created by the community as this has been trained on Adobe stock photos you will have that sort of stock image feels sometimes but you can't still get amazing photos out of it so let's try and take a prompt here I'm gonna put in a rather long prompted first something you could see in in mid Journey we have a 3D fluffy alpaca adorable cute big circular effective eyes long fossil fur 3D render cinematic smooth intricate detailed cinematic vibrant tropical jungle in the background that was a mouthful then you have some options here to the right just a quick input here that these images are great to be honest this is just a first render that I have not cherry-picked now I did prepare the prompt but I didn't prepare that I would can't even choose a seed so this is completely random and you have here first of all you have aspect ratio so you can pick that fairly easy if we change this to 16x9 we get a new generation so you can't keep the same if you have something you like in a square format you can't have the same image you need to generate a new one and here then you have your content type and then I've noticed that picking photo art or graphic is actually the best option if I choose num and try to describe what I want art style whatever it doesn't become as good as if I press the art content type up here you can still get good images but it's much easier to do it with the presets up here then you have some preset styles that you can play with and these two are actually fairly okay if you press the little buttons here all you can see all the available Styles now you can still input whatever you want down here as a style but this is much easier however just a reminder that if you have too long prompts then you can override the settings here so what we have down here is a bit much so it might not work with the Styles here here we have an example where I've input puppy dog playing in grass and I've applied some Styles here first of all we have the content type photo and then we have a color tone pastel color we have a lighting which is dramatic lighting and composition close up and we can change this let's say we change this into shot from below and let's change the style into let's try this synth wave see where that gets us so now most of the images are shot from below we have a synthwave character style now this wasn't great so we're gonna remove the synth wave I'm gonna remove the pastel colors here I'm going to change the color tone to vibrant colors and we're going to generate again and now these are actually very good photos now looking at this we have a very narrow depth of field so we have got some great bokeh coming in here I think the details on this dog are actually very good and the grass looks great so as far as photographs go especially ones that are very similar to stock photos you can create great stuff and if you're new to generative AI it's as simple as just telling the AI what you want in the input field down here I'm gonna try something else we're gonna have a flower in a colorful swirling space nebula gonna remove the Styles here we actually press the clear style and we're gonna change this into art and I want a 16x9 and we're gonna let it generate for us now this is very cool and I like how the flower kind of merges into the swirling space nebula here so I'm very happy with this result and I like you know the Spacey nebula images in general because they're so chaotic and you can get all these little happy accidents however if you change this into a photograph you're getting a much more crisp result well it still looks like an artistic painting because you can't really see this in real life there's some extra Christmas to the edges and everything I think this is one of the best of the bunch but it's all very subjective this foreign is fairly good and with coherency in understanding what you want so if we do something more complex let's have an astronaut riding a motorcycle on a space rainbow we're gonna generate that we're gonna keep the photo style for now so we have an astronaut on a motorcycle that didn't fail in any of the image it's not really on the space rainbow it this one here could be riding on a space rainbow the other two only have has the space rainbow in the background but it's still fairly good compared to what we asked it to do now the big difference is compared to Mid Journey V5 which can get images like this with a simple prompt and this is basically just portrait of a woman in dramatic lighting F 1.4 35 millimeter lens and if we do the same thing here we will get more stock image like results can't load one or more words violate usury guidelines I think I remember from another video that was Lance so we're gonna try again and as you can see while these are exactly what we asked for it's portrays of women in sword or dramatic lighting but it's not at all similar to this kind of output now this is something else when it comes to the artistic Vibe of it that that's not to be said that this can't be used these actually look great and it looks like real people the eyes are great which which is one of the harder things for the AI to reproduce together with hands and that's something we can look at this time holding up her hand I'm gonna add to this and in I've seen three hour images we have something that is fairly good here's a hand with five fingers here's a hand with five fingers the nails are kind of weird and there's some artifacts there and this one's kind of broken but honestly with without cherry picking anything it isn't bad it's close to what you can get in mid-journey now something I noticed with this image is how crisp and clear the skin I mean obviously she's wearing makeup here you can you can actually see that it's not that AI generated lines everywhere X actually looks like again and a photograph of person so while this was supposed to be an image to look at the hands we looked at the skin here instead that's okay here we have generated an image which where the prompt is cyberpunk Man video game art sci-fi City and background cinematic neon lights and I have chosen to put in art and concept art so these are more illustrations or drawings I think that's fairly cool let's try and remove the concept art Let's see we can we find a style maybe 3D art let's try hyper realistic and let's generate that I just love how intuitive this user interface is I mean it's it's very simple you don't have a lot to choose from so you don't have the controller you have in stable Fusion but for a lot of people this is all you need and mid-journey why that is great even for beginners it's still a threshold to get over installing Discord and everything so I think this is a great entryway into generative AI so let's have a quick look at what the other people have created so here's an example of a beautiful flowers and a blob object Frozen or block allies isolated on a warm gray background and I think if you're just learning starting out look at the images available and check the prompts that they have used that's basically how every one of us learned in the beginning of stable diffusion and only we had libraries of prompts where we borrowed from each other and adapted it and played with it I think this is a very cool image except for the moon or the sun here here's a pineapple that's made of rainbow cake inside I noticed however that these four Images are not the same as this one this was actually better this was kinda cute I like this so I hope you've learned something today and I hope you enjoyed all the time together just remember that this is still a beta so you can sign up for Access and they are rolling out access to a lot of people now I've been told by Adobe but it might take a while since a lot of people have requested access if you want to talk more about this let me know your comments in the comments I'm also available on Discord if you want to chat thank you for watching I'll see you in the next video as always have a good one see ya