flux is the new open- source AI art generator that some people are calling the mid Journey killer but is it better than mid journey in this video I'm going to compare flux and mid journey and work out which one is the best we're going to run both of these tools through a number of different challenges so let's dive in and see which is the AI art generator to rule them all we're going to start off with an overview of each of these platforms and then we're going to go into a prompt battle where I put the same prompts into each of these AI art generators for a number of different circumstances we'll then compare the results and see which is performing better after that we're going to talk about censorship we'll take a look at pricing training data and other important aspects when choosing an AI art generator so let's begin flux is a brand new open source model that has been birthed out of the former team of stability Ai and it's doing some pretty remarkable things on their website they have this rather fancy looking graph which displays how flux performs against some of the other Market leading AI art generators and of course because they've made these charart themselves it makes it look like flux is the best of all of the AI art generators but we are truly going to put that to the test in this video and see if their own standards match up to what I find out so as you can see here they say that the flux pro model is better than everyone at everything apart from idiogram at typography so you can see the traits that they're testing are prompt following which is how much adherence to the prompt you get when you insert a prompt and get out an image also the size and aspect variability which is the ability to change the aspect ratio of the images now the degree of accuracy when rendering typography so it means when it's asked to render out a word that it actually gets the word correct W they've also asked for output diversity and for visual quality now on this graph the green shapes are the flux models and you can see that it purports to be doing rather well now you might be interested to see how mid Journey shapes up in this and mid journey is this shape that has very poor typography so this is a rather confusing graph but essentially what it's telling us is that the flux model is very good now the important thing to understand about flux is it comes in three different FL there is the Schnell which means fast model which is the one that's most easy to run on your own machine because it's smaller than the others and it costs less then there's also a Dev and pro model which cost more but give you more creative capabilities in this video we're going to be looking at the schel model Schnell means fast and German don't you know so just to clarify those three different models for you Pro is the best of flux offering state-of-the-art performance image generation with top-of-the line prompt following visual quality image detail and output diversity Dev however is an openweight guidance distilled model for non-commercial applications directly distilled from flux it obtains similar quality and prompt adherence capabilities while being more efficient nothing like the Germans for efficiency whereas schel is our fastest model it's tailored for local development and personal use flux Schnell is openly available under an Apache 2.0 license and the other exciting thing about the team over at Black Forest Labs who are the team behind flux is that they are teasing us with a possibility of an open source text to video model as well so we're going to be putting flux Schnell against what I think is the leading AI art generator on the market mid journey and mid Journey just very recently released an update to its algorithm we now have the v6.1 mid Journey algorithm for all to use now mid Journey 6.1 is coming out with some remarkable images just like these absolute stunners and the important thing to note about mid Journey 6.1 is that it's an incremental Improvement on the V6 model however it's building on some pretty incredible foundations now the team at Mid Journey has said that the 6.1 model is giving us more coherent images especially with arms and legs hands and bodies you know how much AI struggles with the digits of the human hand it is also creating much better better image quality with reduced pixel artifacts more precise detailed and correct small image features which means that when you have a complex scene with perhaps smaller characters that it renders the faces and other elements correctly now it's also saying there's an improvement to the UPS scalers that it's 25% faster and it has improved text accuracy now remember with mid Journey you have to put text inside of quotations it also has a new personalization model with improved Nuance surprise and accuracy and generally they say everything should look more beautiful across the board now mid Journey has also been teasing their future and in their office hours recently they alluded to the fact they are also working on a 3D model version where you'll be able to generate 3D models inside of mid journey and not only that they are also hard at work creating a video model which I am so excited for we've actually had rumors of this for many months now so I'm curious when it will finally be released and they're also discussing the possibility of creating a storytelling tool which would be a very exciting development and something I would love to see how we can tie together both creating stories and images into one place but let's dive in to a prompt battle and here I'm going to take a look at some images from both mid journey and from flux and compare and contrast the differences we're going to discuss the strengths and weaknesses for using these tools in different situations now of course there is a art of prompting in each of these and they respond differently to your spells that you're casting Within These AI generators and it's important to take that into to account when we're looking at these images however I have tried to be fair in some instances I have used prompts that are from the flux showcase so I've used these exact prompts that flux are saying work very well in flux and I've put them into mid journey and conversely I've put a few of my favorite mid-journey prompts into flux flux so for flux I've been using it on replicate tocom which is one of the easiest ways to run the model and I'm using the Schnell version Schnell so the first is a very simple prompt depth of field establishing shot character woman cinematic realistic I will leave a link in the description to all of my prompts and images so you can take a closer look at these comparisons yourself now first up we have mid Journey a little note here so obviously mid Journey outputs four options I have picked my personal favorite and here we have a very seductive young lady staring into the camera inciting you onto a an adventure it's almost like we're on the rooftops of Tokio and she's saying come with me let's fly across these rooftops but what I must point out is the lovely individual strains of hair the beautiful blur bcka background and the Very eye-catching glint in her pupils here also the expression feels very realistic there is a real sense of emotion passing through her however I would say there is something a a little little bit too perfect about the nature of this image but do take into account how wonderful these restricted tones are now we can compare this to Schnell and it's given us such a different interpretation to this prompt it's wildly alternative and it is it has a feel to it there's it's quite a should we say a more morbid dark and almost like a scandy Noir feel to this it's almost as if she is a a detective and she's come to this dark Bleak suburb in Stockholm to find out what has happened the color patter is restrained I appreciate this however the tones are a little bit shall we say depressing there is not quite this Allure or the aesthetic to this image and if we put these side by side and take a closer look for me the aesthetic nature of mid Journey once again wins out and this is for me why it always does and that is because the pure taste the Artistry innate inside of mid Journey always shines through whereas when you're using one of these open source tools you have to be much more specific with your prompting it's an interesting composition I would have to say coming out of chenel as well it's it's odd to have this this man's shoulder here and the image is extremely heavily weighted to the right there is so much black going on here on the right that the composition feels a little bit disorientating whereas here we've got a wonderful contrast between the light area here in the frame and the more midtones coming through on the right hand side so next up we're going to take it to the hands this is always a test of how well an AI art generator can render out the complex anatomical aspects of a human hand now because we look at hands so often and they're at the end of our arms we forget how utterly complex and ridiculous these objects are if you did not have hands and suddenly saw you saw a human hand you would think it was some very strange alien feature they are purely Fantastical mystical magical objects so I used a very simple prompt here once again hand held up to camera blurred citycape in background Sony fx9 50 mm f1.4 cinematic now I put this into mid Journey version 6.1 and the four images it came out with showed me four hands however one of these hands had a rather large amount of digits upon it but the other three performed very well and I selected my favorite out of this and as you can see we've got a a good rendering of a hand the hand proportionally looks correct and the aesthetic of the image is also very engaging now we had a look at the flux image and in the version that I put out first it only came out with four fingers however I thought I would give it another chance because we gave mid Journey four chances here and here it is this is a live prompt entry will our dear flux perform well or will it not the prompt is in the machine the machine is thinking he knows he's being challenged and here we have the result and as you can see it does look like we have four fingers however one looks little a little bit short and it's not a very interesting composition it's a rather odd and unusual way to be framing this image and the four finger certainly does look slightly deformed but it is a huge Improvement in the first output also the diversity between these two outputs are very interesting that they are quite different and that is from the same prompt now next up we're going to challenge The Prompt adherence of these two tools and this is where we're going to put in a complex relational prompt to both the tools and see how they perform at accurately depicting this scene so the scene is three magical Wizards stand on a yellow table on the left a wizard in Black robes holds a sign that says AI in the middle a witch in red robes holds a sign that says is and on the right a wizard in blue robes holds a sign that says cool behind them a purple dragon wow sounds like a wonderful place to be so here we have the two outputs on the left we have the mid Journey version and as you can see that it has unfortunately not quite adhered to The Prompt now on the right we have the flux Edition and it has done exceptionally well now this was a prompt that was showcased in one of the flux Galleries and I thought well I wonder how many times it took them to get this out E I bet this was 100 tries and then they got it so I thought I would give it a go myself and first time flux got it absolutely right so then I thought okay well let's see if M Journey can do better if it can actually perform this prompt if I try a lot of times and I put it in three times so we got 12 images and not one of the prompts accurately depicts the scene however I know a thing or two about mid Journey prompting and one important factor about mid journey is the words you put at the front of a prompt are much more important than the words you put at the end and so as you can see in these prompts the first part of the prompt usually renders more accurately so you can see in some of these we are getting at least it's saying AI is and then it's mucking up the dragon and cool bit so here I would say that flux is outperforming mid Journey it's doing a fantastic job at creating these complex relational prompts and so if accuracy to prompting is something that you are looking for then flux is incredible so let's move on to a character study one thing I love to do is create some really detailed imaginative characters inside of my AI art creations and these are what I'm using to turn into AI films now if you ask interested in the AI film making process I am just launching the AI filmmaker Academy so the prompt for this battle was photo cinematic old Jolly happy King and Crown INF Furs large sword in sheath and we have our two outputs here now can you guess which is which and M journey is the image on the left and you can see here it has this more beautiful tonal range and you can see that the saturation and the HDR the dynamic range is extremely high in flux it almost looks like somebody has put the clarity filter all the way to the top so it also hasn't quite got the Open Arms aspect that we asked for and you can see here that he only has four fingers on this hand so here I would say that they've both done a pretty good job but for me my preference is certainly mid Journey so next up I used another image from one of the flux Galleries and the prompt involved a man and a woman standing against a backdrop it has some complex elements to test the relational nature of the AI art generator you can see that it's asked for the backdrop to be divided equally down the middle now mny has done much better in this instance at accurately outputting an image that matches the prompt that was inputed now it's very interesting for me this shot because I think it reveals a lot about the nature of these two tools so the way it has interpreted these prompts shows us essentially The stylistic Taste the stylistic intention the aesthetic mind of these two tools and you can see that the couple on the right seem to have absolutely no taste in life their sense of style is incredibly cliche and it seems that they are stuck in the early 2000s and that they have a very low budget for buying their clothes I mean even this hairstyle is extremely dated whereas not only are the couple on the left much more attractive as human beings they also have a much more distinguished sense of style and for me you can really get a sense of what it is like to work with mid Journey with it image that comes out on the left and what it's like to work with flux with the image that comes out on the right the image that comes out of flux is certainly more attuned to what you would expect to create yourself if you out photographing people whereas mid Journey certainly has the eye of an art director it is a much more intentional approach to generating art so next up we're going to look at some architecture and this is looking at a skycraper in a futuristic cyberpunk city now the detail is far superior in the mid Journey version as well as the sense of composition you can see that the perspective of the building feels a lot more natural it gives us a sense that it's a much larger building than the one on the right we're in a very strange angle with the flux example it's almost as if we are onethird of the way up the building and perhaps we are a drone flying through the shot there's also an extremely high mountain range in the back here possibly that's a cloud but it seems rather amateurish the details put in from flux you can see we've just got this very odd right angle of a very Stark light it's almost as someone's taken this image and drawn with a neon crayon an extra line on the side whereas some of these fine details com out of mid Journey are absolutely remarkable it's got these huge neon signage coming on the side of the building and what looks like either some antenna eyes poking out that gives it more detail more realism more interest and more variety next up is another prompt from the flux gallery and this was a very beautiful image that came out of flux I very much appreciated tones in this image that we have this wonderful focal point of the orange in the center of the screen it's almost reminiscent of the Sun and this rather happy looking tiny astronaut emerging from the egg now moury slightly misinterpreted The Prompt because we asked for an astronaut being hatched from an egg and I don't know how this astronaut managed to fit in this tiny egg but I do like the glowing nature of the little egg as well as well as the reflection of the egg very clearly depicted in the visor of our astronaut So based on these image comparisons I would have to say that for a cinematic feel mid Journey wins out with depicting hands I would say that they both can perform accurate hand renderings and mid Journey probably just takes this one now with relational prompts this is where flux stands heads and shoulders above mid Journey as well as accurately rendering out typography for character studies I'd have to give it once again to Mid Journey especially when comparing this couple here which is absolutely hilarious to me for architectural studies mid Journey again takes the biscuit and for abstract and surreal interpretations of the world I would say that flux in some ways is imagining something that is more believable and so I would say that actually it does a very good job at interpreting something that would be impossible to generate with with AI now I'd love to know your opinions on these prompt comparisons and let me know if you agree or disagree with my subjective viewpoint but let's move on to one of the most controversial aspects of these AI art generators and that is censorship and this is a key point and one of the reasons why I think flux is incredibly useful and that's because mid Journey said that you're a very naughty boy if you want to create anything that is Gory considered Gore which includes udes anything to do with blood or violence so if you do want to have any dramatic action style war films then mid journey is going to be very challenging for you to create those within now of course also they they do not allow you to put in anything salacious anything containing nudu or shall we say the organs of reproduction and fixation on such things will be punished immediately by the mid Journey team whereas flux is completely uncensored so you can go wild with anything that comes into your imagination now if you are using this online if you're using it in a website and using cloud computing some of the hosts will not allow you to create not safe for work content so I created my examples in replicate and they do not allow you to create not safe for work content however if you go to hugging face which is another place which you can run the flux model then you are allowed to depict the human anatomy amongst other things so in that regard flux has a hand over mid Journey now let's talk a little bit about pricing now the mid Journey basic plan starts at $10 a month and that gives you 200 images a month now I have the $30 a month plan and that gives you an unlimited amount it does give you a limit for fast Generations but I found the relaxed speed to be suitable enough for me and generally I don't usually use up all of my fast hours anyway now you can run flux on your own machine if you have a bucket load of ram if you do that then it's going to be completely free apart from the electricity and the down cost of having a fast machine but you can run it on hugging face or replicate for replicate they're going to charge you about $1 per 333 images and so comparing that to the basic plan of mid Journey it comes out at about 15 times cheaper now for both of these models it's not exactly clear what training data they have used and mid Journey did get into some hot water over the last couple of years for possibly taking a lot of data that it was not supposed to there were some leaked tweets where they were discussing how they had laundered the data by essentially putting it into the machine generating loads of AI art based on famous artworks and copyrighted material and then feeding the AI art that had been generated on the stolen data back into the system and then deleting the original data and also Runway recently has been in the news because they are a very popular AI video generator and it shows that they have a a massive training data set that takes videos from Netflix Nintendo and YouTube and so you would think if you upload a video to YouTube that these AI companies aren't going to take this and train the data of your YouTube videos well apparently they will I wonder how long it will be before I can simply prompt my own YouTube video of myself I would like an AI Samson YouTube video please AI so the summary is that if you are looking for a open- source model that is free to run on your own powerful machine machine then flux is a great option it's also wonderful because it's completely uncensored and if you spend some time defining articulate and precise pumps you can get out some wonderful results now mid Journey on the other hand is more expensive and you can only run it online it is also highly censored and it struggles with complex relational prompts however the pure beauty of the images coming out of mid Journey means that for me it is the outstanding AI art generator of the current time now of course these are not the only AI art generators and there are a couple of others that are important to be aware of for some specific use cases a particular mention goes out to idiogram for its incredible ability to render text it is absolutely stunning at creating Graphics so if you are looking to create graphic design then idiogram is the best option now Leonardo is an interesting hybrid between the Simplicity of mid journey and the complexity of open source models and it gives you that opportunity to take more control over how the algorithm is working but you do sacrifice something in the user experience interestingly about Leonardo is that they were recently acquired by canva so we might expect some huge developments in Leonardo coming up soon especially if canva takes the technology of Leonardo and reinterprets it in its canva way by simplifying it and making it accessible to the masses and another honorable mention is rendered specifically for working with detailed character work where if you want to have a character who is as consistent as possible in different situations then this is undoubtedly the best tool for character consistency but which tool do you like best do let me know in the comments and are there any other AI art generators that you use that I have not mentioned I hope you enjoyed this video and if you are interested in joining the AI filmmaker Academy and being part of a community of creative AI Visionaries then check out the link in the description below I thank you very much for watching I thank you for being here and most of all I wish you a delightful day