Generative Fill in Photoshop! All Your Questions Answered

foreign [Music] ERS here it's an exciting week January Phil is here and is both exciting and slightly terrifying at the same time this new feature allows you to add objects remove objects create backgrounds and extend images let me know in the comments what you think about this new feature so far my goal for the stream is to answer all your questions about the generator fill so leave any questions you have down below and I'll try my best to answer them all also let me know where you're tuning in from I'm streaming from the beautiful San Francisco Bay area in Northern California also a big thank you to our sponsors MSI computers and the Creator z17hx Studio laptop which comes with a powerful 13 gen Intel Core I9 processor and the Nvidia GeForce RTX 4070 GPU with 8 gigabytes of video memory the display is a 16 by 10 golden ratio and it comes with an MSI pen allowing you to retouch directly on screen make sure you check out the link below for all the details I'll start the stream with a quick example in case you're not familiar with the generator fill and remember leave your questions below I want to make this stream as interactive as possible okay here we are in Photoshop and you know what guys I just realized I forgot to change my screen resolution so let me go ahead and do that now that way you can easily see the screen I was just looking at my screen and realized I hadn't bump it up usually I bump it to about 250 so that things are easier for you to see here we go so this is the room and by the way I am keeping a look at the chat I see a lot of questions coming in I see people letting us know uh where they're tuning in from I see somebody in San Jose which is basically down the street for me good to see you we have um Taha from Tunisia but anyway I'll quickly get started let give people some time to come in here and start asking their questions I do see one question that I do want to address but first let's take a look at this so in case you don't know what is the Gen fill AI or general Phil Li AI basically you can select the lasso tool and you can make a selection with the lasso tool so I can make a quick selection here under this window it has it can be a very loose selection and I'll click on generative fill and from here I can now enter a description now I see a lot of confusion online instructional prompts do not work for example make bigger make red things like that don't work so let me show you what I mean by that if I just make a selection over the entire room by pressing Ctrl a and click on General refill and type in make room red and click on generate that will not work again Adobe says in their official release of the notes and instructions that instructional prompts do not work adjectives announce is what works and I'll show you what I mean by that in a moment so when you make a selection like this notice that I get a completely new Red Room this is not really the room that I was supposed to that I was working with but that just goes to show you that instructional prompts do not necessarily work you might have a situation where you have a person standing you select them and you type remove person it will quote unquote work but that's only because Photoshop is just analyzing the surrounding pixels not necessarily because instructional prompts work now adjectives and nouns so basically words that explain things size shape age color origin purpose or material or a noun a word that names a person place thing or idea so that's basically what Adobe says to use when generating prompts so in this case as you sign the thumbnail I can make a selection around this area maybe I'll make it a little bit bigger and I can use the prompt retro yellow if I could spell yellow correctly that's one unfortunate thing about the generator of AI you have to learn how to spell which I'm really bad at retro yellow couch and I'll click on generate and as you'll see Photoshop is going to send this prompt to the cloud so you do have to be connected to the internet to generate this couch and it'll take a few moments here give or take 15 seconds to generate whatever you tell it to generate and here we go here here's our couch we can click on these arrows to see the variations I think I like the third variation best and that's what I'll stay with so that's basically how the generative AI works now I do want to address some of the questions that I saw here that are relevant to this very first example is that available for everyone or in the beta version well it's available to everyone who's a member of the Creative Cloud let me show you how to get it let me just open up my Creative Cloud account here clear Cloud app if you go into your Creative Cloud app you can go to the all apps section here and then I'm sorry not the olap section you on the left hand side here you can go down to where it says beta apps right here beta apps and you can see that photoshop beta and you can install that I've heard that some people were seeing the generative fill but it was grayed out one of the reasons could be that you haven't entered your birthday into the Creative Cloud account they need to make sure that you're 18 and older or at the very least confirm it and I believe you can do that through the B hands tab so that's one of the issues that I've heard people have and this be used to select objects of a person better than conventional tools no not at the moment it can basically do four things it could generate an object it can remove an object it can extend an image and I always forget the fourth one oh create a background and create backgrounds and we can talk about all those in a moment and I saw another question here any way to have the generator Phil replaces placement maps in other words have your brand be displayed on models with the new AI technology maybe map it to a file and have the AI produce a mock-up so that's a very good question Seth not at the moment at the moment it can do those four things I mentioned you might be able to get away with it in some cases but not really you can't really take one of your own images maybe like a client logo or something and then map it to something else at least not yet but I wouldn't be surprised if that would be something that the AI can do in the near future also another thing that the AI is really good at is creating Reflections so I'm going to use the lasso tool here and I'm just going to click around this area of the wall here and I'm just going to use the term mirror with wooden edges like so and then I'll click on generate and then Photoshop is going to analyze the image and then create Reflections that are relevant to the room which I think is incredible and while that's generating let me see if there's any other questions coming up oh I see a really good question there and I'll go ahead and answer that in a moment and it's a very important question actually so let me just finish generating this and I'll address the question that I saw in the chat and for some reason this is taking longer than usual I'm not exactly sure why but it is what it is I guess I can answer that question now does the new AI layer match the resolution of the image no at the moment the AI can only output a 1024 image on the longest side so it's 10 24 pixels on whatever the longest side is of your image so if you want something if you're working with like a 4K image 8K image or something like that it's going to just stretch out those pixels and it's going to look very pixelated there's no way around that now it's in the Beta release my assumption is that adobe is going to come up with a more more high-res version of the AI once it's out of the beta process but at the uh out of the Veda stage but at the moment it's only 10 24 pixels wide on the widest Edge and if you have something larger if you make a selection that is larger than that photoshop is just going to stretch those pixels so you will see some pixelation when you zoom in and we can take a look at that and actually this is stuck I'm not really sure why so I'm just going to cancel oh look at that I press canceled and it showed up but as you can see we have the mirror here with the wooden Edge and look at that is reflecting the room I can click to look at the variations and I think this is just amazing it's getting the reflections that are found in the room so I'm just going to stick with that one I think that one looks pretty good also um another example I showed um just because it's in the thumbnail I'll show it to you guys I'm going to make a selection here on the bottom left and I'm just going to create uh I'm gonna type The Prompt wooden table and I'll click on generate this is going to generate a wooden table but then I'm going to show you how to create the reflections on the floor which I think are amazing um let's see and more questions yeah somebody wrote in the comments cryptocan you have to go through b hands to confirm your age so I saw that that was one issue that you haven't confirmed your age through b hands you have to do so so that adobe is make sure that you're over 18 and you can use the generative fill can you do a composite from user provided pictures unfortunately not yet but kind of so I'm going to show you an example so the answer to your question Chuck Norris is no but there's something you can do that's kind of similar hopefully in the future you can give photoshop your own images and it'll composite them for you but at the moment you really can't do that but there is something else you can do and it looks like my Photoshop is having an issue it's also stuck again I'm going to press cancel and see if it works this time yeah it worked um so I press cancel and it's showing up so I'm guessing it's having some issues but you can see the the wooden table here I'll stick with the second one I think that one looks best but the example that I wanted to show you is that if you select the polygonal asphal tool and just make a selection out of the floor here the wooden floor here like so we can then subtract these objects from the selection let me show you how to do that an easy way of doing that you can enable the object selection tool make sure that sample all layers is checked and then Photoshop will find those objects in the image and you can hold alt and click and subtract from the selection so now only the floor is selected you can go into generative fill and type reflective floor I'm not exactly sure what that's going to look like but I'm hoping to get some Reflections just so that you can see how good the generator fill is with reflections and I think the I think my face was covering the prompt so sorry about that I didn't realize that I was working on the left hand side I just took a look at the screen there but you'll see that the reflective floors came in and look at that isn't that amazing how it just figures out what the reflections are and you can just put them into your scene really really fast it's incredible I mean this is probably the most amazing feature I've ever seen in Photoshop I've been using Photoshop for 20 plus years and I've never seen anything this amazing so I want to hear your comments this is why I'm doing this stream I want to get your thoughts your comments your opinions and hopefully answer some of your questions because I see a lot of things online I see a lot of misinformation so hopefully I can help clarify some other questions that you may have um yes Adobe says it is not allowed to use this feature for commercial use um what about Instagram and other photo platforms like 500px good question so obviously I don't work for Adobe and I'm not part of their legal team but from my understanding is as long as you're not selling the images in any type of way then you should be okay to use them um the goal is for this technology to be used in your commercial work but that will be once this is out of the beta phase so I'm not exactly sure how long that's going to take but at the moment Adobe is saying that this cannot be used commercially my assumption is that that's probably why the resolutions kept to 10 24 on the longest side just because anything larger than that would probably be suitable for commercial work um I live in Mexico and the generative option does not appear the bar appears but not the option um make sure that you are on the beta make sure that you refresh your Creative Cloud account and also make sure you add your birthday to be hands I think those are the three solutions I've seen online for people who are trouble who are having trouble accessing that generative fill um Can generate philby use offline no mrf I cannot be used offline the general fill is taking your image and taking it to the cloud and that's where it does all that compute all that magic on the cloud it would just be basically impossible to do it on your machine it's just so many there's too much computing power that needs to be done so it sends it to the cloud you need to be connected to the internet in order to send it right back how much is the resolution of the wooden table is it that can is that uh um so the resolution for everything that is generated is 1024 on the longest side so 10 24 pixels so here let me give you guys an example and I think this one will really help you understand how this works give me one second here will I find the proper image okay great so we have this image here and I'm going to it expanded and look at their solution the resolution is huge right this is what are we talking about here 8837 pixels wide right and the Gen AI can only do a fixed size of 1024 on the longest size so basically that tiny little square there that's basically that the score that you see here and I'll fill it so you can see that black square or I'll make it blue that blue square is basically all the Gen AI is generating and then it stretches it out so let me show you what that actually looks like right so on a high-res image like this one let me change the style to normal and I'll make a selection here and add to the selection here on the side and then click on generative fill and then just leave it blank again you don't need to type you know fill background or anything like that just leave it blank Photoshop will analyze the sides there and fill it with content but what I'm trying to show you here is that 1024 pixel that's uh pixel square that is stretched out and it's not necessarily a square it's just 1024 on the longest side um so it's going to look amazing right it looks incredible but watch when I zoom in see that see the detail there's there's really no detailer look at the difference see this is the original image here on the right side and this is what the AI generated that's because it took that tiny 1024 square and stretched it out across the entire image now if you're wondering how can I fix that I want to use this technology but I need to have some detail what can I do well two options make your image smaller if you can't do that then what I recommend is one of two things do one side of at a time because now you're stretching a image on a smaller area so you're not stretching it so much you can click on generator fill and click on generate and it's still going to be blurry it's still going to have those issues but not as much and if if that's still not good enough for you then you're gonna have to divide the project into pieces and I'll show you just so that you can see the different options that you have so this is going to have a little less pixelation than the previous example still there but not as much so you can see the difference see that this one gave us a little more detail and if you really wanted to just use the maximum resolution possible that the AI can give you you can use the rectangular Marquee tool select fix size 1024 and basically you can come in here and then just literally tile this all the way to the Top If you wanted to but that's the only way you're gonna get um results without that much pixelation pixelation simply because at the moment you're only stuck with a 1024 on the longest side image and then stretched out to a different area so when I zoom in here notice now see that see how now we have the perfect amount of detail it just so that's what you have to do because what Photoshop is doing is taking something that small and I'll rasterize it just to show you guys so what Photoshop is doing is taking something that small and then it makes it larger right and it covers the entire area and then you get the the pixelation that you saw earlier see how that's very similar to to that that's what Photoshop is doing behind the scenes so 1024 is the largest it can output and then it stretches it out accordingly let me see I see a lot of questions I'll try to get to them all let's see how much uh let's see yeah the resolution sent 24 10 24 on the longest side can I remove the window using the feature yeah so the window here I can definitely remove the window so if I was unhappy with the window I can just make a selection with this tool like so and click on generator fill again when you're trying to remove something you don't necessarily need to type in a prom just hit generate in Photoshop will analyze the image and give you a result so there you go it it will remove it let me see if there's any other questions yes very good question uh Gad Melvin can the general field prompts match with existing Shadows light direction of the image absolutely and actually didn't do that good of a job removing the window here just gave me another window sometimes the AI doesn't work as good but I can try to hit generate again and see if it works and if it doesn't work what I like to do is give it um like a descriptive prompt so I'm going to type in like blue wall and see if that works or I'll just have the word wall and we'll see what what happens here looks like Photoshop is having some issues so let me just type the word wall and see if that helps um what was the question that I was going to show oh yeah matching lighting direction and all that good stuff yeah I'll show you that in just a second okay still not giving me um a result but in theory it should work and actually to be quite honest with you the original image had a bookshelf here on the left hand side and I removed it just by selecting the selection and and leaving it blank but for whatever reason is not working but the left hand side here had a bookshelf to begin with anyway um let me answer the question about the lighting so in this this is a very good example and what I'll do is I'm just going to make the image a lot smaller just so that you know it we don't lose so much resolution when we're scaling up let me just work with this so what I'll do here is I'm just gonna just make this much much bigger and I'm just gonna do an outcrop so that you can see that the lighting matches the Shadows match all of that stuff matches so so you can actually do select invert now from the taskbar here which is super cool generator fill click on generate and give it a second you have a good question uh man deep can any further changes or modifications we made to the generated result of course you can make different kinds of modifications I'm not sure what you mean by your questions specifically but the short answer is you can make whatever modifications you want I'll get to that in a second notice how the Shadows match the area and the lighting and perspective and everything is just amazing the only downside at the moment is the resolution also if you select a side you can actually give it a prompt if you wanted to generate something more specific so in this case I want to generate a lake with distant mountains and the lake and the mountains should theoretically match the lighting shadows and everything that we have going on here let's see if that works and there you go you can see how there's a lake back here and mountains look at that isn't that amazing it's pretty pretty impressive I mean it's incredible that we can do this so fast um so to answer the question about modifying these further so I do have an image here of me actually let me see if I can find it uh this one here and hopefully this will answer your question so there I am and what I'm going to do is I'm just going to make a selection here across my face and I'm just going to type the word glasses glasses there we go now whenever you make a selection Photoshop will generate new pixels over that entire selection what that means is that there is no transparency Photoshop will not use my eyes it's going to replace my eyes and to answer your question I can modify the glasses see that see how those are not my eyes those are not my eyes so if I wanted to use these glasses over my face what I could do is use a layer mask remember everything comes through with a layer mask so I can just paint with a mask with black so I'm modifying the generated glasses right so I'm now modifying them so see I can paint with black and I'm going fast obviously but when you paint with black you hide the contents of the Mask so now my eyes came through and what I could do is below that layer create a new layer and paint with brown or black or whatever color you want I'm going quickly here obviously but then you can now fake the illusion of glasses over my face and I think you get the idea I don't want to spend too much time getting all the details right because I just want you to see the the process right so now that would be basically the process right I took the generated uh glasses removed the eyes that don't work and I can just you know use traditional Photoshop tools to make the generative fill work because it even though it's impressive it's not a one-click solution in most cases and I hear a lot of people saying Photoshop is going to be really easy and people are going to um anybody because is going to be able to do Composites and stuff like that and yeah sure a lot of things are going to be much easier but you still need to know the traditional tools to fix that 10 20 that it doesn't get let me see I'm gonna just scroll up here guys give me one moment I want to make sure I get to as many questions as possible there's a lot of questions let's see can I edit using my home images of course you can edit using whenever whatever home uh whatever image you like it doesn't have to be stock photos and it doesn't have to be generated by AI it can be whatever you want okay I think I've answered a lot of these questions so I don't want to answer the same questions again does it remember what you change yourself so next time am I use that as well I'm not exactly I understand your question but it doesn't remember anything it doesn't whatever I generate it'll never generate again it's always going to generate something different the only thing that it quote unquote will remember is the selection that you make and for example on this particular image I on another day a couple days ago I added blonde hair to myself see that and I added glasses right so it remembers that selection so I can come in here and it remembers that I generated blonde hair and I can click on generate again so that it can give me more results so it remembers the selection and it remembers the prompt but I'm not exactly sure what you mean about what it can remember so remember the selection it remembers the prompt but it doesn't necessarily remember what you generated in terms of like basing it off of that you know it won't base a new generation of something else that it generated unless you create a new selection in like Trace over it it will reference it in some way but what do you think about that should I should I go blonde um let me see any other questions um art said can we make my wife thinner I mean I this actually doing this it makes me thinner so there you go so I have a selection here with a long sleeve I have my body selected and the shirt and everything here and I can just type in leather jacket and nine times out of ten it makes me skinnier so let's see leather jacket click on generate and we'll see what it does the AI work on Smart objects um so the answer is yeah but I'm not exactly sure what you mean by that question and I hate to give like very gener generic and answers so like for example if I were to convert this layer here into a smart object I can definitely you know make a selection on this side and click on generator fill click on generate and it should generate that left side of the image and that original image is a smart object I'm not exactly sure if that's what you meant if the question is could I put that inside of a smart object then the answer is yeah I can select these two layers convert them into a smart object and now when I double click on that layer then I have both the general refill layer and the other smart object within that new smart object so yes it works with us with a smart object but I'm not exactly sure if this is exactly what you meant by that question let's see it sounds like you want to be blunt I don't know if that'll work um I it definitely looks like I'm in a K-pop band cool um another thing I wanted to show you guys that I don't see a lot of people talking about um and I think it's a very powerful feature is an image like this one and sorry I'm making you hungry it's making me hungry but if you have two images right maybe you know you're like a person who shoots photos for like restaurants doesn't have a restaurant but this is just the example and you have two photos but they want to use both of those photos in like their Facebook cover photo you can actually use AI to blend the two images together to merge the images together and the way you would do that um bias by selecting the rectangular Marquee tool and I'll zoom in a little bit closer and then just making a selection between both right so I have one image a selection and the other image generator fill leave it blank you don't have to type anything in and Photoshop will merge those two images and it might not be perfect but that's okay we can work with that so let's give it a second here and there you go look at that it did a really good job merging those images right look at that I mean it's almost seamless uh I mean I would have to zoom in and see what's not good I'm kind of not liking this area here very much but I mean I could work with that I mean that could definitely work and if you didn't like that area you can just generate over that area again and see and to be honest I didn't even click on the different variations so there might have been a variation that worked better I just went with the first one but it's incredible how you can just you know just merge two images like this very very quickly and very fast and for social media work and things like that this is this is just fine I don't have to worry about the resolution issues this is an issue I've encountered a few times it says the uh generated images were removed I believe it said because of uh some sort of violation for whatever reason the AI thought that this was not appropriate I'm not sure why that was but you can you can see it removes only one variation that's very interesting but we have two variations here and if I don't like that that's okay I can come back and see these oh actually I like some of these better this one has the chicken outside of the plate I don't know if I like that I probably like that the best but the point is that you can merge two images together in this way and that was one of the questions I saw earlier they were talking about merging images like your own images and compositing this is basically the only thing we can do at the moment just merge two images but it does a really good job as you can see so you might want to try that let me see um question good question have I tried any AI canvas out painting or in painting before generator fill in Photoshop the answer to it is not extensively I tried out a couple but not enough to give you any real comparisons do you think the creative industry is worried about these Neo AI tools well I don't know about the whole creative industry Mike but I can only speak for myself um I'm not necessarily worried I just see it as the next step in evolution so to speak if you think about Photoshop 3 that's when layers came out people were freaked out about that people used to say I'm so good at Photoshop I can do it all in one layer I don't need more layers that's going to take the art and skill from Photoshop because I'm so good I can do it all in one layer layers came out now everybody's good at Photoshop you know I kind of see that in in the same way as this I don't think it's going to replace everyone I think some jobs will be lost without a doubt but I think that if design is your passion Photoshop is your passion creativity is your passion you'll find a way to be creative with the new tools but it's definitely changing the world that's that is for sure does it only work for English at the moment yes Amir at the moment this is only only works in English and as I mentioned earlier primarily it works with adjectives and nouns descriptive prompts do not work let me give you another example of what I mean by by a descriptive prompt let me see what can I use here I have I thought I had a picture here that I could use for that but I guess I don't I have a an image that I use on my tutorial by the way if you haven't watched my tutorial I talk all about how this tool works but anyway so I think this is the image I use for the tutorial you can make a selection around this dog and let me show you what I mean by up like a prompt that wouldn't work if I say something like make dog bigger that is an instructional prompt and at the moment Photoshop doesn't support it make dog bigger right I'm giving it an instruction and there you go see the dog is not bigger and it's a different dog right so instructional prompts do not work so when you say remove car or remove this or remove that it doesn't usually work sometimes it might actually work because that's the only thing you can do it looks at the surrounding areas and then maybe fills in the person but again Adobe says that at the moment instructional prompts do not work um let me see what else we have okay one other question says um basically a saying that they're spotting areas that or arm spotting areas that I don't like do I think it's going to improve soon I do think it's going to improve dramatically remember this is beta this is not even fully released yet this technology in in terms of adobe and in terms of it being part of Photoshop is it it just came out like literally days ago so this is the worst it's ever going to be if you think about all the tools that have ever been added to photoshop the day they're released in beta or full version it doesn't matter the day they're released that's the words that they'll ever be they only get better from there so if you think about this technology what we're seeing today this is the worst it'll be so in a year two years who knows whatever the time frame is it'll be much much better you know as time passes okay let's see will this be free with unlimited Generations if you have a Photoshop subscription all where it'd be limited to 200 500 Generations do you have any idea of costing so Adobe has not had any official word on what that will be Steve but judging on how expensive compute is like every time we do this it costs Adobe money every time you generate something it costs Adobe money it might not be a lot but if you have millions of people doing thousands of things a day then it adds up to a lot of money so my assumption is that there'll be some sort of fee I don't know what that will be I don't know if it's going to be added to the subscription of the Creative Cloud I don't know if it's going to be per generation I have no idea but unlike say the remove tool or the Clone stamp tool that it's all generated in your computer it's all computed in your computer this is computed by servers in the cloud and that costs money so my assumption would be that there will be some sort of charge but I have no idea what it'll be how much if it'll be a lot a little who knows or maybe it's free I don't know there's been no official word on that um let's say cool I see no other questions let me show you guys another example now that I have everybody here um another cool thing that I haven't if you guys saw my tutorial and you probably have seen something similar but another thing I haven't seen a lot of people do is just composite from scratch right so I have I've created Composites completely from scratch a lot of people start with like an MP I'm sorry with a photo but I decided to start with a nimbi canvas and you can still generate something with an empty canvas right so in this case I can do distant grass field with patches of dirt and click on generate and by the way something I didn't mention is that with other services like mid Journey you have to type in the proms HD 4K and all these other things to make it you know feel like or have more resolution you don't you don't have to do that with Photoshop because two reasons number one is Photoshop is looking at your image and is determining all of that from your photo number two is the output can only be 1024 on the longest side so no matter how many times you type HD and 4K 8K or whatever it's not going to give you any more it's just gonna give you what it gives you so cool we have these um these distant grass fields and I think they work and now I can do the top part here and I can do something like um Mountains mountain range with I don't know impressive clouds or something I don't know we'll we can try that and see what comes up and that's the cool thing about it you can just keep trying until you find something that works it says here that I'm dodging someone's question I don't know pepper there's like a million people asking questions does the narration match then I've answered that question like three times actually so I'm not dodging your questions I don't think you listen properly I've I've mentioned that the output is always 10 24 on the longest side so yeah um no the answer is no it doesn't match your image resolution it matches the output which is 1024 on the longest side no matter how large your image is no matter how many different HD 4K 8K prompts you make all right so there we go impressive clouds mountain range that looks pretty cool and then we can do something here like I don't know like um what can we do here um let's do something like uh 1950s car and then click on generate and there it is is that a 19 I don't know anything about cars so if there's a car Enthusiast in the chat let me know um but yeah there we go does that look like a 50s car you guys tell me um if I I don't think about cars I'm gonna say yeah that looks like a 50s car so then another cool thing this thing can do is um it's really good with Reflections as you saw so one of the prompts that I like using is reflective waves we'll see what it does in this case but the point is is that I'm compositing this photo just from proms it's it's incredible that I can just type a couple things in it and it will just generate them hey MW Roberts how's it going good to see you thanks for the comments and there you go we have some some water not not exactly the best water but maybe I can try something else maybe I can try something like River with oh how about this strong waves I don't know what's gonna what that's gonna look like but hopefully it works I missed your question Mandeep what's your question it says we posted it twice and it went unnoticed but as you could imagine there's a lot of people commenting so I I miss some of the questions what is a good very good question primer rage where is it pulling the imagery from so it's using Adobe stock images and other sources where Adobe has license the images or images that are no longer under copyright and but primarily it's Adobe stock you can say Adobe stock that's probably like the majority of the images and it's just analyzing all those images that are on Adobe stock to generate the result cool nlx78 it fits the 50 60s car scene of Cuba oh that's a good point it does fit the 1960s car scene of Cuba yeah you're right about that so I don't know if this is an issue but oh there we go look at that I think this looks much better actually these waves with strong waves oh this is pretty cool which one do you guys like best I think I like between these two but anyway the point is is that you can generate your own images from scratch you don't necessarily have to start with an image um Primal Rage is asking could you point to your own files stored on your PC not at the moment maybe that's something you'll be able to do in the future but no at the moment is just analyzing Adobe stock images and something else I should mention is that the technology is so good that adobe can actually tell what images it's referencing to generate the result and in um I'm not exactly sure how it was worded by Adobe under release but the gist of it is that in the future there might be some sort of profit sharing plan with Adobe stock contributors so instead of like someone licensing your image and you getting you know whatever the the fee is for someone licensing your image you would get some sort of Kickback for when an image is generated using your reference photo on Adobe stock so I don't know what the numbers are or what the plan is but that is something that adobe is uh playing around with at the moment okay it's a very good question Chuck Norris can I can it add Reflections or surfaces without distorting them or is the same struggle as with the glasses that's absolutely correct same struggle so at the moment it cannot generate transparency so whatever you select whatever your selection is it it cannot generate transparency it'll generate new pixels so if you are working with something that you wanted to generate a reflection and let me see what would be a good image like um I don't know like in this table right if we wanted to reflect trees what I could do is just old traditional Photoshop method right so I'm just gonna make a selection here and I'm just going to use the word like I don't know mirror we'll see what what it does I'm not really sure if this is going to work but we can try it mirror and my goal is that that'll just create a reflection and I can Photoshop it using blending modes or other techniques right just because it wouldn't keep that same surface it'll generate new pixels and it might be okay it might look fine with the newly generated pixels but in some cases you want to keep the actual surface that you have so here we go we have a couple Reflections I'll stick with that one just because it gives me more room to work with and what I'll do is I can just you know use one of the blending modes like maybe screen or lighten I don't know whatever whatever works best in this case maybe like soft light and then what I'll do is I'll fill the layer mask with black um and then just paint with white on the areas where I want to reveal that reflection right and this is what I was talking about earlier the AI doesn't do everything for you at least not yet so you do not you have to know the traditional Photoshop Tools in order to make your designs work so it's not doing everything for you just yet it's just saving you a little bit of time and then you know for example if you don't know about blending modes and about opacity then this might not necessarily be something that you could do but if you know about blending modes and opacity then you can apply a reflection to that table and I think that was your question Chuck Norris if I'm not mistaken do I have to give a disclaimer that the image is AI generated I'm not exactly sure what you mean by that Saeed um I'm assuming you mean when you generate something you post it online I mean I guess I guess you have to um depending on what you're doing with the image I'm not really sure what you mean by that question sorry very good question MW Roberts as a teacher of Photoshop aren't you a little bummed that techniques that took time to master can be achieved by novice with no expertise in Photoshop not really um example I made earlier was about the whole layers thing like for example let me give you guys a perfect example of that there was um people who were very very good at you know creating text in one layer and then using um you know other techniques to create a shadow and now all I have to do is double click to set a layer and click on drop shadow and then just you know I have a drop shadow right it's like a perfect drop shadow there you go look at that perfect drop shadow just with a few clicks anybody can do that and that used to be very very difficult prior to photoshop three like it would take a lot of skills to do that so some people got really really good at making these kinds of Shadows and then next thing you know you can do it with just a few kicks a few clicks so in that context I'm not not really worried is I assume that that's going to happen can can we add a chandelier a line with a yellow light bulbs on the room and if yes then does the air adapt to the lighting conditions accordingly um so the short answer is yeah I mean here I'll add a chandelier here and God I don't even know how to spell a chandelier so let me see if I I hopefully you spelled it correctly um that's one of the downsides about this thing I've learned how to spell I don't know if you guys are good spellers chandelier with what did you say with yellow light so it's not going to change the lighting of the entire scene obviously because I'm only making a selection on that particular area but theoretically it should change the light within that selection and if you're wondering well why don't you select the entire thing well then that brings us to the problem that we saw earlier with the Red Room that will just change everything because anything I select Photoshop will regenerate and it won't be exactly like what is behind it so there we go a chandelier with yellow lights there you go um oh with the yellow I forgot the word lights let me let me change that again with the yellow lights I'll click on generate again thank you so much Chuck Norris foreign yeah so there you go it it it's not going to change too much in this case but you know there you go but this is exactly what I was talking about earlier with knowing the traditional Photoshop tools right because if you generate this and in your opinion there should be more light up against the wall well you can certainly do so right you can paint with you know a similar color like like you know this yellow here whatever color it is go to the side of the layer change the blending mode to like uh you know I don't know color Dodge uncheck transparency shapes layers reduce opacity and now you know you can create the light right so again my point being is that just because this can generate something it doesn't mean it does everything for you so you still have to know some photoshop techniques uh Richardson media doesn't match the original resolution of your photo from the images you're generated by the general refills no 1024 on the longest side um if you just join the video you can just go back to like minute 10 or so we talked about how it stretches it and I showed a few examples so unfortunately it does not match your high-res files it only does 1024 on the longer side very good question PJ Z and also thank you for all the comments you leave on my podcast I started a new podcast by the way today's Creator if you haven't listened to it already and I recognize your name from your comments pjz um so yeah the answer is I have and I think I can talk about public this publicly because it's already live but let me show you guys um a professional piece that I worked on let me just find it here and I'll show it to you guys and I will answer that question have you ever have you guys ever seen the show um Ghost Adventures I've never seen it to be honest but uh I created the key art for this so this is um this is for the Discovery Channel and the Travel Channel and I've made this and one of the issues I had was that the poster was basically a portrait and I needed to extend it and it took forever it took a really really long time it really it literally took me I don't know the better part of a day just to get the water the you know everything it seems simple but it actually took a long time and I would have used the Jane AI on this if I could because it would I would have been I would have had it done in like a minute but that's one of the ways in which I'm planning on adding this to my workflow for things like that obviously extending images that's another way that I would use this new technology I probably wouldn't use it for like the main object like I don't know if for example if what I would probably use it the most is to generate like background stuff so if if this were a portrait and I had like my main person in front of this scene and then the items here in the in the living room or whatever this room is in the background I would use to generate some objects maybe objects in the background to generate interest like the mirror that's probably how I see myself using it but it's only been around for about a week so I haven't haven't had that much experience with it yet well I've been playing to be frank with you I've been playing with it for almost a month now I'm part of the Photoshop pre-release team and we've had it for about a month but still not enough time to like fully come up with workflows just discover a few tips and tricks yeah you can definitely create your own assets absolutely that's a very good question um and I'm sorry if I mispronounced your name jeharu does the AI work in other color profiles other than RGV or those that work on profoto and 60 bit as well why don't we find that out let's uh let's go into us we'll go 32-bit and we'll do Pro Photo and see if that works I'll click on create I haven't tried it to be honest I'm going to try it for the first time here with you guys so I'm just going to make a selection here and I'm just going to type oh yeah so I guess not it works on RGB 8 or 16 bit so there you go grayscale lab and CMYK are not supported and what it what else did it say there and the mode has to be RGB there you go thank you for the question I rarely work on those color spaces so I wasn't sure let's see yeah that's um and I'm sorry an LX 78 I'm not exactly sure what you're referring to but I'll read your your comment and reply to that Jesus what I meant earlier is that with many techniques on media when you take time to examine with your eyes or software you can spot what is off absolutely you can definitely spot what is often on any everything I generated as soon as you zoom in you can start seeing some imperfections right I mean you can see how it's it's blending the pixels here a little weird and all of that but I go back to the same point I made earlier this gets you 80 90 of the way there and you have to use other tools to bring it all together and and I'm sorry nlx78 I don't recall your other comment sorry about that um would you consider just taking graduation photos in green screen and making AI backgrounds okay so Amish first of all if you're doing photography I personally wouldn't use green screen because the green of the screen puts green on your subjects and you have to do more work so if you are going to replace backgrounds like that I would use white or gray it's just easier to work with when it comes to photos if you're working with film then green is the way to go or blue now to answer the second part of your question or the Euro question I wouldn't use green screen so I was making a comment on that I not I mean it's always better to get things in camera right so if you can get it in camera great um if you can't then you can use a real photo but now you can use the AI generated background the only downside at the moment with Photoshop is that the AI generated background will be high resolution but if you're working but if you're only using the AI for backgrounds you can stretch them as much as you want and blur them to kind of create the shallow depth of field effect that might be okay to do but you know I don't necessarily I I would still try to get it in camera if you can yeah let's let's show more examples good good point I'm going to show one more example um let's see okay this is a cool one with this image I can come here on the on the side here and just make a selection and I can use the prompt building with reflective windows there you go for some reason it's taking a little longer than usual I'm not really sure why but look at that isn't that incredible how the reflective Windows just reflect the cars that are on the street the buildings are across the street I mean this is just I mean I just think I can't even speak because it's so incredible like I would have that would have taken me hours and hours and hours worth of work before so it's it's so crazy let me see and by the way guys so the whole point of this stream was for me to um answer your questions if you want to learn more about how this works make sure you check out my YouTube video on this I'll paste a link to it down below after I'm done recording this stream but if you go to my YouTube channel Photoshop training Channel and you check out the latest video here which is General Phil in Photoshop right that one right here is a complete guide so my goal for this video was just to show you everything I know about how this feature works so make you sure you try it out and actually I just thought about something I haven't talked about so it's really difficult to show to explain how this works just because I haven't found like a great example case for it but basically if you purchase a q key and then you fill with black that means that nothing is selected right and then you've paint with white that means you're making a selection of a hundred percent wherever I paint it right well if you press Q again fill with black and change the percentage to something like I don't know like 20 or something like that and then you paint the AI will use this as intensity not opacity but intensity so my goal I hope is that I can paint some wet areas here so we'll see if this works now press the q key General fill and type in wet ground click on generate so it should it shouldn't create such an intense effect we'll see if that works but that's another thing I learned while I was at Adobe San Francisco about a week ago learning about this with their team is that it um that opacity is more like intensity when you make a selection and there you go didn't do and I'm not surprised right because the intensity was so low that it it didn't really create a wet ground let me show you now if I do the same thing I'll press the q key to enter the quick mask mode pin with white or fill with black first and then paint with white white is going to just have full intensity right because that's a hundred percent black is zero percent so basically the same selection obviously it's not 100 the same and if anybody has figure out like a great example for this let me know but I haven't found an example that I like for this uh generally fill and now I can do wet floor but I'm showing you just in case you can find a good use case in your projects but the point is is that opacity on the selection affects the fill and let's see what this looks like yeah see the difference so the first one was I believe 25 and then this one was at a hundred percent so just for fun why don't we try a 50 one and see if that makes any difference so press the q key fill with black with the brush still active double click on the foreground color and change the brightness to we'll try we'll try 60 see how that works and then when you paint actually that might be too much let's try we'll go with 50 why not we'll go with 50 percent and then when you paint it will not select all the pixels it's like 50 intensity on that fill so we'll see what that looks like and I know my selection is not the best in the world but this is just for trial and error so press the q key again actually I don't like that at all let me let me paint a little bit better there we go and then I'll do wet floor once again what did I do wet ground or wet floor in the previous one I forget uh we'll just do a floor and just give it a second here let me see if there's any questions that's right uh solarun I've only been working with photographs but this also works with artwork so if you have like an anime style drawing or something like that this will also work with that thank you nlx78 have a good night is oh a very good question Robert Williams is there a way to adjust the size of a generated object that's an excellent question I didn't like that last prompt so I'll try I'll finish this example and then I will answer your question Robert Williams because that is a very good one generator field can work with 2D illustrations yes it can okay so you see the difference here so this is at um I think we did what do we do 50 so 50 intensity 100 intensity and this one's like 25 intensity so all this just to say that the opacity of the selection affects the generative fill so I that's something I haven't seen a lot of people talk about but I wanted to point it out again I haven't really found a great use case for this but you know maybe maybe something like this would be a good use case I guess that it's not as wet as with everything else anyway um so the question that I wanted to answer was from Robert is there a way to adjust the size of a generative object so the answer is yes but it's also no and let me show you what I mean by that um let me see I have a photo here that could work great for that actually no that's not what I wanted foreign that one's fine um I can use this image that's fine so in this image I generated this red truck and you can see there's other red trucks and you know what maybe this is a truck that I like but I want to make it larger I'm going to just duplicate it just so I don't lose that original truck so Ctrl J to duplicate that's command J on the Mac and I'm going to make this larger but when I make it larger Watch What Happens it um it makes everything else around it larger because as I mentioned earlier Photoshop generates pixels around the transparency as well so there is no transparency is what I'm trying to say it gener whatever you have selected it's going to generate pixels so when you increase the size it also increases the size of the background so it doesn't work right what you can do at this point is click on generate and Photoshop is going to generate that selection in the new area in the back room will be fixed but unfortunately you can't keep that original image so in this case I'm not going to keep that same red truck I'm going to get a similar red truck but not the same one so unfortunately you can scale them and this is the workaround and see there you go see the background is fixed the truck is bigger but it's not the same truck right so the other option is to make oops sorry about that the other option is to make your truck larger so there's my larger truck and then with the layer mask just paint with black and you know kind of do what you used to do in the past right you would just mask out the things that don't work and I'm going faster but I think you get the idea so you can regenerate but that'll give you a brand new truck or you can just scale up and then mask away the areas that don't work if that makes sense let me see is there a way to it okay Control Alt probably won't okay good question Michael uh Michael Gilbert is asking are you typing a shortcut key to bring up the generator fill dialog box um no so when you make a selection Michael the this little window appears by default if you don't see it go into window and make sure that you have the contextual taskbar enabled also if it's hiding somewhere it's not appearing directly over the selection that's probably because you have it pinned somewhere else so just um pin it or click on reset bar position see how it just brings it right back so maybe for whatever reason you might have it pin up there when you create it you don't see it and it's up here when you find it just unpin it and then reset the bar to this position and it will appear directly below the selection whenever you create one also I should mention that there's two other ways of getting to the generative fill you can right click on the selection and choose generative fill and it brings this window where you can enter your prompt you can also go into edit and choose generator fill here it is I don't like it I don't like doing it that way it's too much work going into edit right clicking in the selection so I never use it for me it's just easier to create a selection and click on generative fill here on the contextual taskbar okay can you assign a keyboard shortcut instead of using the bar menu that's a very good question pjc I haven't tried it so let's see if we can if we go into edit and go into keyboard shortcuts um let's see edit generator fill here it is yeah so I guess you can do a keyboard short but that'll be for the for the generative fill window so I guess you can do it that way again I don't like that window so I probably wouldn't use it but if you wanted to you can certainly do so as you see here something I haven't talked about publicly that I've been kind of playing around with and I'll show you guys just because you guys have spent like an hour listening to me talk here uh let me let me close all these windows and show you something that I've been working on um with this whole generative fill and I thought about this idea last night so it's not completely thought out yet so you're gonna get you're gonna get to see a work in progress right so let me just see what image will be good for this um I don't know maybe these dogs I don't I don't know if it's gonna work with this image or not but uh what I did it what I realized is you can actually create an action using the generative fill so what I did is I took let's see there we go we're gonna we're gonna crop that right there so that is basically the area that the generative fill uses for um to generate an image right so if I created this action I called it expand test and basically what it does is it just keeps going around in a circle generating those areas to to expand an image and then it costs an error so I have to go back and and test it to see where the error happens but you'll get the idea as you see it it'll do one side fill it move up do the other side it didn't do that good of a job there but um the point is that I just created this action that just goes around generating different pieces to expand an image so it works with actions as well um but it's not doing that good of a job in this case oh you know what here I discover one issue so if you do the if you are gonna do this one of the issues I discovered is that um it doesn't do that good of a job with lock layers so that was one of the issues that I came across so let me try that again so I'm just going to make this super small again and I'll put it here and I'll crop it and I need to unlock the layer for some reason lock layers it doesn't work as well I don't know why but I'm going to try my expand test again and we'll and hopefully this will give me a better result so Andrew um oh look it generated a weird dog on that side that's so funny um Andrew's asking this uses Adobe stock assets can generated images be used commercially without a stock account so at the moment Andrew everything is not allowed for commercial use however when this is added to the full version of Photoshop you should be able to use it commercially so at the moment no but in the future yes I mean you won't need an adobe stock account and technically if you are a member of the Creative Cloud you have on Adobe stock account but I I got your question um but yeah at the moment it's not meant for commercial use um yeah I've answered this question already um I don't think that art design is gonna die because AI is you know doing a lot of the the work I just think that we're going to have to learn how to be creative in other ways but there's always going to be room for Creative people and designers maybe not in the same way that you're used to it today or were used to it in the past but we've seen this happen many many times the development of the camera the development of the computer itself if you think about it designers prior to computers how to do everything by hand computers came along desktop publishing came along and they probably felt the same way that you know design was gonna die because now everybody could use a computer you know so it's I see it in the same way that those people or I'm sorry that I I see AI the same way as as those increases in technology anyway I'm going to cancel this because I don't want to stay here all day but what I was trying to show you guys is that I came up with this this action that just expands uh keeps expanding expanding expanding expanding expanding and my goal is to see how big I could get it but um I ran into a problem when I think you can see it right here I ran into a problem with the with the action so it's not not fully fully done yet but it's something I wanted to show you guys since you've been watching the stream for this long um oh good question Matthew is there an issue because it's not getting in any of the of the original image selected yes very good observation so you can actually see that there's like these little tiny edges there and that's something that I I do need to adjust in my action that it overlaps just a couple pixels with the original image very very good observation yeah um so clumsy Dragon are you able to blend two different Skies together with this you absolutely can you probably were not here earlier to see how I merged these two photos these are two different photos and I was able to merge them together by creating a selection in the middle and having the AI fill in the blanks you can definitely do that with Skies as well when action is working well can you share it please I'll probably make a YouTube video video on it and I'll definitely share it Mike um but that'll probably take me some time because I haven't figured out exactly how to get it to work as you can tell but I only started working on it last night I just literally was sitting down having dinner the idea plopped into my head and I came and worked on it for an hour and then I had to go to bed but I'll try to work on it a little more and hopefully hopefully that that will work and I can make a YouTube video on it um Richardson media we have already talked about this but I'll briefly answer that question in case other people have missed the answer um this will be based will this be based on a credit system or unlimited prompts unclear Adobe hasn't made any official announcements my assumption is that there might be a paid element just because of the nature of this technology it costs a lot of money to generate these things so my assumption is there'll be something I don't know what it will be I don't know if it'll be cheap I don't know if it'll be expensive I don't know if it'll be part of the Creative Cloud I don't know but if I had to guess probably let's see a good question earthy is it actually possible to add highlights Shadows or color Corrections on a realistic character or subject so the answer is is no because if you have a realistic character or subject and you add something to it you're going to cover it much like the example I showed earlier where I put glasses over my face wherever I made that selection it generated new pixels they were not based on the pixels in the background so in that context no you wouldn't be able to add you know highlights and Shadows on top of your character the closest thing you'll be able to do is if you didn't see this earlier you can go back and see how we added that reflection to a table that probably be the closest thing you'll be able to do at the moment okay so good question here Michael can you specify depth of field within the prompt for instance to make the background out of focus so at the moment Adobe is analyzing uh let me rephrase that at the moment Photoshop analyzes the image of your images out of focus on the background it should Place something that's out of focus so let's let's try that um with an image here I know I have a couple images here oh actually I think I have the good one I have a good one here I have a couple here and everybody in the background is out of focus so maybe let me just add the word dog and see if we can get it to be out of focus back here so I'm just gonna place a dog there General Phil and I'll type in the word dog and theoretically speaking it should be a dog that's out of focus because of where I'm placing it in the scene and Photoshop is analyzing the image but who knows it might not work yeah definitely didn't work in this case okay well that one worked yeah so contextually so this tool is meant to be contextual which is why you don't have to type in 4K HD or any of that so this is about as best as as answer as I can give you there's no specific prompt that I could use that will guarantee an autofocus dog or whatever object you're trying to add yeah that's a big issue portraits by Daniel that you cannot replace the background with the same image on multiple images and I hope that adobe does change that because I I do see that as a drawback in other words whatever you generate you generate ones and you'll never get the same thing ever again so it'll be great if we can have a seed and then generate a background from different angles same background with no differences only the angle uh let me see what uh what's another question here when you use the tool do you upload good question um no I do not upload my own image um oh actually sorry I misread your question sorry about that uh jeharu so yes you do have to be connected to the internet and yes whatever I do to this image it goes up to the cloud the compute you know cloud system does this thing and it sends it back to my computer and no they are not using your images for their AI training model that would be illegal and if you find out that adobe is doing that you can sue them but no they're the training model is based on images they have rights to including Adobe stock um I know that some of you may be thinking well doesn't Adobe stock allow you to upload AI content from other sources and the answer is yes but those are not used in the training model um no trading model is perfect by the way can you sneak in a photo that was generated by another source on Adobe stock probably but you know that's they're it's almost next to impossible to be 100 sure that they haven't used an AI generated image from another source but you know in the perfect world yeah only real photos not AI generated images let's see yeah good question Tom so Tom is asking can you add that you only want realistic photos for the generative text Fields sometimes it gives me really fake computer generated images so the answer is no unfortunately like for example if I wanted to add um I'm just gonna say like like a person here like I'll say like I don't know just say man man walking man walking it'll probably look really computer generated um this AI is not very good with hands with faces with some of those things so they are going to look pretty computer generated and there's no prompt you can give it that it can assure you that it's going to give you something that doesn't look computer generated unfortunately now see in this case everybody's out of focus so there you go see like there's we just have a man walking in the background and everyone's out of focus what if I add them here so if I had him in front of this guy and I'll just put man hugging I'll just type into man hugging see what it does yeah I agree with you purchased by Daniel I do I hope that the quality gets higher than 2024 at some point soon yeah um rexus Prime there's a lot of things that adobe is blocking at the moment in terms of content um for example you're saying that you're working with um with a nude image of a model and and definitely at the moment Adobe is not going to allow anything like that it's really difficult to discern you know what like for a computer to discern what art is or like you know child pornography or something like that so if the computer is just gonna you know just be safe and not allow any of that but I hope that adobe figures out a way on how to solve that issue because I know there's a lot of people who work with models not necessarily nude models just maybe bathing suits or something and and the AI is really picky about that at the moment which is why I think um I I don't know I'm just assuming but I think that adobe is allowing a lot of people into the Beta now so that they can work out some of these Kinks and hopefully when it's finally in the full version of Photoshop it it it doesn't have those issues yeah I agree with you Tom there's it I'm just going to hit cancel and see what happens yeah it couldn't find another man hugging so I just did a another hand actually made him buffer look look at that arm it looks like he's been working out um that's so funny yeah I'll just I'll just put him I'll just add another man walking in front of him man walking we'll see what happens oh and actually um I want to show you very good point there earthy um I I yeah there's it shows men walking behind them but not in front yeah so you don't always get what you want so at least now I think your job is safe because if your client wanted a person standing in front of this guy the AI definitely can't do it but I I saw a really important comment there that I want to address so this is an image that we already looked at but just so that you guys can compare and by the way this is the this is all generated by AI right so the original one is that one and just so that you can see the difference I'm going to use the good old content aware fill because I've seen some comments online of people saying this is just content aware Phil and no this is not just content aware fill that's content aware fail and this is the generative AI I mean seriously I it's a night and day difference right content door fill generator AI there's no comparison so yeah very good point uh earthy thank you for that comment that's a very good point rexus if it read the metadata yeah that's that's a good point yeah but I couldn't agree with more with you more and actually talking about metadata something that I don't know if it was just a happy coincidence or not but I was working with this image here and I don't know if it's going to work now but um there we go it's an it's a pyramid of a photo I shot in Mexico and I just thought I went my cell phone but I was able to remove the well first of all extend the image remove the people add this Lake advice really cool Stone Bridge and then I started adding different things but one of the things that I added before adding all of this right when I got to this point prior to the water right about here and I don't know if it's going to work I I made a selection and I use the word Cowboy just cowboy and I was wondering if it was reading the metadata of my photo because it rendered a Mexican looking Cowboy so how would it know I was in Mexico does it recognize the pyramid and it knows I know in Mexico I don't know I don't know what it's going to do in this case but on that first try it was very surprising so I did okay definitely not a cowboy yeah see like like this guy almost looks like a like a like a Mexican charro a little bit I don't know but the point is is that I was wondering about that also something else that I tried here is I made a selection and I just type in the word pyramid um and it generated another Aztec looking pyramid rather than oh she didn't recognize your pro that I did I misspelled pyramid how do you spell pyramid pyramid anyway um it generated a lot of Aztec looking pyramids it didn't generate Like An Egyptian Egyptian pyramid so I'm wondering if does it recognize the photos or reading The Meta tag I don't know but see they're they're not Egyptian looking pyramids so my assumption is either it's recognizing the photo looking at the Mana tag or both I don't know thank you for everybody telling me how to spell pyramid yeah but anyway so I think that at the moment the AI is very good as you can see and you've seen a couple examples in this stream and generating water generating Landscapes I think it's very good at all of that water Landscapes Reflections mirrors all that it's just incredible some of you may be used the really popular plugin which I think is called the flood plug-in which I thought was an amazing plug-in but I think that this feature just pretty much made that plug-in useless so yeah try Egyptian pyramid alright cool I'm going to try Egyptian pyramid so here we go prompt generate fill Egyptian pyramid generate actually that that little selection was probably too small yeah I agree the water is insane that doesn't look like an Egyptian pyramid or maybe it does I don't know you guys let me know that's not a pyramid at all and that one I guess that one is an Egyptian pyramid yeah so there you go so as you can see it's it's not perfect it's not great it's amazing in certain cases in other cases not so much but again this is just Beta And as we mentioned earlier and I think I saw someone in the comments say this is the worst it's ever going to get so it will not get any worse than this yeah very good question uh jiharu you cannot uh this is this is meant to generate new content and that includes removing something right because if you're removing say a watch from a from a from a wrist you're generating new content over it so yeah it's it's meant to just generate new content you cannot use it for noise reduction or like you know I can't make a selection over this entire image and say something like add rain right like we talked about her earlier it doesn't do those type of prompts we'll see what happens but yeah it's only meant to generate objects at the time being and yeah it doesn't yeah again so the way you have to think about it is see it didn't work I said I had rain and and it didn't add rain at all right um You can create overlays I guess let's try that um rain overlay rain over lay well I don't know if it's gonna work but to answer your question and whose question was I answer it um yeah earthy it doesn't replace colors Reflections or anything like that you have to think about that whenever you make a selection Photoshop is going to generate everything on top of that selection so it's not good for generating Shadows or generating Reflections or any of that and yeah again we didn't get a really good result here with the rain overlay um but you know maybe we can create the effect that we're looking through like a window or something I don't know but yeah doesn't doesn't really do well with brain yeah any other questions guys I think we've been on for quite some time now and I just wanted to make sure that a lot of the questions that you guys had were answered uh as I said earlier in the Stream there was a lot of misinformation going around and I just wanted to clear up some of those questions also um I gotta mention our sponsors MSI Creator z17 HDX studio if you're looking for a brand new laptop you make you make sure you check these guys out I've been using MSI computers for three or four years now and they've been amazing the laptop was Best in Show at CES 2023 um it's a powerful laptop especially if you're doing any type of graphic work not just Photoshop these laptops are optimized so they are more efficient with blender Photoshop Premiere Pro and other design applications Link in the description make sure you check them out it references only Adobe stop images to create the content it will not copy other styles like um mid-journey right because mid journey is just using basically every image on the internet to train its AI model so therefore it's quote unquote better but they're generating content from images they don't have licensing to Adobe is only using images that they have licenses to which is why primarily it's Adobe stock and that means that when this is fully released you'll be able to use it commercially and you won't have any issues with copyright so you cannot say in Photoshop copy this style or copy that style to generate content now one thing you can do is you can put in a computer there I mean computer you can put in a an image of whatever you want and then add things on top of it and some of the Styles might be similar and in that context and yeah you can copy a style but it's not necessarily copying a style it's just copying the image that you're using did you try blending most with the rain uh yeah so so I don't know if you saw there but I had the the blend I was playing around the blending ones but it's not going to look realistic in in this case so I could use this this particular overlay for something else but not definitely not for rain yeah you can create a picture from scratch if you if you didn't see if you just tuned in I did there earlier in the screen in the Stream so just rewind the video and I created a complete picture from scratch and I don't know if I have it open here but I had a complete photo created from scratch very good question or comment here Colin I completely agree for people who make a living from uh Photoshop AI will take away the basic grunt work when comping or retouching but the quality and adaptive creativity will always be a human quality and I completely agree with that sentiment will Photoshop update normally to integrate the Gen AI the answer is yes I just don't know when the regular version of Photoshop at some point will have gen AI awesome Trevor I'm glad my video helped yeah by the way if you haven't seen my latest video make sure you check it out uh I showed it earlier I don't know how to bring it back um but anyway yeah here you go if you go to if you go into uh YouTube at photoshoptrainingchannel.com or not.com just YouTube at Photoshop training Channel you will see the generative fill in Photoshop it's an ultimate guide I basically you guys know my videos no fluff straight to the point everything I knew at the time about gen AI so it's all there if you just want to figure out how all this works by the way I actually did make one mistake in that video and I misspoke in the video I said that the Gen AI generates an image is 1080 it's not 1080 send 24 I'm just so used to saying 1080 you know when I'm talking about like video sizes or whatever so I actually misspoke in the video so I said 1080 it's 10 24. but as far as I can tell that's the only mistake I made on that video oh man Rex is I do talk to Adobe but I highly doubt that they listen to me but I'll definitely mention it thank you so much Tom I appreciate that good question about that um I um oh God fruit and similar fluid and similar photography um everybody knows AI has big problems with correcting uh with having basically can Photoshop fix the imperfections that AI has with hands unfortunately not photoshop's hands are no better than other AI generators yeah that's right um the 1024 as a Trevor you're right that you have to just do pieces you know piece together 10 24 10 24 10 24 to get over the long side limit but you know that could take a long time depending on your image but yeah that's definitely a solution to that issue at least for now yeah you're right Lori uh mid-journey does hands extremely well now absolutely I agree with that yeah I can you can definitely add people but as I mentioned before it's it's not the best it's it does people don't look realistic people that should be in focus with a lot of detail right so you're mentioning this lady doing yoga so I can you know make a selection here and then say something like man in yoga pose and I misspelled man oh my God a three-letter word man in yoga pose so it it probably won't look realistic but we'll see so you can definitely add things but the problem with adding people is you know face doesn't look very realistic hands and look very realistic at least not yet I'm sure that this is going to improve soon but yeah look at his feet see that look at his hand look at his feet a face so faces hands and feet don't work very well with Photoshop yet see that no John you cannot use your own library to generate images unfortunately I hope that's something that photoshop adds in the future Gabe um there's no official word as to how many generative AI tries Adobe will give you so it's all speculation at this point but my assumption is that there'll be some sort of limit I don't know what that limit would be but again that's just me speculating I have no idea have I used stable diffusion I I haven't personally used it the only other one I've used with some level of you know intensity is mid-journey but I haven't really used all the other ones which uh which version of Photoshop has the AI I mentioned that in the beginning but if you missed it you can go into the Photoshop beta just go into the Beta apps click on beta here and update it um if you want a specific version of the app then this 24.6 okay everybody well thank you so much for joining I really appreciate everybody being here um I don't see any new questions I just see some of the same questions being repeated so if you have a question just feel free to watch the rest of the stream I've answered a lot of questions um but yeah um and actually you know what Trevor I'm going to answer that because you make a very good point if something doesn't work use traditional tools like the patch tool to fix the issues absolutely right but that's that's how I recommend using AI use use it to get you 80 90 of the way there and use traditional tools to get you to where you want to be anyway everybody thank you so much I really appreciate all your support feel free to give a thumbs up if you enjoy the stream and share this and my other videos with friends if you enjoy them thank you so much and have a wonderful weekend see you all soon bye

Channel: Photoshop Training Channel

Views: 100,543

Keywords:

Id: VX41gHpGUSg

Length: 93min 36sec (5616 seconds)

Published: Fri May 26 2023