Nice. I am regular user of Leonardo AI. First time to see this tutorial with so in-depth illustration among the style and charactor references. Still a lot to learn.
Thanks for the LeonardoAI vids. I find starting with photo real for initial generation and then switching to Lighting XL and character ref works really well.
This was fantastic work. I went ahead and purchased the Leonardo AI prompt database, because Leonardo is the main AI generator I use and based on how much it's advancing it just makes sense. Another thing I noticed while you were generating images was how each preset has a different preset style based on the preset you pick. The anime preset style drop down options had me 🤯. This has been a big help with the different projects I'm pursuing. Thank you.
Thanks for the video :) I did several tests as soon as the update was done. It doesn't work with all the images. I tested with my AI influence generated on Fooocus. An 18-year-old blonde American girl. I used the reference character High and tested several styles and models. On all the images generated by Leonardo, on none of them, the girl looks like my reference image. Even by putting in the prompt the information of the characteristics of my AI influencer. On one image, she becomes an Eskimo, on another, she is a 35-year-old woman, on another, she is black, etc... I did about twenty tests, and none of them look like the reference image. And my reference character was always High and I didn't use any other reference image, always only the character reference. So, it really depends on the people in the photo. Even if they are people generated with AI.
My favorites are AlbedoBase XL, Diffusion XL or Kino XL for models + 3D Render for preset. Then to add some more fun, I like to use these elements: 3D Sculpt, Cute Emotes, Kids Illustration, Glowwave and Toon & Anime.
Wat sup! 😎 1 time long time... Meaning this is my first time making a comment, but I'm a long-time subscriber to your channel-love it, keep grinding! 🙌 I'm looking to get the most bang for my buck with the Triple Prompts Bundle. Would this be the bundle you’d get the most for your buck? If not, please let me know. Quick question: I'm pretty familiar with Leonardo AI using legacy mode. Great video, by the way-got the consistent characters down! 🎥 But I'm having a hard time finding pose-to-image references I can use because I'm doing storyboards. If you have any advice or prompts I might be able to use or where I can find pose-to-image reference pictures, I would greatly appreciate it. Thanks, and keep up the hard work, my brother! 💪
I hear you, but it's well worth the price. For everything you get, there isn't a better deal. IMO, the best paid apps are Leonardo AI, Freepik and Canva.
I have a question about character reference. I'm using a close-up that I generated myself. But now when I want to generate a half-body portrait in my prompts, I only get close-ups. Is there a way to generate half-body images? Best regards
Yes, use another reference image, but this time add pose to image as well. For your pose to image photo, choose an image that has the half-body style you want. Then you many have to adjust the strengths of each image guidance to narrow down how you want it.
So what would happen if you did all of this with a trained model? Say you did tbis with Scarlett and used a model tbat was trained using her images? Would help with consistently
You definitely need to make sure your aspect ratio (image dimensions) can fit a full body shot, like 2:3. Also, put "full body shot" at the very start of your prompt. Describe what the shoes will look like too. That will force the AI to include that, which will lead to a full body shot. If all else fails, and probably the best way is to use Pose to Image (as an image reference).
Thanks for the info, would you possibly know why I’m having this issue? I have a paid subscription, and I’ve tried using the character reference with a very real looking persons face. No matter what options or models I choose full quality ect, the images it creates look like an almost cartoon version of the initial image. I cannot get a realistic looking face image with the Character reference to come out of it? Is there a setting I need to turn on or off for realistic non cartoony smooth skin images? Any help would be appreciated. 😊✌🏼
This is definitely achievable. Do you have my Leonardo prompts database? If so, just follow some of those prompts and presets. If not, it really depends on the combination of models and presets being used. Make sure PhotoReal is turned on, try Leonardo Kino XL, then try cinematic and portrait as the preset. Use some of these keywords and artists/director's names: Documentary style, cinematic, Michael Subotozky, Peter Lik, Andres Serrano, and Meryl McMaster. When you use that combo with some of those names, the results are incredibly lifelike.
@@metricsmule Thanks for replying, I do appreciate you taking the time. I have been enhancing prompts with gpt, and do not have your prompts! Gpt does a decent job overall but obviously I’m doing something wrong. But as for Leonardo creations, I’ve tried every model and setting with the reference image of my 3 girls, and an image for the poses. It always puts one of the faces onto all three of them when it creates new images. I’m using lightning (I think that’s what it’s called) because the others seem to make them look very artificial. How much are your prompts and would they help fix this for me? 😊
Sure. 👍 I plan on making one on how you can create an actual comic book with consistent characters and speech bubbles. It will be all anime style. I'm still working on the final details and best prompts to use, but once testing is done I'll start the process
The information provided is usually worth the effort to watch but the fact that in every video the rush to complete it in the shortest possible time is palpable. At least you speak English and clearly so I can get a transcript to follow.
great video but please quit the strobe flashing effect in your video transitions. Its hard on the eyes and very unpleasant. your info is excellent. thats good enough.
☝️What @leonardoai said. They're good at making things right, so if you haven't, they'll credit you with 1k tokens (if you haven't already been credited).
Nice. I am regular user of Leonardo AI. First time to see this tutorial with so in-depth illustration among the style and charactor references. Still a lot to learn.
Glad it was helpful!
Thanks for the LeonardoAI vids. I find starting with photo real for initial generation and then switching to Lighting XL and character ref works really well.
Great tip!
Maybe do videos about niche ideas. Example would be horror graphic book, fantasy book, children book images and etc.
This was fantastic work. I went ahead and purchased the Leonardo AI prompt database, because Leonardo is the main AI generator I use and based on how much it's advancing it just makes sense. Another thing I noticed while you were generating images was how each preset has a different preset style based on the preset you pick. The anime preset style drop down options had me 🤯. This has been a big help with the different projects I'm pursuing. Thank you.
Thank you, glad it could help!
I like how well you explain Character Reference. Cheers!
Thanks very much, cheers to you too!
Thanks for the video :)
I did several tests as soon as the update was done.
It doesn't work with all the images.
I tested with my AI influence generated on Fooocus.
An 18-year-old blonde American girl. I used the reference character High and tested several styles and models. On all the images generated by Leonardo, on none of them, the girl looks like my reference image. Even by putting in the prompt the information of the characteristics of my AI influencer.
On one image, she becomes an Eskimo, on another, she is a 35-year-old woman, on another, she is black, etc... I did about twenty tests, and none of them look like the reference image.
And my reference character was always High and I didn't use any other reference image, always only the character reference.
So, it really depends on the people in the photo. Even if they are people generated with AI.
Best teacher ever. Thank you so much.
Thank you! 😃
What is the best preset model and preset style for 3D animation character consistency? (Not Anime). Thanks for the great content!
My favorites are AlbedoBase XL, Diffusion XL or Kino XL for models + 3D Render for preset. Then to add some more fun, I like to use these elements: 3D Sculpt, Cute Emotes, Kids Illustration, Glowwave and Toon & Anime.
This video is seriously helpful and awesome. 🎉
Glad you think so!
Wat sup! 😎
1 time long time... Meaning this is my first time making a comment, but I'm a long-time subscriber to your channel-love it, keep grinding! 🙌
I'm looking to get the most bang for my buck with the Triple Prompts Bundle. Would this be the bundle you’d get the most for your buck? If not, please let me know. Quick question: I'm pretty familiar with Leonardo AI using legacy mode. Great video, by the way-got the consistent characters down! 🎥 But I'm having a hard time finding pose-to-image references I can use because I'm doing storyboards. If you have any advice or prompts I might be able to use or where I can find pose-to-image reference pictures, I would greatly appreciate it.
Thanks, and keep up the hard work, my brother! 💪
Thanks for your detailed explanations. Now the options "character reference" is not available for Leonardo free version. a pity.
I hear you, but it's well worth the price. For everything you get, there isn't a better deal. IMO, the best paid apps are Leonardo AI, Freepik and Canva.
@@metricsmule thanks for answering!
Do you do any prompts with text based designs? If they are on the airtable I can't find them?
I have a question about character reference. I'm using a close-up that I generated myself.
But now when I want to generate a half-body portrait in my prompts, I only get close-ups.
Is there a way to generate half-body images?
Best regards
Yes, use another reference image, but this time add pose to image as well. For your pose to image photo, choose an image that has the half-body style you want. Then you many have to adjust the strengths of each image guidance to narrow down how you want it.
So what would happen if you did all of this with a trained model? Say you did tbis with Scarlett and used a model tbat was trained using her images? Would help with consistently
For sure. I think I'll try that. Good mention.
@@metricsmule Can I ask if you know a trick to get full body portraits?
You definitely need to make sure your aspect ratio (image dimensions) can fit a full body shot, like 2:3. Also, put "full body shot" at the very start of your prompt. Describe what the shoes will look like too. That will force the AI to include that, which will lead to a full body shot. If all else fails, and probably the best way is to use Pose to Image (as an image reference).
I have been searching for a character consistency AI platform. Hope you are right. Will check it out
Thanks for another mind bending vid 😎
Best update ever!
Thank you!
Awesome content buddy. Is this how people superimpose certain characters on…Pepsi cans, Doritos bags, etc?
It can definitely be done.
Thanks for the info, would you possibly know why I’m having this issue? I have a paid subscription, and I’ve tried using the character reference with a very real looking persons face. No matter what options or models I choose full quality ect, the images it creates look like an almost cartoon version of the initial image. I cannot get a realistic looking face image with the Character reference to come out of it? Is there a setting I need to turn on or off for realistic non cartoony smooth skin images? Any help would be appreciated. 😊✌🏼
It doesn’t matter now, I’ve found a few sites that turn these leonardo plastic looking skin images into authentic looking ones. 👌🏼😊
This is definitely achievable. Do you have my Leonardo prompts database? If so, just follow some of those prompts and presets. If not, it really depends on the combination of models and presets being used. Make sure PhotoReal is turned on, try Leonardo Kino XL, then try cinematic and portrait as the preset. Use some of these keywords and artists/director's names: Documentary style, cinematic, Michael Subotozky, Peter Lik, Andres Serrano, and Meryl McMaster. When you use that combo with some of those names, the results are incredibly lifelike.
@@metricsmule Thanks for replying, I do appreciate you taking the time. I have been enhancing prompts with gpt, and do not have your prompts! Gpt does a decent job overall but obviously I’m doing something wrong. But as for Leonardo creations, I’ve tried every model and setting with the reference image of my 3 girls, and an image for the poses. It always puts one of the faces onto all three of them when it creates new images. I’m using lightning (I think that’s what it’s called) because the others seem to make them look very artificial. How much are your prompts and would they help fix this for me? 😊
Can you make a video using anime style characters?
Sure. 👍 I plan on making one on how you can create an actual comic book with consistent characters and speech bubbles. It will be all anime style. I'm still working on the final details and best prompts to use, but once testing is done I'll start the process
You did great job thanks but please bro if possible make it quickly when Leo users have paid features free like from 24 may to 30.❤❤❤🎉🎉🎉
Thank you!
The information provided is usually worth the effort to watch but the fact that in every video the rush to complete it in the shortest possible time is palpable. At least you speak English and clearly so I can get a transcript to follow.
That works.
You dont look that bad as a woman :) great tutorial vids btw. Saves me a load of trial and error.
Lol thanks!
Ya, pero usando funciones de pago. Si no,no puedes usar referencia de personaje.
great video but please quit the strobe flashing effect in your video transitions. Its hard on the eyes and very unpleasant. your info is excellent. thats good enough.
More Free pik videos please
Will do. I have another I'm working on for a new feature that they currently have in beta. You'll love the new release.
@@metricsmule waiting for it. When are you releasing this video?
The new version takes much more tokens. Thats not so good
Hey Michael - there was a token pricing issue which has been fixed, and all paid users were credited 1,000 tokens.
@@leonardoai no, I didn't received any additional tokens. And I am on the artisan paid plan
☝️What @leonardoai said. They're good at making things right, so if you haven't, they'll credit you with 1k tokens (if you haven't already been credited).
these are pro features .. its not free