Это видео недоступно.
Сожалеем об этом.
MidJourney Has Competition (And It's Free To Use)!
HTML-код
- Опубликовано: 14 авг 2024
- Leonardo is probably the closest competitor that I believe could eventually pass MidJourney. In this video, I give you a complete overview of how Leonard AI works.
Discover More From Me:
🛠️ Explore hundreds of AI Tools: futuretools.io/
📰 Weekly Newsletter: www.futuretool...
😊 Discord Community: futuretools.io...
🐤 Follow me on Twitter: / mreflow
🐺 My personal blog: mattwolfe.com/
🌯 Buy me a burrito: ko-fi.com/matt...
Resources From Today's Video:
leonardo.ai/?r...
Outro music generated by Mubert mubert.com/render
#AINews #AITools #GenerativeArt
People at Midjourney really need to get a user interface like this. I'm so sick of using Discord to generate my stuff
Yes man it feels like peeing in the public.
@@friendlyvimana i belive you can import the bot to a private server
@@friendlyvimana you can generate privately, but I just mean the discord interface is so limited. For example, I hate having to open my generation in the browser, then save it to a folder. And having to click 1 by 1 for upscaling each of the 4 images. There should be an option for automation to handle this stuff. And more broadly, all this incredible stuff shown in the video should be added to Midjourney inside a standalone desktop app or browser interface. God knows they've got the money
@@friendlyvimana 😄😄😄
I read somewhere that they were working on such a UI. Can't remember where.
LeonardoAI improved so fast, this will be like a combination of midjourney and stable diffusion ... I really love the free 150 credits per day which is enough for people like me who does not generate much AI image generation but also does not want to spend a penny.
I don't understand why mid journey does not offer the same free tokens per day deal. I guess they just don't want new people to play with their tool and join their ecosystem...
@@ares106midjourney support costs per day are too big. Free users would only increase the cost while reducing the speed of work
shhhhhhh Leonardo might hear you and cut off the free tokens
@@rizztraininghamster i suppose, I wouldn't know the first thing about how to run such a business, at least as a consumer it seems like a bad approach from my vantage point, but maybe they know what they are doing.
@@ares106 Midjourney does give a free trial of images. Leonardo is probably losing money with how much they are giving away in order to get people to use them. They will probably get more restrictive in the future.
"Photoshop of AI" - That sold me instantly. The UI, the tools, the unique model creation, all of this is just fantastic, exactly the direction it needed to go..Writing complex prompts is just too niche to be the target state of AI art - You can quote me.
Team Leonardo here 💪Thanks for the good guidance, Matt! Keep it up :D
The website seems glitchy sir
@@alexanderangay8634 didnt have issues with it
When my time ran out on Midjourney two weeks in, I shifted to Leonardo, and got this great image I liked. Once my monthly hours reset on Midjourney, I brought the image into Midjourney with “Describe” and then used those prompts, and my Leonardo Prompts to come up with the best images I’ve ever gotten. Totally jaw dropping.
Definitely going to try this. Some of my best midjourney images came from gpt-4 prompts. Looking forward to mixing and matching all 3, hadn’t heard of this one until just now
What do you mean about your time running out? What’s that about!?
@@travelwell6049 Free model only allows 1 editing tries per month (I think it's 1 it may be a few more) so time referred to is time until next free edits
In my opinion, if this gets just a little better, Midjourney is done. I would much rather use this tidy interface than the discord style prompting. And their editing tools are incredible. Reminds me of Nvidia's GAUGAN
Midjourney will also get only better so no way in the short term
@@simonsanchezkumrich8489 of course midjourney will get better too. But i feel like they will eventually reach a point when the difference would be too small to notice. But what stands out to me is the features of Leonardo. The fact that it's not on discord but actually has its own ui. It has a prompt generator that helps you write prompts, not only acknowledging its own limitations but helping you overcome them without too much extra work. The ai editing and the ability to expand the frame. The fact that you can do this with real pictures. The masking. The fact that it's free. This is so much more than midjourney. The fact that the image generation is a little worse means so little when theres so much more
@@prashantsarkar821 that's right 👍🏻 probably both will be amazing and useful but each will have different uses
They posted in their discord that they are prioritizing invites for those who join the discord. there is a link to it once they signup for the waitlist on the site you linked. Thanks as always bro!
I've been playing with leonardo, it's actually pretty impressive, and the online web UI is really good and easy to use! They've done a great job imo
I hope the stable diffusion community can catch up to this. Usability is so important, its not only the features, or flexibility
InvokeAI does everything shown here plus more.
Invoke, ComfyUI and A1111 are the ways to do it. Plus there is google colab to go around your computer limitations.
I was dumb and paid for the base MJ before I was granted access to LeonardoAI. I enjoy LeonardoAI interface so much more. Love the interface and tools it has.
I'm definitely happy with my Stable Diffusion setup, I think I finally got it down to work the way I like it, so I don't think I'll be switching over to Leonardo as my main image generator (perhaps for extra images not part of my main workflow) but, I do like the extra features it has; that canvas tool could come in handy for my own AI generated images someday, or even regular images of my own that I'd like to expand on.
The canvas is like outpainting on SD.
Photopea extension for SD is very powerful.
Canvas is basically a copy of OpenOutpaint
The canvas tool looks like a direct copy of Invoke AI, which is definitely not a bad thing.
I use bit all i can and enjoy, passing also pictures and prompts from one to the others, often with stunning results. Limiting myself isn't so much my thing, especially for AI tech changing that quick now.
This UI looks like a dream come true. Here are some people that really understand UI design!
Playground has already a very similar UI, it's not entirely free, but you can generate lot pictures for free each day till you reach the limits. Lot modifiers and possibilities to. Then sure you can suscribe a plan, to get more picture per day/faster, and make pictures privat, and other stuff.
I've been using Leonardo for 2-3 months and there are updates every few days, it seems, usually adding new features and improvements. So the Devs are constantly working on it and it gets better all the time. My hard drive is fast filling up with awesome artwork. 😁
Are we getting 150 tokens for lifetime? And then we should buy more?
@@AniForgerYT per day 150
@@MaxiAnime-nf8pu That's great. I thought it was only 150 free credits and then you need a paid subscription for more.
@@martiddy nope 150 per day he too said it in the vid
Thanks a lot
Stable Diffusion is, by far, the most powerful and flaxible image generation. At the cost of being very complicated to use it opens a whole new universe of customizations with checkpoints, LoRAs, Text Inversions and many more ways to fine tune the output to match almost exactly what you have in mind.
My waitlist spot opened in 5 minutes. Thanks for bringing this one to our attention!
nice to see more commercial competition to midjourney - even though i mostly just use auto111.
Thanks for your contents Matt!
Cool. I entered the waiting list.... and subscribed to your channel.
I’ve been using Leonardo for a couple months now, and last week I realised something. The uptime has been rock solid. I was using midjourney back before stable diffusion and there was a lot of downtime in the beginning.
Also the controlnet and in painting with Leonardo is so convenient 👌
The waitlist and 150 credits per day were probably a good idea, because it helps them rate limit, to maintain this uptime
AI art creates such rich images that helps someone be able to write detailed scenery and add better context to a story.
That's one of the reasons I'm drawn to AI Art. I have such a hard time describing settings for world building and this makes it so much easier!
Custom models is HUGE! This is really exciting.
For whatever reason RUclips does not put your videos into my feed and I look for them. You are now officially the only channel that I have notifications turned on for. You are hands down the best source for AI information that I have come across. Thanks for putting these videos out.
Leonardo is fantastic! I have been using it for a while now. The models are amazing!! I love that your can train your own models and the canva option is really amazing.
I have been on a waitlist for months now
The proplem with 3d figure is it doesn't have a proper UV map, it needs to be unwrapped 1st
Btw I wouldn't consider Midjourney king of realism, a lot of its images still have that artistic feel even when I fill the prompt will hella lot of (realistic) tags or keywords. With Stable diffusion I've been able to generate realistic images that can legit fool people into thinking it's a real person or place + the fact that I can change a cartoonish/animish pictures into realistic or anime pictures into realistic is a huge W and also love how I can make cosplay of newer characters with ease using LyCORIS, and obviously control net through which I've been able to choose a great picture with a lovely pose and use it's pose + hands to generate 1000s of variation from anime, cartoonish, semi 2d, realistic, RPG, and all that
MidJourney is like buying pre-assembled legos.
Agreed. I've shown images generated by Stable Diffusion to people as a sort of blind test and they straight up thought it was a real image. Even upon closer inspection, they couldn't tell unless I'd told them it was AI generated. They look that good. You just have to find the right prompts and model, but it can do it! :)
Jep I really like they have an interface makes it user friendly to use. Using discord for MidJourney is really cumbersome and all the text is mashed together, making it hard to get an overview.
Just want say thank you Matt! You made a video awhile back about future tools website which led me to Leonardo before it got popular. Its been a dream come true because i can use it to bring my dreams to life. I'm hoping to have a prototype of my board game by the end of the year. So again thank you big time 😊
3 Videos a week is a good rate IMHO. And congrats on the subs!
You're a LEGEND! Kepp it going! Love your stuff! Your literally saving me years of my life with this information. Keep it up!
This is more user friendly than midjourney. It is easier to handle.
Great demo, Matt. Question, the "extending the photo" feature is fantastic. I can think some helpful use cases for that. I'm on the waitlist. I'm wondering if it would replace some MidJourney faces that still come out distorted?
It can be used to fix faces to some degree of success.
Id say no. The output is not as good as MJ so your not going to get the same styling.
If you look closely at the wolf outpainting part of the video. Look at the boulder the wolf is stood on and then look at the boulder its generating.
In photoshop theres a stamp/clone tool. They need to add this so the generation can pick up the style. If not then its useless.
@@armondtanz it's beta, it will continue to improve further as time passes by
Yep that was most impressive to me too.
I've had early access for about two weeks. It has changed the game for me. Thanks for this awesome tutorial. I will watch it several times. Cheers, Scully. By the way, I'm a narrator, and I use this stuff to create my overlays that I read to.
15:15 The obj you uploaded looks like it does not have UVs. it's basically like taking the model and unfolding it to a flat canvas. The AI looks like it is generating a texture for the canvas but the model hasn't been unfolded to that canvas yet. When you see the model as a solid color that usually means all the UVs are at the same point (the top left single pixel of the texture). I don't know for sure that this is the issue because I do not have access to leonardo yet.
What is the best AI photo tool to use to generate photos of myself dressed in costumes with specific facial features and motions? I want to be able to generate hyper-realistic images.
This a paid/sponsored video? You should really disclose that up front
to have the ability to add to a IRL image is impressive but also clean asf.
Esse canal é pura inspiração! Trás muito conteúdo novo pra quem curte IA! Parabéns Matt Wolfe! Saudações Brasileiras✌
"the photoshop of ai" will probably be adobe firefly I bet
I finally got approved for Leonardo last night and I have to say I could not produce images of the quality you were getting, other than when I used the "Little Monsters" setting and generated tiny Kaiju. I would really like to create LORAs or what's that other one called? ROOPs or something? I have not been able to get anything to work that creates those. I'm an old dog struggling to learn these new tricks as fast as I can. I need a teenager here to explain things to me LOL
Would love to see a video on how to generate consistent characters using Leonardo
Maybe the reason way your results in the 3D model are not showing correctly is because the models haven't UVs ?
Great info. Thanks for this.
I have been using Leonardo and ur totally right. They are building a great structure and a real competition especially when it comes to digital art and stuff other than photorealistic scenes
Good. Absolutely ludacris that MJ doesn’t let you generate using automation or even let you have more than one account. They told me if everyone at my work already owned a mj account our company would not be capable of having its own for business purposes because one of us would have to have two accounts. Like what tf are they smoking
That Canvas feature is top tier
Leonardo Diffusion in such a clever name, I love it
I have both paid leonardo and midjourney accounts. Midjourney 5's photo realism can't be matched as Leonardo hovers around Midjourney 4's level of realism. With that said Leonardo is by far the least restrictive/moderated of the two. Midjourney 5's AI Moderator tends to flag a lot of content something they really need to work on. Leonardo definitely offers a lot more render options and granular level control pre and post rendering.
your channel has made my world so much better
Unfortunately, you have to join Discord, and I can't get access to it.
Good stuff Matt!
Great video and it's nice to see this improve. I tried it awhile back and wasn't too impressed but it's always great to see improvements to free stuff. I'll stick with A1111 for the time being as I can generate as much as I want and it's giving me what I want. I also use Prompt Hunt for when I get an idea while on my iPad. 20 free images a day is great but it's also very cheap for 500 a day.
Obviously, it's subjective but I wouldn't call MJ realistic, at least not based on most of the images I've seen as they've all been stylised. I don't see the appeal of it myself, especially with it's terrible Discord based UI.
My word? Have u seen the realism on MJ? they are off the chart.
MJ5 is default is just realism.
I recommend u have a go. Make sure you select MJ5 in settings and everything you type will look super real. I did some mock ups on a soda can. The hand holding the can was amazing.
I wanted a blank can ang MJ wasnt doing a blank can. So i just brought in a picture of a hand holding a blank can and it got it.
I think to get best results use:
Describe
Image promts
Photoshop
What is prompt hunt?
@@armondtanz It's a website for AI image generation. YT doesn't like links so googly it and it should pop up as one of the top options. 🙂
That was really cool about the canvas. Wasted several tokens but eventually was able to fix a poorly cropped image generation. Thanks for sharing this. :D - Heidi
Thank you Matt ❤🙏
personally this is much better than MJ... 150 credits a day is more than enough for someone like me that just likes to randomly create things whenever I'm feeling creative and imaginative
calling it "the photoshop of AI" couldn't be more accurate, and I'm super excited to see what happens to this in the future!!!!
Another brilliant video - you are an asset to the AI community!
Leonardo should allow for people to rent out their gpus via an app, for use within their software setup. They could let you say, give away your gpu processing power to be used by others, in exchange for more points.
Just make it so that the gpu rates are fair, not 1 to 1.. But maybe 1 to .75. This way people can still utilize their gpus, AND they can help scale this project.
i wish u went more indepth into the training side cause thats what sets it apart from othersites like playground, they are the only ones who offer this i believe. playground offers more free images and it has canvas and cool filters/models so i'm reluctant to move over to leonardo but the ability to train models is really drawing me to leonardo
Completely agree, i use Playground nearly on a daily base, this with other AI-sites of course. Playground has lot nice models, and similar UI, canvas and more. The only thing is the training part. Training costs also lot performance, so i'm very curious about this part to.
Surscribed by curiosity, will see how long it takes till i can access it, even if i'm not located in the US. 😂
Hi, can you actually create a corresponding full-body picture from a face/portrait picture? If you add further properties such as full body shot etc. you get a completely different picture of the impression. And in the canvas you can't reasonably extend over such large areas. Best regards
Im in mid journey user, but that AI canvas and Net was awesome interface. If mid journey not letting user to have control like this. I think others will for sure find alternative like leonardo which look very promising as well.
Yea. And their censorship is off the chart. I tried to create a gangster pimp surrounded by bikini models and i couldnt upload any pics of women in bikinis. Flagged as nsfw images.
Matt, your videos are always so helpful. Thanks for doing what you do.
Leonardo is the gpt4 of images. The quality is untouchable.
Super useful, I've signed up for the early access, looking forward to exploring training it on my own styles! see what I gets 😃
Love it and been using it for 2 months. The options and interface is amazing!
At @8:30 that is a GAMECHANGER!! MIDjourney needs to add that...Wow...OVERALL it seems harder to use than MJ.
Leonardo, AI really climbed really quick.. I have a comic book Model that i can create the most beautiful comic book accurate character models.. I love it! Thanks for this!
15:37 When this small feature will work... DJIIIIZZZZZZZZZZZZ that's INSANE
"Thank you for sharing this video. It was interesting to learn about the competition between Away and Midjourney in the luggage industry. As someone who is new to the world of entrepreneurship, it was helpful to see the challenges that startups face when trying to establish themselves in a crowded market. I appreciated the insights provided on how Midjourney could differentiate itself from Away to gain a competitive advantage. Overall, a great video that I learned a lot from."
Comments above was generated by chatgpt . Too lazy to watch the entire video so chatgpt summarized it for me and made comments on my behalf figures crossed it was good comment about the video 😅
Texture might be for material texture, and not a complete UV Map for any object.
The only good thing here is the canvas part where you can paint the area you wish to add or edit. Its basically the same as stable diffusion Inpaint. The the pro also is that it has its own server, while stable diffusion localhost it eats your computer resource.
Maaaan! where the heck is your Patreon!! your stuff are just amazing, you deserve a lot of support!
i think the 3D modeling didn't work because your model of a person is all one surface, whereas a normal model of a person would have the hair modeled, the outfit shaped onto the body, the shoes, etc... you just have one surface of a human.... that COULD be the issue, idk really... I know nothing about 3d modeling but this seems to make sense with what you outcome was :) I hope this help[s
oh and THANK YOU for EVERYTHING you have taught me about AI...I have learned SO MUCH from your videos, thanks bro!
Leonardo AI is fantastic, I've been using it for months. It's not quite as powerful and versatile as certain other popular AIs but it's more than capable of producing content every bit as good as the best out there.
And it's free.
Leonardo can come up with fast nice looking base images from the available trained models but when you try to deviate from the model your image is based on it becomes a chore to use those types of images with welknown photography or other artistic styles.
You have a new subscriber cos of this video. Thank you for doing this little tutorial
I started using Bing image creator last night. I was on for 2 hours just creating art. It was completely free and no credits needed.
I use it also, and in my eyes the AI generation is very similar to Leonardoai and MJ.
@@louisd95714 I will give that a shot.
if it can generate the hands correct. I am in Leonardo
maybe in order to generate a texture for a 3D model, the 3D model must first have UV's? maybe that would be the reason why it failed to generate the textured you wanted
agree!! Leonardo is really Amazing.
I suspect that the free 3d model wasn't correctly UV unwrapped. I assume that it's generating a square tile and using the UV island's as guidance... So it's not that your prompt didn't work. It's more likely that the model was broken
So Leonardo is just Stable diffusion with a simple website but an upcoming subscription
I need an AI LLM app that can check whole documents of 200-plus pages. The current GPT4 character limits are a real problem for me. I don't want to split off chunks all the time. Any suggestions?
ChatPDF?
Try mpt-7b-storywriter, it says it can handle a context length of 65k tokens
Thank you I have an account but I didn’t know how to use all the features thank you for this video
Thanks man, i gave a try and is really nice, as a 3d artist i see as a nice tool to help and a find it more intuitive and resourceful than midjourney
There many others doing the same. Leonardo push for publicity more than perf. Training models is the only feature most other sites don't have... at moment.
The discord is a mess, and when it comes to price per image it's similar to other sites, MJ excluded of course, that's fully Discord anyway. It's easy to give lot token when each image cost more than on other site.
For instance not conviced at all by what they provide.
For my part i will stay on playground, catbird and others that works well for me.
Thanks for the tip on Catbird
This really fights midjourney!
I wish MJ had all those features, especially the training models feature... we need the ability to make consistent characters!!
Gonna check out this Leonardo, maybe I'll even switch to it
MJ are working on features like this, they're just too busy focusing on Version 6 and 7
as always . well done sir !
Matt Wolfe you should try the texture generation using a 3D model with UV coordinates. The model you used is meant to be a base for sculpting, that is why it's not UV mapped and can't show any textures.
Thanks for the video.
I don't know man. Using Stable Diffusion, with a custom model, LORAs, and proper prompting gives very realistic results as well.
Really excited about being able to train AI on my own art
It's been like that, especially if you get the prompt right - it can end up being on par with midjourney.
Midjourney slightly still better off the jump without having to do much
I started using Leonardo yesterday. Very very impressive but has scope for improvement. Still far better inages than from automatic 1111.
Leonardo Ai is the bigger brother of Playground Ai. I just wish that Leonardo would have the same amount of daily credits as Playground.
Waiting to get whitelisted for a weeks. I am considering paying 1 month of Midjourney, because the AI world is so fast that a few more weeks could mean a huge loss of experience with ai generative art. But Leonardo looks great (if you have acces).
best thing about leonardoai is that you can train your model for free!
Thank you Matt. Excellent, again. Sunny greetings from Germany.
I really appreciate you too! Thanks for all this info!
This is truly amazing, and I can't wait to try it out. But I can't find any information on where the intend to go in the future. Running a platform like this isn't free. I assume that eventually it's going to cost. But how much? Can we eventually be able to download and run this on our own computers (even if they charge a fee for the app)?
The isometric tiny rooms should be used in the Sims 1 remaster, lol
3d artist here . the issue is with the character is that it doesn't have any UV maps since the model u uploaded is a High poly model