Hey everyone! Since this was posted 5 months ago prior to the paid plans being announced, please be aware the free membership is now 150 tokens per day and 1 model can be trained and retained/mth.
Thanks for the great review! Just wanted to add a few quick notes 🙂 1. The usual render time for 4 x 768x768 images is actually closer to 15 seconds, we had a huge influx of users yesterday due to a number of YT videos dropping. 2. We're working to improve outpainting, you should actually be able to use it with images generated with other models to extend images. Sometimes it helps to erase some of the edge before outpainting, it can do a better job that way. It sometimes also helps to simplify the prompt when extending the image. We also find the negative prompt to generally not be necessary when outpainting. 3. Unzooming is currently in beta and is meant to actually generate more of the underlying image, sort of like pulling the camera back. We're improving this too, there's just a lot happening in parallel 😂 4. The current free tier allowance is being adjusted to 500 tokens a day, and 1 token will soon equate to a standard 512x512 render. So 250 x 768x768 renders per day.
Thanks so much for taking the time to clarify. I'll mention these things on the next video. I'm really looking forward to see where Leonardo AI goes, very, very promising! 🙌👊
I'd love to hear your thoughts on Leonardo AI. Are you just as impressed as I was? I decided to cover the data set training separately to make this video shorter. Hoping to release it this weekend! Stay tuned. 👍
I'm really enjoying it, it has so much to offer. To think where it will be in a few months. I appreciate you stopping by and taking the time to leave feedback 🙌 I'll have to check out your video too!
Nice! I use Stable Diffusion on my home PC, so I recognize a lot of those features. Glad to see an app that is trying to keep up with all of the open source extensions and models. Everything is moving so fast.
Indeed Robert. I still feel that running it locally gives you the most flexibility however these web ui's are improving so fast making it more accessible to more people that may not have the gear to run it locally. Definitely a positive direction for all of us! What are you using? Automatic1111, Invoke AI?
I've been wasting tokens, trying to describe specifically what I want, and relying on image to image...this is a game changer for me. Far fewer wasted tokens. That's superb.
This is a fantastic AI generator. Not sure if you're aware, but the seed option can, like MidJurney generate images tailored to the selected seed so the new images from the seed are generated similarly. One thing I would love for these AI generator platforms is offer a preview, so the image can be reviewed before the actual generation. THis way, there would be no wasted images, but you currently can't beat the pricing.
Hey Louis! I appreciate that thank you. Yes I'm aware of seeds and how they work but good to know that it's like MJ. I'm going to do some tests for "consistent characters" since this is something our community wants more info on. Hoping Leonardo AI can produce what we are all looking for.
Another AI scam, it's main attraction is to the lazy, that's why these type of Pictraps are so enticing, No talent needed, It's not art by any definition, just another art scam.
@@davewills6121 Agreed, and it’s perfect for me. Not everyone is talented to be an artist. I have tremendous respect and admiration for those who are talented.
Hello my friend, thanks as always for supporting! Means a lot! I'm messing around with a trained model today and will likely record in a few hours to hopefully post by the evening. Stay tuned! Have you signed up yet?
@@davidbackham100 ok cool! I know they might start limiting access soon as this is still beta and with the recent influx, it won't be long before they limit the service. Paid option is on the horizon but if they are smart, they will still offer a decent free option.
You're very welcome! Those are for emphasis but it doesn't work on Leonardo AI just yet, I was copy and pasting a prompt that I commonly use when running SD locally. They did mention to me though that they will implement weighting eventually.
Would be good if you supplied the prompt you were using... I assume there are several entries in that prompt that would have been interesting for beginners.
This AI stuff is amazing. Like I’ve said before, totally over my head, but on discord or in your last video you were doing something that turned photos into comics. What was that? I had an idea for a sequence in one of my running videos. Hope you’re well, bro!
Hey bro great to see ya here! Was it the one on creating your own avatar? But yeah this stuff is moving fast and it's had to keep up! hahaha! Great to see you doing your thing on your channel man, keep going! or Running!
@@MonzonMedia That sounds right! I'll search for it. Thanks for the support...just posting for the fun of it at this point. I've realized I enjoy messing around with filming and editing, but no real passion for any type of niche. It changes depending on the day haha Have a great weekend!
Hey Dirk, I appreciate it. Yeah these guys have pretty much been paying attention to what users want. Feels like midjourney, with the ease of use of playground but leveled up! Going to be exciting to see how they improve it!
@@MonzonMedia I got my invitation yesterday. So far my impression is that playgroundai generates more quickly (and has more free generations). Upscaling is much slower in leonardo. Quality wise I didn't notice much difference when using the stable diffusion 1.5 model. I didn't see facial restoration in leonardo(?), which is a big thing to me. But with a finetuned model, you may not need that so much. Finetuned models generate better quality, but the results are very similar when using the same prompt. A big advantage for leonardo is that you can train a model. If you are productive then a limitation of 10/month may still be a downside. For training a character, you basically just need one full length picture (so far I tested this by uploading the same picture several times). And then you can get a fairly consistent character in different posing (back profile, side profile, sitting, lying down, smiling etc) which keeps the same type of clothes(!). The AI does tend to change the colors of the clothes, so I specify the color in the prompt. Outpainting is not so obvious, because I got a lot of unusable results, by trying to turn a partial image of a character into full length. In playgroundai you can also simulate outpainting by uploading a picture within a larger frame and marking the parts where something needs to be added. Based on my limited experience so far, I might go for a combined use of playgroundai and leonardo. For background removal I use adobe express, where it's free, fast and very accurate.
Hey everyone! Just a heads up there is a temporary hold on approving whitelisted registrations due to the heavy influx of users. Please checkout the community tab for details. Please be patient, it will be worth the wait!
Thanks for the demonstration! Which AI do you recommend for creating very simple coloring book pages for kids ages 2-4? Bc with Midjourney the coloring pages still turn out too detailed for toddlers
Hi Chris, honestly any AI generator can do it, it's a matter of prompting that style you are looking for. This one is a good place to start, Playground AI is also of good value.
Sure thing! There was so much to cover and I didn't want the video to be too long. I will definitely cover more on Leonardo AI. Also with the influx of new members, they site was slowing down, occasional hang ups here and there. Once it settles down it'll be easier for me to make content. Appreciate the feedback. Stay tuned!
Honestly I haven't thoroughly tested models for architecture and landscapes but I will, best to experiment. Start with standard Stable diffusion 1.5 and then try some of the custom models like RPG or Deliberate which have a photorealistic look. There are also community models you can try. 👍
I think they are already there, it's just a matter of people catching on. Although it's not perfect, they are way ahead than most! Super excited to see where they take this! Appreciate the feedback bud!
If someone can make a model that includes realistic hands and fingers, then it would become more useful than midjourney. I have attempted to make perhaps 100 portraits involving hands on midjourney but they always look mutated.
Stephen, that's what we all want!!! hahaha! I will say though, a lot of the recent models for stable diffusion are getting better with hands. Honestly, whomever can figure out how to get those consistent hands will be a legend! 😁
Hello, brackets are used for emphasis, it's a way to give more weight to a particular word. However it only works on certain platforms, like Automatic1111 for example. I just didn't bother deleting it when I used this prompt on the video as it has no effect in Leonardo AI at the moment. Hope that helps!
I'd like to see how it handles things which trip up AI image generators. Multiple things specified in the same image - "8 year old Asian girl cycling through a park and a female 8 year old Asian angel with pink wings stabbing her with pencils". Odd prompts - "statue of a smiling fish standing on his tail and tipping his wide brimmed hat." Night café refused to even try that second one and Playground insisted on including a human or second creature as part of the statue.
This is still stable diffusion, so it has the same limits that stable diffusion has. Stable diffusion is not good at making prompts with multiple different people in it, it mixes the tags up and will make two people with characteristics of the tags of both.
Could be a few factors, the prompt, negative prompt and mostly the model. Not sure if you are using the standard stable diffusion model? Use the platform models like Dreamshaper, RPG or even the custom leonardo ones. Lot's to choose from.
@@MonzonMedia I use prompt generator, Dreamshaper and other customs. I change up the customs as well as use other prompts from the generator and still the results looks terrible.
Yes my friend, you have to go to the discord and go to the introduction channel, wait 10min before you can post and then once you introduce yourself a bot will respond with a “priority early access” post. From there you just have to follow the directions which includes filling out a form. I know it’s a convoluted process but it will be worth your while. If you have any issues you can ask in the Discord. 👍🏼
Yes I agree, but this Leonardo is offering some unique features that playground isn't offering at the moment. I am especially interested in the custom model training. But if Playground adds this training option then it will continue to be my number one choice.
This is typical with trained models. If you are talking about "people" I often like to add ethnicity, or a combination of certain traits to change up the look.
its now 250 tokens that they changed in 1hr but they refill daily feb 17 it was unlimited then to 1000 token then changed again to 250 token in a matter of 3 hours so I hope they dont change again. seemed like CLICKBAIT because most of the people they got flooded with thought it was unlimited so we all joined the they did this so who know what will happen next .BUT IT IS A NICE APP
Yeah I know, but you have to understand that they got bombarded by 10 of thousands of new users so it's a way to control their server load. I don't think they were expecting that kind of growth so fast. Gotta love TH-cam hahaha! They did announce recently they are making adjustments to accommodate all the new users. I'm sure it'll settle down soon. They clarified that it's 500 tokens if using 512x512 and 40 steps or 250 tokens for 768x768 at 40 steps. It's still very reasonable but I do hope they keep the free option reasonable for most users. Personally I would pay for a pro membership to have full access since they offer way more than MJ or anyone else. For me it's the training capabilities and outpainting that is very valuable. Let's see how it goes! Appreciate the feedback bud!
Yep, tried it, blind girl can't get past the Discord Captcha Not very... intelligent.. use r interface design... artificially or otherwise. How hard is it for millenials to program a simple web interface where you upload photos of yourself, and it gneerates more based upon the provided photos. Wow. Just wow fail. I miss the real programmers from the 1960's 70's and 80's...
@@choppergirl Have you tried playgroundai.com? You just need to sign in with a google account and off you go! 1K images/day with some minor limitations.
Hey everyone! Since this was posted 5 months ago prior to the paid plans being announced, please be aware the free membership is now 150 tokens per day and 1 model can be trained and retained/mth.
Thanks for the great review!
Just wanted to add a few quick notes 🙂
1. The usual render time for 4 x 768x768 images is actually closer to 15 seconds, we had a huge influx of users yesterday due to a number of YT videos dropping.
2. We're working to improve outpainting, you should actually be able to use it with images generated with other models to extend images. Sometimes it helps to erase some of the edge before outpainting, it can do a better job that way. It sometimes also helps to simplify the prompt when extending the image. We also find the negative prompt to generally not be necessary when outpainting.
3. Unzooming is currently in beta and is meant to actually generate more of the underlying image, sort of like pulling the camera back. We're improving this too, there's just a lot happening in parallel 😂
4. The current free tier allowance is being adjusted to 500 tokens a day, and 1 token will soon equate to a standard 512x512 render. So 250 x 768x768 renders per day.
Thanks so much for taking the time to clarify. I'll mention these things on the next video. I'm really looking forward to see where Leonardo AI goes, very, very promising! 🙌👊
@@MonzonMedia hi could you please upscale the image too? Could you also tell us to what pixel resolution can the image be upscaled to? Thanks
@@musicwave8217 I believe it's x3 or x4 but I'll verify. The final pixel size depends on the size you started with.
@@MonzonMedia thank you so much 🙏
I love this
I'd love to hear your thoughts on Leonardo AI. Are you just as impressed as I was? I decided to cover the data set training separately to make this video shorter. Hoping to release it this weekend! Stay tuned. 👍
Cool to see you also took a look at Leonardo! It's a very fun tool to use:) had a lot of fun creating a video about it.
I'm really enjoying it, it has so much to offer. To think where it will be in a few months. I appreciate you stopping by and taking the time to leave feedback 🙌 I'll have to check out your video too!
Nice! I use Stable Diffusion on my home PC, so I recognize a lot of those features. Glad to see an app that is trying to keep up with all of the open source extensions and models. Everything is moving so fast.
Indeed Robert. I still feel that running it locally gives you the most flexibility however these web ui's are improving so fast making it more accessible to more people that may not have the gear to run it locally. Definitely a positive direction for all of us! What are you using? Automatic1111, Invoke AI?
I've been wasting tokens, trying to describe specifically what I want, and relying on image to image...this is a game changer for me. Far fewer wasted tokens. That's superb.
Dreamshaper 3.2 is my favorite
It’s a great model even though it’s a bit older 👍🏼
I used this for a week now and I'm really impressed ... This tools is really good 👍👍
Good to hear that! Yes great feature set available. 👍🏼
Loves your negative prompts... Too many fingers wkwkwkwkww... Loves this AI app and your walk through video. Many thanks
You’re very welcome! More to come soon 👍🏼
I’m in awe. THIS is what I’ve been needing
It's amazing isn't it! Exciting times for us indeed!
Fantastic! Thanks for the easy walk through
My pleasure! Be sure to check out the other 2 videos on Leonardo AI where I go more in depth on the other features
@@MonzonMedia will certainly do check them out
Hello good person…. I am loving your videos! Leonardo looks great
Hey Jim! Give it a go my friend...this is definitely one to watch out for. Thanks for supporting my channel, means a lot to me. 🙌👍
Very nice video
Thanks for checking it out! 🙌Be sure to check the others on Leonardo AI
This is a fantastic AI generator. Not sure if you're aware, but the seed option can, like MidJurney generate images tailored to the selected seed so the new images from the seed are generated similarly. One thing I would love for these AI generator platforms is offer a preview, so the image can be reviewed before the actual generation. THis way, there would be no wasted images, but you currently can't beat the pricing.
Hey Louis! I appreciate that thank you. Yes I'm aware of seeds and how they work but good to know that it's like MJ. I'm going to do some tests for "consistent characters" since this is something our community wants more info on. Hoping Leonardo AI can produce what we are all looking for.
A preview isn't really possible, as it has to generate the image to know what it is going to look like.
Another AI scam, it's main attraction is to the lazy, that's why these type of Pictraps are so enticing, No talent needed, It's not art by any definition, just another art scam.
@@davewills6121 Agreed, and it’s perfect for me. Not everyone is talented to be an artist. I have tremendous respect and admiration for those who are talented.
You never disappoint man, eagerly waiting for your testing models ....
Hello my friend, thanks as always for supporting! Means a lot! I'm messing around with a trained model today and will likely record in a few hours to hopefully post by the evening. Stay tuned! Have you signed up yet?
@@MonzonMedia yes i have signed in
@@davidbackham100 ok cool! I know they might start limiting access soon as this is still beta and with the recent influx, it won't be long before they limit the service. Paid option is on the horizon but if they are smart, they will still offer a decent free option.
@@MonzonMedia exactly , Sir
Great review, thanks. Could you please explain what those multiple parenthesis do in your prompt? Thanks!
You're very welcome! Those are for emphasis but it doesn't work on Leonardo AI just yet, I was copy and pasting a prompt that I commonly use when running SD locally. They did mention to me though that they will implement weighting eventually.
Would be good if you supplied the prompt you were using... I assume there are several entries in that prompt that would have been interesting for beginners.
Oh snap! I almost always do...let me add it to the description! 👍
This AI stuff is amazing. Like I’ve said before, totally over my head, but on discord or in your last video you were doing something that turned photos into comics. What was that? I had an idea for a sequence in one of my running videos. Hope you’re well, bro!
Hey bro great to see ya here! Was it the one on creating your own avatar? But yeah this stuff is moving fast and it's had to keep up! hahaha! Great to see you doing your thing on your channel man, keep going! or Running!
@@MonzonMedia That sounds right! I'll search for it. Thanks for the support...just posting for the fun of it at this point. I've realized I enjoy messing around with filming and editing, but no real passion for any type of niche. It changes depending on the day haha Have a great weekend!
Nice introduction to this new tool. Looks a bit like an upgraded version of playgroundai. I am currently waiting for the invitation.
Hey Dirk, I appreciate it. Yeah these guys have pretty much been paying attention to what users want. Feels like midjourney, with the ease of use of playground but leveled up! Going to be exciting to see how they improve it!
@@MonzonMedia I got my invitation yesterday. So far my impression is that playgroundai generates more quickly (and has more free generations). Upscaling is much slower in leonardo. Quality wise I didn't notice much difference when using the stable diffusion 1.5 model. I didn't see facial restoration in leonardo(?), which is a big thing to me. But with a finetuned model, you may not need that so much. Finetuned models generate better quality, but the results are very similar when using the same prompt. A big advantage for leonardo is that you can train a model. If you are productive then a limitation of 10/month may still be a downside. For training a character, you basically just need one full length picture (so far I tested this by uploading the same picture several times). And then you can get a fairly consistent character in different posing (back profile, side profile, sitting, lying down, smiling etc) which keeps the same type of clothes(!). The AI does tend to change the colors of the clothes, so I specify the color in the prompt. Outpainting is not so obvious, because I got a lot of unusable results, by trying to turn a partial image of a character into full length. In playgroundai you can also simulate outpainting by uploading a picture within a larger frame and marking the parts where something needs to be added. Based on my limited experience so far, I might go for a combined use of playgroundai and leonardo. For background removal I use adobe express, where it's free, fast and very accurate.
Hey everyone! Just a heads up there is a temporary hold on approving whitelisted registrations due to the heavy influx of users. Please checkout the community tab for details. Please be patient, it will be worth the wait!
That's so cool
It is indeed!
@@MonzonMedia Unfortunately I have to wait for my credits to be reset
How to create pose to pose image in leonardo ai with new update v2 and high contrast
Hi Sudheer, I just did a video on this the other day. Check it out here. th-cam.com/video/nasZTz3_Pq0/w-d-xo.html 👍
Thanks for the demonstration! Which AI do you recommend for creating very simple coloring book pages for kids ages 2-4? Bc with Midjourney the coloring pages still turn out too detailed for toddlers
Hi Chris, honestly any AI generator can do it, it's a matter of prompting that style you are looking for. This one is a good place to start, Playground AI is also of good value.
I want to see what you can do with the image to image option. I want to take a figure that I draw and do character design
Sure thing! There was so much to cover and I didn't want the video to be too long. I will definitely cover more on Leonardo AI. Also with the influx of new members, they site was slowing down, occasional hang ups here and there. Once it settles down it'll be easier for me to make content. Appreciate the feedback. Stay tuned!
Which models are best for architecture or landscapes?
Honestly I haven't thoroughly tested models for architecture and landscapes but I will, best to experiment. Start with standard Stable diffusion 1.5 and then try some of the custom models like RPG or Deliberate which have a photorealistic look. There are also community models you can try. 👍
more changes, and they will surpass midjourney for a mile! great app!
I think they are already there, it's just a matter of people catching on. Although it's not perfect, they are way ahead than most! Super excited to see where they take this! Appreciate the feedback bud!
I think Leonardo has potential. However you can't create variations of a specific image unless you have a secret trick
There is a "remix" option that does the same thing in your personal gallery. Would be nice to have it in the same image generation area though.
If someone can make a model that includes realistic hands and fingers, then it would become more useful than midjourney.
I have attempted to make perhaps 100 portraits involving hands on midjourney but they always look mutated.
Stephen, that's what we all want!!! hahaha! I will say though, a lot of the recent models for stable diffusion are getting better with hands. Honestly, whomever can figure out how to get those consistent hands will be a legend! 😁
Hi. Is it call Playground at the 06/2023?
uuummmm not sure what you mean, this is Leonardo AI. Are you thinking of Playground AI? Different platforms. 👍
Why are there brackets in the prompt?
Hello, brackets are used for emphasis, it's a way to give more weight to a particular word. However it only works on certain platforms, like Automatic1111 for example. I just didn't bother deleting it when I used this prompt on the video as it has no effect in Leonardo AI at the moment. Hope that helps!
wow the portraits of the women look identical to my trained and merged models results on the setup i have xD hahaha
It's pretty impressive and the fact you can train on this platform is huge!
@@MonzonMediathat's cool! i actually train and merge models on A1111web gui though but i will check that out then :)
I'd like to see how it handles things which trip up AI image generators. Multiple things specified in the same image - "8 year old Asian girl cycling through a park and a female 8 year old Asian angel with pink wings stabbing her with pencils". Odd prompts - "statue of a smiling fish standing on his tail and tipping his wide brimmed hat." Night café refused to even try that second one and Playground insisted on including a human or second creature as part of the statue.
I'm not sure if any public image generator would let that through.
This is still stable diffusion, so it has the same limits that stable diffusion has. Stable diffusion is not good at making prompts with multiple different people in it, it mixes the tags up and will make two people with characteristics of the tags of both.
I signed up for the early access, know how to use it?
I go over it in the video here th-cam.com/video/ii_fEY_e5nQ/w-d-xo.html
How did you get those quality images in one shot, I've been trying with simple letters and I got was trash images?
Could be a few factors, the prompt, negative prompt and mostly the model. Not sure if you are using the standard stable diffusion model? Use the platform models like Dreamshaper, RPG or even the custom leonardo ones. Lot's to choose from.
@@MonzonMedia I use prompt generator, Dreamshaper and other customs. I change up the customs as well as use other prompts from the generator and still the results looks terrible.
I wonder if the site is not open in Turkey, it sends it to the discord channel
Yes my friend, you have to go to the discord and go to the introduction channel, wait 10min before you can post and then once you introduce yourself a bot will respond with a “priority early access” post. From there you just have to follow the directions which includes filling out a form.
I know it’s a convoluted process but it will be worth your while.
If you have any issues you can ask in the Discord. 👍🏼
@@MonzonMedia
Can you make a video about it too?
Playground is easier . No forms etc. to fill.out.
True, but well worth signing up as there are more features to use. It's not very difficult. 👍
@@MonzonMedia What do you think about Lexica ?
Yes I agree, but this Leonardo is offering some unique features that playground isn't offering at the moment.
I am especially interested in the custom model training.
But if Playground adds this training option then it will continue to be my number one choice.
Hey there. It’s amazing platform. I do like it so much. I have one problem that I need help in promoting. Any one can help me pls
What kind of help? What's the issue?
It's cool but until you can go private I won't use it.
I'm sure they will h ave to implement it especially since they allow NSFW
@@MonzonMedia oh I didn't know that lol but yeah if they do it will be a must have
You can go private if ask one of the creators to give access to the feature!
@@Soulera805 oh really? Thanks for letting us know
Аll the drawings are the same in style , as if they were made by one artist . 😢🙂
This is typical with trained models. If you are talking about "people" I often like to add ethnicity, or a combination of certain traits to change up the look.
The result will be better if you add the names of different famous artists, mix them together..
its now 250 tokens that they changed in 1hr but they refill daily feb 17 it was unlimited then to 1000 token then changed again to 250 token in a matter of 3 hours so I hope they dont change again. seemed like CLICKBAIT because most of the people they got flooded with thought it was unlimited so we all joined the they did this so who know what will happen next .BUT IT IS A NICE APP
Yeah I know, but you have to understand that they got bombarded by 10 of thousands of new users so it's a way to control their server load. I don't think they were expecting that kind of growth so fast. Gotta love TH-cam hahaha! They did announce recently they are making adjustments to accommodate all the new users. I'm sure it'll settle down soon. They clarified that it's 500 tokens if using 512x512 and 40 steps or 250 tokens for 768x768 at 40 steps. It's still very reasonable but I do hope they keep the free option reasonable for most users. Personally I would pay for a pro membership to have full access since they offer way more than MJ or anyone else. For me it's the training capabilities and outpainting that is very valuable. Let's see how it goes! Appreciate the feedback bud!
but discord...
What about it?
Yep, tried it, blind girl can't get past the Discord Captcha
Not very... intelligent.. use r interface design... artificially or otherwise.
How hard is it for millenials to program a simple web interface where you upload photos of yourself, and it gneerates more based upon the provided photos. Wow. Just wow fail. I miss the real programmers from the 1960's 70's and 80's...
Sorry to hear you are having issues, you can always reach out to support. What are you currently using to generate AI images?
@@MonzonMedia Nothing, I can't get any of it to work.
@@choppergirl Have you tried playgroundai.com? You just need to sign in with a google account and off you go! 1K images/day with some minor limitations.