@@mato579 Nice! I was just talking to.a buddy about that-- I'm Mac based, so SD is a massive pain to install, but I might finally break down and do it over the weekend. I know its do-able, its just not a very clear-cut process, and will likely take a few hours. Sigh.
@@thatguytravis3933 a-a, bluewillow is far worse, the output it produces is nowhere near in quality to the likes of midjourney or stable diffusion on local powerful pc machine
@@TheoreticallyMedia FBI: "We understand you've been generating robot-on-robot action flicks?! Tim: "Well, err..." FBI: "I was wondering if we could show one at our upcoming fundraiser?"
Oh, the first time I tried Leonardo, I got stuck in a glitch that was hilarious. No matter what model I picked and what completely different prompt I gave it each time (cup of coffee, church, mailman, etc), I kept getting the same weird rainbow blob creature over and over that I’d created first. I actually wish I could recreate the glitch, but apparently I picked a time when a few gazillion other folks were trying Leo out as well, and the program was completely overwhelmed.
I’ll say, as critical as I sometimes am with AI art, I always remember that it is Infant technology. And, I gotta say, sometimes those weird glitches are truly hilarious. Like, I’m laughing about it all day, kinda funny! My other favorite thing is when it parses in a literal way. Like, you prompt: “man with glasses” and you get a guy HOLDING a glass of water. That stuff always amuses me.
I'm a big defender of MidJourney, but I do think it depends a lot on what you're trying to do. For overall quality, MidJourney is definitely in the lead. I'll look at SD stuff for a while and then look at any random profile on MidJourney and go "wow!". And MidJourney has more tools than many realize for non-"one-shot" like using image prompts and the remix feature (changing the prompt as you select variations, and even adding new image prompts during the remix). The advantage of SD is in being able to train your own models, the control you get with the new ControlNet stuff, and in/out painting. Looking at workflows of people that use SD extensively, they tend to use inpainting a lot. This takes more work, but you can do stuff like when you in-paint over a face do it at full resolution and then it scales it back down. Even in V4, MidJourney tends to have worse faces if they are small in the image or there is many people, and you can't fix that within the same program. And people are starting to get cohesive animations going with SD. One thing to consider is actually combining the two. Like producing some images of a character in MidJourney and training a model for it in SD to do re-posing or in-painting. SD does seem to be catching up in some areas, though (at least until MidJourney V5). Maybe not in Leonardo, but using particular models at particular settings and resolutions. I recommend looking at this video to see some of what is possible on the cutting edge: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-AlSCx-4d51U.html
totally agree. You might be interested in a video I just posted-- 10 things we Love/Hate about Midjourney. And I'll say, I'm an MJ subscriber, and a big fan of the work they've done, BUT-- there is a considerable amount nipping at their heels these days. We might be looking at a few months before SD overtakes them, or at least generates equal to-- once that happens, along with In/Out Painting, Pose to Img etc-- I mean...that might spell the end of MJ. Again, to be clear, I'm not rooting for that in any way!
The whiskey dude from Leonardo… that’s almost legendary🤣🤣 i had to laugh so hard, but i was doing cardio in thr gym while watching it😂 ur comments made it even better😂 great video!
I love the paper art style also, I have been doing paper art for the last couple weeks in Midjourney and am very happy with them. I am going to have to try it in Leonardo AI.
It’d be neat to create one in Midjourney or Leonardo and then run it through a circut to “print” it. Admittedly, I don’t own a Cricut, so I don’t know how hard that would actually be…
@Theoretically Media I have the cricut and without a pattern it would be hard. I don't do the curling type of paper, I just ask for things like vibrant paper trees and it spits out amazing pictures.
Leo is also evolving in a different way. They recently (yesterday) added pose to image feature that's definitely an interesting and different change. And though I use both Midjourney and Leo, I can say if you include other ai tools/site then leo and midjourney both will be in first and second place mostly in my type of art (procreate character artist
100% agree. That’s what great about Leo being (for now) free. There’s no need to choose, it’s the best of both worlds! I’m really excited about the pose to image, I was in the Midjourney Open Office chat earlier today and hoped to ask them about that, but alas, I did not get a chance to ask a question.
I forgot to mention that in the latest video, someone asked about Control Net in the MJ Office Hours, and the response was basically: “that’s cool, we might look into it in the future” So, it doesn’t sound like it’ll be anytime soon.
Thank you for this vid, I just signed for early access and joined their Discord. Very curious about the tool, I'm currently using a local install of Stable Diffusion, we'll see how Leonardo develops over the next months.
Glad you caught it! I do need to install SD local at some point, but I’m on a Mac, and that is a real hassle. I know it can be done, but I also have to schedule a few hours of futzing. Please let me know how you find Leo vs Local, once you get a chance to dig in. I know local installs can take a lot of extra maintenance, and I think that’s where these services will prove their worth in sub costs. If time is money, and local takes time…it might be worth subscribing to Leo?
@@TheoreticallyMedia Once you've got all the required/advised Python tools and your Cuda libraries installed, maintaining local is just getting your files updated via Git. I use Automatic1111's fork. I have no experience with a Mac install though, I'm on PC. About costs, local is worth for me because I do a crapton of tests and that would consume a helluva tokens of a paid service, so I had to bend my brain into the more techy approach. Of course it depends on Leonardo costs and in their options, but if it's tokens, then I'll continue with local SD.
Oh seriously?! That was yours?! That’s awesome! Fantastic work! I’ve used they model a few more times since the video, it’s really great! Many thanks to you!
See, I actually like the Discord platform. Mostly so I can generate images on my phone, while I'm "on the go"== these days, i actually don't mind getting stuck in a long line at the grocery store!
@@TheoreticallyMedia I am totally into the discord instead as well. It allows me to sort things into groups as I work on various projects. And mobile absolutely I do the same thing lol
@@prizzmatik Like a bonehead, I only recently started using multiple channels for different projects, and my god, what a game changer that has been for my productivity! Previously, I would be working on different projects in the same stream, and like a hyperactive rabbit, I kept getting sidetracked by various outputs! Plus, c'mon, who doesn't love the Boo-Boop of a Discord notification?!
@@TheoreticallyMedia Oh yeah I am the same way. Easily side tracked by oh, lets change this and then another spark thought. If I couldn't have a different channel for various projects I would drown for sure. Its easy to queue up everything and wait for the messages to come in lol. And so I think that's where leanardo is a bit of trouble. No way to sort or pack projects into different boxes.
I heard that Midjourney uses fine-tuned Stable Diffusion, but does so some kind of secret proprietary magic with prompts, settings, and initial images. Stable Diffusion can get good results, but it requires more detailed prompts, negative prompts, and lots of trial and error.
I'm not sure if they actually used SD code for the test/testp modes they had between V3 and V4, but they've been very clear that V4 is their own proprietary system. They actually said in a recent call that while they'll probably have zoom out pretty soon, doing actual inpainting is harder in the system they have and something they're still working on.
I feel the best of Leonardo is you just need a single word (called prompt) or simple sentence can make great result. Compared to others AI that need a whole paragraph of words to get awesome results.
I'd say Midjourney is about the same-- I mentioned at some point in the video, I'm not a fan of SplatterPrompting, or GorePrompts. 4 Paragraphs of contridictory buzzwords like 8k, Wide Angle, 700mm Lens, RTX Graphics, Natural Photography, Closeup, Panoramic. Haha, That stuff drives me nuts!!
MidJourney does this much better than Leonardo. It's actually when you want something ultra specific that MidJourney lacks the power to make further changes to the generated image.
Leonardo Signature/ Leonardo Creative are Leonardo's "own" models. "Deliberate 1.1" and the rest from that main list are curated from outside and downloadable for use in a1111 and invoke, etc. Then there's the community and other models. Also, pretty sure there was no glitch with the elden ring and robot v zombie results. pretty sure you accidentally hit img2img on the robot v zombie pic. If you look at elden "init strength", you'll see it's .5 for those two images, that means it was img2img
Yeah, someone mentioned the glitch thing in the comments, that makes sense. I'm going to dive in.a little more for Signature and Creative in a follow up video. To be honest, I wasn't overly impressed with Leo Creative, but I'll spend some more time with it, to see what I can get out of it. I'll say, what I've been having a lot of fun with is Pose to Image in Leo, and then using that as an image reference in Midjourney. I did a followup video on the channel featuring it, and it's just been a real blast.
I think the best one of all is stable diffusion on you own pc with all the toold comming out and different trained models because its open source and that you also use ot to create videos.... i think midjourney and leonardo are great tools for sure with some nice functionalities but... its not even close to stable diffusion when you have the right extentions ,tools and model installed on your pc and the great part is that it completely free forever. It might be a bit more complicated due to all the parameters and its a bit complicated to install but once its done its the power in your hands.
Oh, agreed-- But, there's a considerable amount more knowledge and technical know-how to get that up and running, and maintaining it as well. I think AI Art will eventually settle into a version similar to PC Gaming vs Console Gaming. Mass adoption will happen via a simple to use tool (actually, my latest video on Midjourney vs Dall-E's new model talks about this)-- but there will also be a class of "power users" who will be creating images with custom models.
@@TheoreticallyMedia Thats true the toold are still a bit complicated to install and use, but if ive been able to use it lol everyone can ahahahaha. But leonardo and midjourney and also dall-e are still great tools if someone need cool images without a learning curve to big. Next step i cant wait ofr is ai that are able to create convincing 3d models that you can use in a 3d movie or game.
That glitch you got was actually your inappropriate zombie image being loaded into the image to image setting. I've had this happen to me before too when I click on the generated image by accident without even realizing.
Comparing Midjourney with millions of created - and learned - images with Leonardo in it's infancy is pretty much useless comparison. Let's wait and see how Leonardo evolves as it get's practise and more data under it's wings: it's already way better than Midjourney at the same 'age'. Midjourney is the forerunner and thus has a lot more 'experience' and datasets to use in it's practise.
Oh, 100%. With the new Pose-To-Image, plus Canvas, and a few of the other features I mentioned, I think Leo stands to be a good competitor to Midjourney-- Which I'm all for, as that is fertile ground for innovation.
One thing to keep in mind is that Leonardo is just one of many web and offline wrappers around Stable Diffusion. If you look at the SD ecosystem as a whole, it has a huge amount of activity going on. In terms of models that are finetuned, dreamboothed, lora'd, and mixed together. Then there is various add-ons like ControlNet and pix2pix and Offset Noise just in the last few days. It'll be interesting to see how it all shakes out soon...
Yeah. I’m playing with it more and will likely do a follow up video, something like “Leonardo for the Midjourney user” They’re different tools that do the similar things, and I think this video was an interesting experiment in a 1:1, but I’m really starting to feel that Leo and SD offer some really big advantages over MJ, which I think the MJ crowd should be aware of.
My issue with most of what Leo produces especially realistic looks like it was rendered in unreal engine, Quality is great but you can clearly tell its not photo realistic. Its like watching an old TV show in 4k, it just looks out of place. Now saying that, I am optimistic in Leos future and look forward to the competition against midjourney.
Honestly, that's kind of my take on Stable Diffusion overall. There's a followup video to this one, where I use Pose to Image in Leonardo, and prompt it with Unreal, and the results were pretty good...I mean, a lot of extra limbs, but it did look good! There's some catch up to go, but my thought it that it won't be long. I actually kind of worry about MJ in the future, they probably need to start pushing a bit, but they seem to spend a lot of time talking about launching a magazine...which is a really weird passion project for 2023?
I do believe "Leonardo Signature" is their main model, and is much more artistic similar to MidJourney style, also they have "Leonardo Creative" which is trained on stable diffusion 2.1, which is also pretty good.
Hey those hands are nice. Hands in my outputs are always a mess even with negative prompts. I think you should change some of the models you used because same prompts as you gave me very close to midjourney levels and a lot less extra limbs.
@@TheoreticallyMedia Depends on what you want to create of course but I get great results with RPG 4.0 (especially useful in my case since as an aspiring fantasy writer, it helps visualize my characters) it has worked great for me with armor wearing characters too. Leonardo creative and Character Portrait are nice too. Gotta make sure the prompt is good and settings can be changed so keep that in mind when generating. And even with those models you can do something like I do. I create a few versions of great looking faces, then I use other models to generate better looking armor if the other model doesn't and then when I have enough images create my own model and train it with the images I already generated until I get it to produce exactly what I'm looking for. The fact I can edit in canvas as well makes it better and since we are purposely refining the results ourselves it means the AI is learning much faster. I will say Midjourney still gives better results when the prompts are not that specific and detailed but I see Leonardo getting better fast due to all the human input it's getting.
I find it incredibly hard getting the proper results in Leonardo. Usually it just doesn't create what the prompt says at all. Also, anything with significant detail just fails.
I keep getting the extra limbs! Haven’t seen that in awhile with MJ. You might be interested in Pose to Image, I just did a video on that, but even then, I was getting the extra limbs.
Last week I asked Leonardo for a "contemporary vector art illustration of a tree". Three of the returned images were covered in iStock logos. This has really put me off using it.
In my view, I find Midjourney's Images to be very pretty and delicate, but unfortunately, they are not within the budget of an NGO graphic designer like myself who lacks the necessary resources. Therefore, I will continue to use Leonarda's images and focus on improving my skills. Hopefully, in the future, I will be able to afford Midjourney's images. Thanks for sharing~!
Great start of Leonardo. Im impressed of the quality and possibilities so early by release. Midjourney is the best but leonardo is actually very close. Stable Diffusion is only with all the add ons and tools on the same level, but only if you spend a lot of time to set it up. Im actually very happy with midjourney and in anticipation what v 5 will bring us, but leonardo with free tokens daily beats midjourney in overall.
I’ll say, there is a lot of futzing with SD Local, and in some ways, I’m pretty happy with just paying a sub fee to let someone else handle that. I have that eternal futzing gene, where I’ll spend more time optimizing than actually doing, if given the option. Sort of like PC gaming, where you spend more time playing with your settings than you do actually playing the game. It’s why I like MJ, just prompt and go!
I think as time moves on, we're going to see more innovation coming out of Leonardo as well. I've got a video talking about that tomorrow-- Although, I think the monthly subscription details are dropping tomorrow as well! We'll see!
More than "Midjourney vs Leonardo" it would be "Midjourney vs tuned Stable Diffusion". Stable Diffusion has been a bit stagnant for some months, while Midjourney has been having updates. Where Leonardo really shines is in the interface, in Midjourney it's almost nonexistant. I wish the people from Midjourney would abandon Discord, it just limits a great tool in many ways. Btw the picture in picture format you use for your video is huge, it gets in the way of what you are doing. Please make it smaller and 4:3 ratio, thanks.
Yeah, this was a bit of an older video-- at least in AI Time, haha. It's over a month old, which in the AI world is like 10 years. I've got some follow ups on the channel, discussing more about the advantages Leo has over MJ at this point (Pose to Img being a big one)-- and yup: I changed the Picture in Picture look as well! Thanks so much for the comment!
I'm going to spend some time in Bluewillow this week. Seems to be popping a lot in the comments. I really only looked at it at launch, and it wasn't....great? But, a lot has happened since then, so I really feel like I should take a closer look.
I think Blue Willow uses SD models like Leonardo does, but does some magic where it selects the model based on your prompt. Like if you use papercraft, maybe it automatically uses the papercraft model. Which is kind of neat, but at the same time, it is nice to have control over the process.
I wouldn't say killer quite yet. Results are great and on the level of my experience with Midjourney. I love the training modules too. Just waiting till it's open for everyone to see what the the final prices will be. If it's less than Midjourney's ridiculous plans then I'll fully switch
Agreed. For me, if the price is right, I might do a double subscription. If you check out the follow up video to this (Leo Pay), I’ve got a look at Pose to Image TO Midjourney, and I was really impressed with those results. I’m pretty sure I’m subscribed to a few services I’ve forgotten I signed up for, so if I cancel one of those, I should have the coin for Leo. As long as it isn’t stupidly overpriced.
@@TheoreticallyMedia I don't think we'll have long to wait before these tools are completely free due to competition. The availability of open source models, high-quality results, and increasing ease of running these tools on your own machine or web platform will spell the end of pricing.
Leonardo is an encapsulated and tweaked implementation of Stable Diffusion. What that means is that it’s advancing technologically roughly at the speed of Stable Diffusion, which has an incredible open source developer community. Look at ControlNet as an SD extension with incredible power - that will come to Leonardo long before such features show up in Midjourney. Leonardo is rushing to finish implementing features and extensions of SD that already exist - in painting and the canvas just dropped a week ago. So in speed of evolution, there’s no comparison.
100% agree. The landscape is rapidly shifting, and in 4 months (or, maybe less?) I can see MJ taking a tumble, if it doesn't start innovating. I still maintain that they're at the top of the mountain in terms of aesthetic image generation, but becasue of the open source nature of SD, improvements are popping off in all these crazy directions. I think Leonardo's challenge will be to figure out how to implement and streamline all these advances into simple UI. But, if they get that together, plus some really, really good models? Yeah, that'll spell the end of MJ.
@@TheoreticallyMedia Leonardo has some huge advantages. 1) the Discord UI is really kludgy. Leonardo is elegant. They really have thought hard about how to simplify what is a very complex and unfamiliar set of processes 2) Leonardo is leveraging two user communities -- not just Leonardo users, but SD users. And the Leonardo users "training" their own models - that's huge. The number of new models arriving is extraordinary. They also are getting new models from the SD community generally. So, for example, you can see that ControlNet is coming - Leonardo are iterating their "Pose from Image" capabilities. That's _much_ faster than Midjourney, which has been content to deliver "our stuff, when we want it" 3) NSFW - anyone who's been around the Internet a while recognizes the level of interest in adult material. Leonardo permits it for private consumption, with an NSFW tag. . . . means no one else can see it, but its created a huge demand 4) It has a game like "just one more turn" addictiveness to it. People stay up all night doing one more to see how it turns out. 5) model generation is kind of a gold rush at the moment, drinking from a fire hydrant. Midjourney isn't equipped for that. SD, Huggingface, Civitai and the entire ecosystem around it . . . they're in fire hydrant mode - Leonardo is the UI that can exploit it, with less of the chaos of roll your own. (Not many users want to be figuring out why Xformers won't install, for example)
One thing I like midjourney is you don't need a crazy ass prompts to output an image no insane negative prompt or regular prompt. Leonardo works the same way stable diffusion does with is prompt heavy requirement which I find extremely anoyying
Yup! Funny enough, I think when you do provide a really honed prompt, you can get some amazing output. I just put up a video with a formula for creating Cinematic Prompts, if you want to check it out-- It's kind of the "Logic" behind an MJ prompt-- in my opinion at least!
Yes if i need inspiration leonardo can throw a bunch of them since its free. But if i want some more usable pictures then midjourney will most of the time give better result.
Agreed. I’ve been playing around with some stock image stuff, and for sure, Midge is the tool for that job. Ironically, if you’re strolling through a stock image site and literally copy paste the description as an /imagine, there are a few uncanny similarities. Point being: Midjourney makes sellable art.
I'm gonna head over to Bluewillow this week to dive in more. I haven't tried it since launch, and I'm overdue to see how it is developing. The userbase seems to be really passionate about it, and I love that!
Sorry, very new to all this- is it just as good to start learning in Leonardo as it is to start learning with MidJourney or even some other ai image generator? Meaning, before you spend money to subscribe to MidJourney after 10 free images or whatever, is it just as good to learn fro something that’s free (or free for more use before you have to start paying?) Interested in multiple answers on this, thanks! 🙏
Hey Ari! Well, I think MJ has a pause on free trials right now, on account of some abuse shenanigans, but they might have solved that by now. There is a good amount of difference between the two systems. I feel that MJ has an easier learning curve in terms of writing prompts. It'll do a lot of handholding for you, but provide you with stellar results. That said, it can be a challenge to get very specific things out of MJ. Leonardo requires a LOT more prompting, and relies heavily on negative prompts as well. So, not only do you need to be very specific about the things you want to see, but you also have to be specific about the things you DON'T want to see. To be honest, Leo offers a pretty decent free tier right now. Enough for you to play around in and see if it is for you. I'm currently subscribed to Midjourney, but pop images back and forth between the two (using Leo's free tier), but I've been considering adding in the low tier of Leo as another sub. But, yeah-- I totally hear you. The sub model can get out of control quickly. I just went through last month and cancelled a bunch of subscriptions I rarely use, or had even forgotten I had. It got me like $40 a month back!
@@TheoreticallyMedia Hey thanks for the quick reply! It almost sounds like learning the ropes with Leonardo could make a person even more adept with MidJourney-does that sound right? Because it pushes you harder on understanding the “Art” of writing prompts?
Art is subjective, I just find Midjourney over-stylizes. The first 3 you did I thought Leonardo was better. The toast one was a good example. It gave you exactly what was asked with excellent quality. Midjourney added extras that may or may not be wanted. Also, I'm no fan of Discord, so I like the Web UI.
I believe you currently get “free” tokens every day as they are in beta. That said, within a week or so, they’ll likely announce their pricing plan. I presume there will be a monthly tier? But who knows!
I think there are some other alternative sites out there. I'm not familiar with all of them, but Mage I believe has a monthly plan and availability to multiple stable diffusion models (though not training your own that Leonardo has). Just be cautious looking at other people's work with NSFW enabled, since there is a lot of that on the platform.
NEEDING HELP : just went to website and it sent me to page and set up discord (i set all that up) but i still cant figure out how to get to Leonardo itself...im using my phone but idk if that is the issue.. Any clues?
Try going over to Leonardo.AI and logging in. You should get a banner that says "I'm Whitelisted"-- and when you click on that, you should be in. If you don't get that, you might be in the "Waitlist"-- as they closed their doors for a bit, due to getting hammered with new users.
@@TheoreticallyMedia thanks i figured it out soon after send above said message... Ill have to play more when i set up my computer... Not very easy to work using a phone... Hoping this will help my Print on demand Store make some sales again, thanks and will subscribe and watch more of your videos
Agreed, it is really early in. I'm sure that in a few months Leo will have made major strides. That said, I don't think it'll "kill" MJ. I sort of see different branches of a tree. I think both will eventually fall into their respective lanes, have their own strengths and weaknesses, and user bases. I'm all for this as well, as it keeps both companies focused on innovating!
In one respect Leonardo beats Midjourney hands down: It's (still) free. Granted, however, that a lot of work still needs to be put into Leonardo before they can charge a fee for it. I also love the AI Canvas tool as well as the many community models and the fact that I can create my own models.
Agreed-- It'll be interesting when Leonardo decides to pull the trigger on going paid. It's always a tough transition to move a user base from Free to Paid, without having the bulk of them bolt. Particularly in this area, since new tools and services are popping up left and right. Also, really agree on the Canvas and Community Models. If I need some zooming out/outpainting on MJ, I'll bump it over to another site. Keeping all-in-one on MJ would be fantastic, but I'm not sure that's where their focus is right now. As far as the community models...that can be a bit hit or miss. I'm not sure if they've put in a ranking system yet, but from my casual stroll, I'll say there was a LOT of junk models. That comes with the territory, I suppose!
Hi guys, right now it's looking like Leonardo will introduce paid subscription plans soon, while most likely also keeping the current free option available. But it hasn't been finalized yet.
Leo is slowly rolling out the whitelist, so if you haven’t, make sure you sign up at least. I do agree that MJ has that “pop” but SD is starting to catch up. And since it looks like Leon will go paid soon, but still offer a free tier, I think it’ll be good to know both, and kind of bounce back and forth! I just did a video on Leo’s new Image Prompt (it’s the red thumbnail with the gal holding a katana), you might want to check that out to see how far SD and Leo have gotten!
@@TheoreticallyMedia I've signed up for the Leo waitlist. I'm going with MJ monthly for now. I've had difficulty getting the results I want with SD, I have trouble getting accurate faces. It could be my inexperience that's to blame. I also hooked up with Blue Tree thanks one of you videos. Thanks!
I honestly think Midjourney does have hidden prompts and negative prompts for it's creations. So depriving your generation of negative prompts is giving MJ an advantage. You could argue that for an end-user this shouldn't matter. Well that's a question of control, and control matters according to the US copyright office for being able to copyright something. With inpainting, outpainting, controlnet, negativ prompts and if you use the local version of stable diffusion, LoRA, embeddings and hypernetworks as well as hundreds of different scripts, there is a difference. MJ feels like a really nice lootbox in comparison.
I have a follow up video that I did, but basically, they’ve closed their doors for a hot minute, since they got overwhelmed by a flood of new users. My guess is, they’ll be back next week.
UX wise, Leonardo is sooo much better than shitty Discord. But the outputs currently are so vastly different from your review... If MJ teamed up with Leonardo and used their web UI but their ai.... Wow. Otherwise hoping to see leonardo improve over time.
Agreed. To be fair, I kinda like the "command line" typing of a Discord interface, I know I'm in the minority there, but it tickles the creative writer and the technical sides of my brain. But, I realize the the majority of people, a really solid UI with sliders that make sense will be the key to winning the "AI Imagery" war. I think with a few more breakthroughs in SD, Leo has a real shot at becoming top dog in the space. Although, Dall-e 3 is looming in the background...
Your prompts in Leonardo are very basic without detail. Midjourney automatically adds those things. In Leo, you want to be specific. They both have their strengths and weaknesses. If you want to expand a Midjourney image in Midjourney, you can’t. If you generate an image in Midjourney, then take it into Leo’s canvas editor, you can. It’s very difficult and time consuming to get the same model in different poses or facial expressions in Midjourney but very fast and easy in Leo. Midjourney is more beginner friendly but if you know how to prompt by understanding lightning, design style, themes, art mediums, shaders, perspectives, color palettes, etc. You can create beautiful images in Leo for free.
Indeed-- And to be fair, this video was fairly early in with Leo's release, and I think it serves as a pretty good 1:1 of MJ and Leo-- at least from an MJ perspective. I'm starting to dig more into Leo now, and seeing how the workflow differs. On the list of "To Do" videos, is a follow up to this one that will be something along the lines of "Leonardo for the Midjourney User" There IS one follow up video that I did to this one, where I used Leo and MJ in conjunction with one another (Pose to Image) and I thought the results were pretty fantastic. I don't think there's a "One Ring to Rule Them All" application or interface yet, and I really think the best way to go about it right now is to use these tools in tandem with one another. It's just a super fun time for experimenting!
It is starting. Corridor Crew just did a video where they made an Anime short using Stable Diffusion. Now, granted there was a LOT of extra work involved, but it proves that is can be done. Gen-1 from Runway is in early beta, and that is kind of like rotoscoping on steroids. I totally agree though, I think whoever cracks Video is going to be King of the Hill. As far as Text to Video....eh, that's probably going to take a bit longer. Don't get me wrong, what I've seen so far is pretty impressive and neat, but its a long way off from something I would actually watch for entertainment.
@@monikastefan7947 It might be that they're keeping it under wraps until the paid plan starts? Maybe they'll continue offering free, but with no commercial use? I guess we'll find out soon...
Who is gonna come out as Ali?!!! Funny that you mention that fight, I JUST found out that it was watched by over a BILLION people, live! For 1975? That is INSANE!
@@TheoreticallyMedia wow I didn't know that.I give Leonardo a few months and the prompts should be better. Will midj do some major updates?, we shall see
@@dynamomixx Midge has been teasing their v5 model. I'm actually listening to their "Open Office" chat right now, but I missed the beginning, where I think topline announcements would be-- so, I've got nothing right now. Leo, I think will develop and get better-- BUT, the rub is gonna come when they start charging. That's always a problem with the Free to Paid model!
Upcoming. The Bluewillow community seems really passionate about it, and I love that. In the follow up to this video (linked at the end) I give BW a shoutout; Gonna spend some time there this week!
I was saying in another comment, I gotta circle back to Bluewillow soon. I checked it out really early, like basically at launch, so I'm sure it has improved a ton since then.
I can't disagree there. But, SD moves closer and closer everyday. It's going to be really interesting to see what MJ v5 has in store for us. Given the latest developments on the SD side (Control Net), I'd think the MJ team has to be feeling some pressure. Not to self promo, but if you'd like I've got a video on the channel (10 Things we Love/Hate about Midjourney) where I go into it a bit, if you'd like to check it out!
More or less? But I think their aim/goal is to have an easy to use UI for people that don't want to mess around with installs and the back end. We're basically looking at a gold rush to the "Spotify for AI Art"-- a really simple to use AI art generator, that everyone can use.
I'm new to AI art, and I really enjoy it, but I have never in my life been exposed to so much CP-related material in my life, whenever I browse some other user's channels, feeds, libraries, models, etc. It's extremely creepy and disgusting, and it will leave a permanent stigma on this type of art until it is properly dealt with.
Agreed. I kinda called it out a bit in this video. There’s a lot of…male gaze? I presume that’ll level out as the population of AI creators starts to diversify a bit. (Says the old married white dude, who made the video! Ha!)
It's been mentioned, so likely. I think this is mostly a public beta right now. Or at least a smart play to get everyone invested in the system, so when they start charging, you're more likely to join their subscription service. Personally, I think they are going to be a bit before they pull the trigger on paid. And given they're building off a free community, I think they'll probably go the route of continuing to offer a free, but limited, option. Just speculation on my part.
Midjourney should start planning to increase their 25 free limit immediately. Alternatively, AI tools like Blue Willow will gain in popularity over time. I am using Blue Willow and taking part in their daily theme challenge is fun for me. Have you tried BW?
I tried BW really early in-- like, launch week, and (at the time) it wasn't quite there yet. That said, there have been so many comments lately about it that I feel like I need to schedule some time to go back over and check it out for a proper review. I'll say, the community of Bluewillow users are SUPER passionate about it. I do love that!
Yea, the downside of Leonardo - i doubt you can download your model to re upload to other places, i note that Leondaro looks like it's similar to SEEK, and others running on Invoke. Ehh i prefer spending money on Google colab, doing it myself and having fun XD
Oh, I'm SURE you can't download and move models around. I wish I had the time/funds to really get into Collab Notebooks, but there's only so many hours in a day! Glad that there are folks like you that are going the independent route though! That's where innovation and new ideas come from!
Oh, I got in yesterday? That's weird. Although, I did just check the Discord and apparently they have paused user intakes for now. I'm sure it won't be long before they're back up though. Did you get in?
I think they opened it back up, or at least did for a bit. I would say everyone should still try. It just might take longer than 24 hours to get in (for me it was a few days).
The problem with midjourney is for the average person to get started its soooo complicated. U need to know discord, then how to add a server, then do all the little side tasks to get up n goin, then u enter a weird prompt language world that definitly not user friendly. They r probably missing out on 70% of their customers because of their overly complex user interface and prompt logic. Whereas Leo is web-based and u JUST TYPE and hit ENTER. Remember guys 96% of the world has a below than average IQ
I get you. There is a bit of a learning curve to MJ, but I do think the Discord platform is worth getting into, as it offers so many other amazing communities/topics, once you start exploring it. Curious, do you think it's worth doing a "Midjourney for Dummies" tutorial? Ha, and your 96% stat reminds me of the old quote: "Two Things Are Infinite: the Universe and Human Stupidity" Basically, you AREN'T wrong!
@@TheoreticallyMedia no u dont get my point - discord is without a question superrr useful and beyond! What i am saying is the average HUMAN which is the biggest slice of the market as far as customers will NOT use discord, let alone the complex prompt logic MJ uses - therefore they will be turned off and will go lookin for a much easier web based instant text prompter - people basically want something like what shutterstock did - u enter the website and instantly an empty search bar awaits you to just type ur prompt and boom! all images created - this is about 80% of the population -
@@thebicycleman8062 ahhhhh, I getcha! And you know, I might honestly think of that as a feature more than a bug! Over the years, I’ve had so many jobs, ranging from Photographer to Help Desk/IT, where I’ve thought: “do people have any idea how easy this is?” And then realized: “No. and thank god, or I’d be out of a job.” I have been paid quite handsomely in the past to run Windows updates and reboot machines. Haha.
Well, for this week at least. They just announced: "We will be introducing paid plans later this week for users who want more generous allowances and private generations." So, the slow price rollout is starting.
@@jochemc.6915 seems to be the case from the looks of it. Faster and private gen for paid, slower and public for free. My guess is that tokens will slowly increase in price, to the eventual 25 daily images to try to convert free to paid. The old Frog in Boiling water method.
@@TheoreticallyMedia We'll see, hopefully later this week or next week. I'm pretty active on the Leonardo Discord (great community btw) and so far it doesn't seem to me that the devs are as money-hungry as Midjourney for example, MJ's pricing and "free plan" are a joke. I'm personally thinking the pricing plan(s) will be similar to PlaygroundAI which is quite generous.
@@jochemc.6915 I'm going to start spending a little more time over in that community, agreed that everyone seems pretty cool over there! I've got some thoughts on the pricing plan (which is obviously just speculation at this point) coming up in a video for later today! (plus covering Pose to Image!)
I hate fallend it Killer, it isnt a killer. It is a good Otter Tool to generate nice Ai Pictures. Midjourney is Limited to 25 Pictures, and i ve no have a Card, you have no Chance to enjoy it. I like Booth Tools:))
I agree. That's in part why I made this video. There were so many others in the RU-vid "hype" cycle of "This is GONNA SMASH MJ!" -- I wanted to take a slightly more level headed look. Just a breakdown of Leo, and a simple prompt based head to head!
all these monetization methods are bad. Paying monthly fee is expensive. They do not make another TV show. They add other computations by using AI and remote working and sitting in front of PC. Task is much cheaper than reason why Netflix has subscription. Let us see if they can just make a show that is critically aclaimed. Why force us to pay all these tokens and whatever else they come up. Soon this will all be out there for free anyways
Free is one direction, but I think some company will come along with a modest fee and a killer UI, and that’ll be the “Spotify of AI art” Something easy to use, offers a lot, and has a low fee. That’ll be the company that cleans up.
That's a first! Although, I kinda at least hope you're thinking 80s Marty? Hmmm...Maybe I should grow a beard? The one comment that cracks me up to this day: I did a Reddit live stream, and someone said: "This dude looks like Robert Downey Jr, if Robert Downey Jr still did drugs." Still laughing!
tried patterns with leo, no matter what I do, and I've tried everything, the patterns turn out terrible. seamless patterns are not seamless. prompted images do not exist or turn into blobs. unusable even as a modern art piece... giving up on leo for sure
I don't mind it being web based, either. That said, I do like MJ's Discord interface, mostly since I can prompt and create while out and about. I'm not sure that would be very fun to do on a mobile web browser!
Try going back to the website and clicking the “I’ve been whitelisted” prompt, I think they’re a bit buggy right now, as they’ve been upgrading to a much larger user base.
Yeah, looks like they've gotten slammed with new users over the last few days. It is back up for me now, but they have apparently paused new users for the time being.
Midjourney is pretty prohibitive with some styles of art or even with some forbbiden prompts, for example "pin-up", I prefer Leonardo A.I over Blue Willow or Midjourney by far.
So this video was published really early into Leo. I’ve done some more videos on Leo since, and I’ll say for the most part is has won me over. MJ is still my bread and butter, but I’ll hop between the two quite often.
I'll say, it's been awhile since this video, and Leo has made a LOT of improvements since then. If you haven't been over there, you might want to check them out. They're still lagging behind MJ at this point, but it is getting closer.
I don’t think I said invites? Well, regardless, you’re right- at least for right now, I did a follow up video where I mentioned they’re closed for new users currently. BUT: someone just mentioned in the comments that they’re open again?
