Awesome! I was looking for something like this for moths! I'm using Leonardo since last year, but never tried all the stuff it offers. I always tough that "train models" was something to do with the style of the image, not with a character, so I never go there and try it :) Thank you mate, you made my day brighter!
Thanks a lot! I'm glad you found the tutorial helpful. There are many ways to go about consistent characters, but these two methods got me the best results in Leonardo.ai. Best of luck to you and have fun!
Outstanding tutorial, so comprehensive and yet simple to follow with nothing left out. The teaching pace was perfect and you are to be commended on your style. A rarity in the field of free online tutorials. Thank you Ben!
You're very welcome! That's so nice of you to say. Means a lot and I really appreciate the kind words. I put a lot of work into these tutorials and I'm glad to hear that you getting some value out of them. Thanks again and best of luck to you!
A great video, I followed you every step of the way with every prompt you did. It was really good and very informative regarding the very final finish with her very blue eyes. I did the alchemy upscaling and it was perfect. I wish I had a photo to show you try it it’s really worked out well. I took away the blue eyes and it was brilliant. Anyway you have a new subscriber thank you
You've explained how to create consistent characters so well and clearly, thank you for the valuable tips! I’m curious about your thoughts on making the locations just as consistent. For example, ensuring that the character's background and the environment stay the same, but are shown from different angles throughout the story. How can this be achieved, so that the setting also aligns with the narrative and characters, while being presented from various perspectives? Would you consider making a video about this?
Thank you so much for the kind words and thoughtful suggestion! That’s an excellent idea for a future video. Creating consistent environments with varying perspectives can definitely elevate storytelling-I’ll explore ways to tackle this and share practical tips soon. Stay tuned!
This is a great tutorial and probably the best one that I’ve come across online that really explained how to do this properly, also I think the eyes were extra blue because it was still left in the prompt, despite using the model which the eyes were already defined, so it may have enhanced the blue in her eyessimply because it was repeated. At least, that’s my assumption.
Hi Ben. This tutorial is amazing, thanks. How can you have multiple characters that are created using seed characters in the one setting? If you can point me in the right direction it would be greatly appreciated.
Hi! Thanks so much! I really appreciate it. I haven't tried to create multiple characters, but you can try the following: 1. Set a seed for the initial character design. 2. Adjust prompts to specifically state that there are multiple characters while maintaining consistency. I hope that helps. I'm not sure we are there yet, but try that and let me know.
Wow, I'm really impressed with the photos, especially the last photos that we see in the video, but on this one we don't see a prompt, wouldn't you have a place where you store your prompts to share them with us so that we can have some ideas 😮
Hi! Thanks so much! For this "model" I used the following prompt: female instagram model, short shoulder length white hair, tan skin, blue eyes, athletic build, happy, wearing a pink miniskirt in the middle of a green pasture. I would just change what she was wearing and the scene she was in for other image generations. I hope that helps!
Thanks for this excellent and helpful video. I've been very frustrated with using AI to create consistent characters for my children's books. The scenario is the answer to my dreams!
Thanks for watching! Yes, I have an updated tutorial that you can watch here: th-cam.com/video/gv0wm-_ajgM/w-d-xo.html I don't cover consistent characters, but it goes over the new interface. I hope that helps!
Hi! Thanks for watching! Try this. 1. Use the new image generation interface, not the legacy mode. 2. Preset: Leonardo Lightning 3. Preset Style: 3D Render 4. Contrast: Medium 5. Don't turn on the Advanced Setting "Photo Real" 6. Prompt: pixar character of a man in the woods I was able to get some good results with those settings and prompt. Good luck!
Thank you for this. It was so easy to follow-up. I am a free subscriber and could not use the trained model aspect but was able to get good results with the seed approach. Thank you
Awesome video, After going around in circles trying to find out what they are and how to use fixed seeds, I finally stumbled across your video On a side note, I sometimes make whiteboard animation videos and I've been looking for a way of generating the exact same style of artwork Maybe I've been looking in all the wrong places but I'm yet to find the right method. if you could point me in the direction of a decent video, that would be brilliant
Hey! Thanks so much! Have you tried the "style reference feature"? Here is a good breakdown on how to use it in Leonardo.AI. leonardo.ai/wiki/applying-style-reference-to-an-existing-image/ I hope that helps!
All characters looking directly camera. What should we do if we want to see the character looking into the distance, or at a book, or at a picture on the wall, or talking to someone?
@@blogwithben Thanks for your answer. Before watching your video, I tried similar prompts as you said but they didn't work. I was actually looking for a solution to this problem so I watched your video.
Thanks for watching! I have yet to test Midjourney's consistent character reference, but you could be right. I've just found Loenardo's to be easier to manipulate. I need to test Midjourney though. I'll do that this weekend! Great suggestion!
Thanks for this however I noticed my model looks different in full body poses. It's like the facial features deviate away from the original trained model the further the face is away from the "camera"
Hi! Thanks so much for watching! I've found that there is some variation in the body types as well. I have faith that it will improve over time, but for now it's kind of test and experiment as you go. Thanks again for watching!
how to create consistent characters on multiple characters, when using leonardo ai for multiple characters to show consistent we are getting wrong output. Is there a way to have two or more character who are having multiple poses
That is a great question and I haven't tested out 2 characters yet. You could try writing extremely descriptive prompts for both characters and naming them. Then use that seed image and use their names in the future prompts to see if they carry over to the new prompt.
I have only seen this one video of yours and learned so much. Have you discussed the upgrade feature? It takes 30 credits per image, and the results are amazing.
Thanks for watching! I have not created a new yet, but I plan to. LeonardoAI has come out with some really cool new features since this video and I plan to cover them in an updated tutorial soon. :)
I m totally new to Leonardo but boy am I glad I was introduced to it and also found your video. This is lightyears ahead of Firefly, Copilot image generator or even MJ (in terms of user friendliness). Two questions: after watching this vid - using the seed method produces MUCH better results. Why would you train a model? I thought it would be better but it was WAY worse. 2nd - Can u update this vid as Leonardo introduced the image guidance feature which apparently makes this whole process easier and better (?)
Hi! Thanks for watching! I just wanted to present both options, seeds and data sets, for the purpose of the tutorial. Additionally, I am working on updated videos. Coming soon! :)
Thanks. Great tutorial. For using characters in storytelling we need consistency in outfits as well. So, presumably, if you prompted her wearing a white T-shirt, denim jeans and white sneakers, you could generate enough images of her in that outfit to pretty much guarantee consistency of both face and outfit, right?
Yes, exactly. However, it won't be perfect. I've found that you have to really drill down your prompts to get the outfits consistent. There was always a little variation in the clothes when I tried that. Great question though. I think the tech is getting there and we will be able to do what you're needing very soon. Thanks so much for watching and using the tutorial. I really appreciate it!
Hi Ben! Great video, really useful. But I'm looking to tell a story with images. Somehow like storyboarding in a comicbook style. Whats could be the process? Should I train the dataset with comicbook style images or more like a human-like model?
Thanks for watching! Why not try both? You could generate two separate but consistent image styles by using the image generator and then train two models with those two separate datasets to see which one would work for your story boards. Do you think you could do that? Great idea though. I love the idea of using these characters for a storyboard!
great video! i want to use this tool to advertise clothing. is there a way to upload a specific t-shirt design or something and use it on the models I create?
Thanks for watching! Leonardo has an AI video generator included with the paid plans. This allows you to add movement to your images. leonardo.ai/ai-video-generator/
Thanks for watching! Yeah, a lot of companies do that with their subscriptions plans. They add features to entice you to purchase a subscription. However, I think it's worth it. :)
That's a great question. You would more than likely have to put the "type of product" they are holding in the prompt and then add the branding via Photoshop or some other graphic design tool. Thanks so much for watching!
My biggest issue with using Leonardo for character models which I have done for over a year.Is that if you change the actual generation model then it will lose the actual features for the character and adapt it to the style of that model. When Leonardo updated their photo real the diversion 2 it screwed up the output of all my characters.So I still use version 1 for them. I like the output better.Anyway after you run it through universal upscale post edit.
Question............... How can I get the motion pictures to have good movement? I noticed that when I create a motion picture it looks weird are deformed. Is there a way to get that to stop and get good motion pictures to create a moving story?
Hey! Thanks for watching! It's definitely not perfect yet. I think it will just take some time for the technology to catch up to our expectations. It will get there soon though. Just look at how far this tech has come in one year. Keep experimenting and trying different prompts. That's what helps me get better results.
It was great, thanks But how about if we want to create consistent family? For example, parents and three children and have them together or individually in different conditions and scenes, it is possible?
Thanks for watching! I've actually gotten that question quite a bit and I haven't had a chance to test it. I don't think it's possible at the moment, but I suppose you could try to create a consistent image with four main characters and use that seed as the base for future prompts. I don't think it's possible with the trained model though. At least not yet. :)
I think I'm getting the hang of creating a more consistent character, but can't get her to change her expression no matter what I prompt. Is this something I should worry about during character creation or is this done in model training?
Hi! Thanks for watching! I would create a dataset with images of your face. I tried it, but it wasn't perfect. Experiment and see what works best for you, but that's how I would go about it now. :)
Really great tutorial! Question ... If I am building a longer story about characters in a workplace setting, is it possible to use two or three trained models in one scene? Can I just use the models names and get a consistent look?
Hi! Thanks so much for watching! I've actually gotten that question quite a bit and I haven't had a chance to test it. I don't think it's possible at the moment, but I suppose you could try to create a consistent image with three main characters and use that seed image as the base for future prompts. I don't think it's possible with the trained model though. At least not yet. :)
Thank you for the tutorial. I was wondering - is it possible to train a model that is full length? What about clothing? Do I have to train that as well with my model? I'm not that into modern clothing, tho sometimes modern clothing is more appropriate - like for weddings or baby showers. I like fantasy clothes and armor - adventurer leather gear, warrior metall armors, veiled Goddess / God clothing, armors looking like game outfits (like Skyrim and so on). And a bit of Goth, Steampunk, sci-fi.
You're welcome! Yes, you can train full-length models and include clothing styles during the training process. For fantasy, Goth, Steampunk, and other unique styles, you can curate specific images to guide the model. Let me know if that helps!
What a great video!! I have a very important question. How do I create one image with two characters? For example, I am telling a story about a elderly man and a young traveler. First I generate the two characters walking on a path but I can't create a picture of them both talking to each other. Then I get other characters every single time. So my question is; can you generate two characters and make scenes with them? For example walking on a path, sitting on a bench. Thanks in advance I really love your content
Hmmmmm that's a really good question. I've never tried that, but it would be interesting to test. I would try to make them their own unique characters in the prompt and then use that seed for the origitnal image of both of the characters and see the new images make both characters consistent in the new image generations. Thanks for watching and good luck!
@@PersonalGrowthHubb @blogwithben I'm in a similar situation. A possible solution is to create each character (transparaent backgrounds), create a scene (no characters) and then use Canva layers etc to create the "whole" image. A lot of work, but at this stage it might work.
Nice. Habe es ausprobiert und ein Referenz Foto als Vorlage .trotz prompt würden nur Portrait Fotos generiert. Obwohl ganauer prompt war gegeben das sie in ganzen zeigen sollte .
Thanks for watching! That can be somewhat tricky, but it would rely on your prompt. Try to be as descriptive as possible when describing the clothing and keep it the exact same in every prompt.
Thanks for watching and great question. Yes, I would train the model separately for each character. I don't think we can create consistent characters in the same setting at the moment. Hopefully in the near future though!
Hey! Thanks for watching! The resolution for this aspect ratio is 2560x1440. TH-cam automatically adapts to the size of each video, and may add padding for optimal viewing for some video and device aspect ratios.
Thanks for your videos, can you tell me if in Leonardo it is possible to put clothes and accessories on my model by passing a cut-out photograph? I would tell him put on this shirt, put on these pants and pass them with a picture. I have to make a lookbook for a client and I would like to wear the brand's clothes and accessories. A thousand thanks
Hi! Thanks for watching! I'm not sure we are there yet in terms of the technology and the output of the images. That is a great idea for future ideation, but I don't think you can do that now. At least the results probably wouldn't be ideal at the moment.
How about consistent outfits? How about consistent hairstyles (and hair length)? How about consistent body types (“athletic build” seems a little too broad of a term to me that doesn’t consider things like height, bone structure & fat distribution)?
Hi! Those are all great ideas. I think you would need to be explicit in your prompts when it comes to those details. So, after you create your seed image, just add those specific details to the prompt to keep it consistent throughout,. At least that's how I'd go about it. Thanks for watcing!
I have a question: when i generate for example 8 Images. And take thad Seed. Is that Seed for all the 8 Images? It is possible to take the Seed from a chosen Image, only one?
I tried to do a Silkie chicken and it didnt train it at all. Each of the them just look like the same photos (30) I posted to do the training. But were blurry. Very disappointing to say the least.
That's a bummer. This could be due to insufficient image quality or variety in the dataset. For better results: 1. Use high-resolution images with diverse angles and lighting. 2. Ensure Leonardo AI's training settings are optimized for the model type. If the problem persists, let me know!
Hi! Thanks for watching! I've gotten that question a lot and I don't think we are there yet. Maybe next year! Keep testing and see if your prompts/seeds can get your desired result though. :)
Hi! Thanks for watching! I don't think that is possible at the moment. Great idea though. I'll have to play around and see if you can use two models together.
For me it world for 3-5 pictures tjen he mixes up stuff. I am really frustrated, u can see it in my videos. They are mixing everything up and then I see other videos and think wow that could be me but boo no chance because it’s not working
Hi! Thanks for watching! Leonardo has made some updated since your comment. I would try your prompts again and see how they turn out. Sometimes it takes experimenting with different words and the way you phrase them in your prompts. Keep at it! :)
is it possible to upload a pic of a woman's face (ie head & shoulders) and apply a tanned style to her without changing her facial features ? Thank you for excellent teaching.
Thanks for watching! You would probably need to create a data set of that particular woman to keep her characteristics consistent and then you could generate a prompt using that data set but make her tan in your prompt. Does that make sense?
@@blogwithben no i mean there already is and it's visible in your video (it's the second item in the list of categories), i was wondering why you didn't chose it (but given the answer i guess you missed it 😂)
I don't exactly agree with your definition of a seed. A seed is just a number used to create a noise pattern. So it's not actually an image itself. So if you create and image you like with that seed then you can use the exact same prompt along with other settings to recreate that image. Then you can change colors or other slight modifications and get slightly different results. I was confused when I first started because people keep saying "seed image". So in my mind I thought it was some random image of a bird or landscape or person.
Thanks for watching. I get what you're saying. Your seed image set in Leonardo AI uses a fixed set of numbers, but it then represent the seed from what all other image generations are based off of. That's how I interpret it and teach it. I hope that helps and thanks again for watching!
Thanks for watching! You're right, there are other options out there, but I like Leonardo.ai. This will only get better over time too! Thanks again for watching!
I know of a third way to create consistent images, would anyone like to know ? and if you go in to your personal feed you can download numerous pictures at once, so go to your personal feed and on the top right select "select images" , select your images, and to the top right, now click where it says download and it will download a folder with the images you selected..
Hi! Thanks for watching! The training model just let's you have a "set" model instead of having to use the seed value every time. However, I see what you're saying. :)
This is a tutorial for creating models on the old version of Leonardo, how about showing us how to do it on the new version. Oh, that's right... the new version doesn't do models.
Thanks for pointing that out! The new version of Leonardo focuses on different features, but I’ll make a note to create content highlighting its current capabilities. If there’s a specific feature you'd like me to cover, feel free to let me know!
Awesome! I was looking for something like this for moths! I'm using Leonardo since last year, but never tried all the stuff it offers. I always tough that "train models" was something to do with the style of the image, not with a character, so I never go there and try it :) Thank you mate, you made my day brighter!
Thanks a lot! I'm glad you found the tutorial helpful. There are many ways to go about consistent characters, but these two methods got me the best results in Leonardo.ai. Best of luck to you and have fun!
Thanks!
Wow! Thank you very much. I really appreciate that. :)
GOOD JOB BRO ! It's well explained, there isn't too many out there who can describe Leonardo features like this, with such dexterity. Thanks a lot !
Appreciate you watching! I'm glad you found it helpful!
Outstanding tutorial, so comprehensive and yet simple to follow with nothing left out. The teaching pace was perfect and you are to be commended on your style. A rarity in the field of free online tutorials. Thank you Ben!
You're very welcome! That's so nice of you to say. Means a lot and I really appreciate the kind words. I put a lot of work into these tutorials and I'm glad to hear that you getting some value out of them. Thanks again and best of luck to you!
Very specific tutorialThank you for sharing and also hope you can make a tutorial about Skylar, they have a lot of different style and functions
Yeaaaahhhh Ben!!! This is awesome!! Greetings from Spain
Awesome! Thank you! Greetings from California!
A great video, I followed you every step of the way with every prompt you did. It was really good and very informative regarding the very final finish with her very blue eyes. I did the alchemy upscaling and it was perfect. I wish I had a photo to show you try it it’s really worked out well. I took away the blue eyes and it was brilliant. Anyway you have a new subscriber thank you
That's so awesome to hear! It sounds like you definitely have the process down! Nice work and thanks for using the tutorial. I really appreciate it!
Thank you very much for this detailed and very well explained tutorial. Your video has actually made me want to try it myself.
That's great! Leonard AI is a lot of fun! Glad you enjoyed the tutorial and have fun!
You have a gift of teaching!
Glad you think so! I really appreciate that. :)
You've explained how to create consistent characters so well and clearly, thank you for the valuable tips! I’m curious about your thoughts on making the locations just as consistent. For example, ensuring that the character's background and the environment stay the same, but are shown from different angles throughout the story. How can this be achieved, so that the setting also aligns with the narrative and characters, while being presented from various perspectives? Would you consider making a video about this?
Thank you so much for the kind words and thoughtful suggestion! That’s an excellent idea for a future video. Creating consistent environments with varying perspectives can definitely elevate storytelling-I’ll explore ways to tackle this and share practical tips soon. Stay tuned!
@@blogwithben Thank you, and I’m looking forward to the tips!
This is a great tutorial and probably the best one that I’ve come across online that really explained how to do this properly, also I think the eyes were extra blue because it was still left in the prompt, despite using the model which the eyes were already defined, so it may have enhanced the blue in her eyessimply because it was repeated. At least, that’s my assumption.
Thanks so much! Yeah, you're probably right. I'll experiment with different prompts and see if that makes a difference. Great suggestion.
Very well presented. Bravo! Thumbs-up
Many thanks! I really appreciate it. Thanks for watching too!
Wow simply amazing tech, great instructional video on Leonardo A.I.😊
Glad you liked it! Thanks so much for watching!
Hi Ben. This tutorial is amazing, thanks. How can you have multiple characters that are created using seed characters in the one setting? If you can point me in the right direction it would be greatly appreciated.
Hi! Thanks so much! I really appreciate it. I haven't tried to create multiple characters, but you can try the following:
1. Set a seed for the initial character design.
2. Adjust prompts to specifically state that there are multiple characters while maintaining consistency.
I hope that helps. I'm not sure we are there yet, but try that and let me know.
Wow wow; a tutorial that actually teaches!
Spectacular job!
Glad you liked it! I really appreciate your kind words. Thanks so much for watching and using the tutorial.
@@blogwithben
Program seems to struggle with consistency wit animals....
Wow, I'm really impressed with the photos, especially the last photos that we see in the video, but on this one we don't see a prompt, wouldn't you have a place where you store your prompts to share them with us so that we can have some ideas 😮
Hi! Thanks so much! For this "model" I used the following prompt: female instagram model, short shoulder length white hair, tan skin, blue eyes, athletic build, happy, wearing a pink miniskirt in the middle of a green pasture.
I would just change what she was wearing and the scene she was in for other image generations. I hope that helps!
Thanks for this excellent and helpful video. I've been very frustrated with using AI to create consistent characters for my children's books. The scenario is the answer to my dreams!
Great to hear! I hope it works for you! Best of luck to you with your children's book! That's a great idea!
How well did this tutorial work for you after all this time?
Hi man, great video! Really useful! I wanted to ask if you have done an updated video now that there is the specific funtion to reference characters
Thanks for watching! Yes, I have an updated tutorial that you can watch here: th-cam.com/video/gv0wm-_ajgM/w-d-xo.html I don't cover consistent characters, but it goes over the new interface. I hope that helps!
Would like to see a video with photographs of yourself being tweaked kind of overwhelming for a beginner like me, thanks for sharing very cool ⭐🙏🔥👍
Great suggestion! Thanks for watching too!
Hello, can you use two models trained in the same image?
Thank you! great video and explanation, will be experimenting. What settings would you recommend for creating 3D Pixar style characters?
Hi! Thanks for watching! Try this.
1. Use the new image generation interface, not the legacy mode.
2. Preset: Leonardo Lightning
3. Preset Style: 3D Render
4. Contrast: Medium
5. Don't turn on the Advanced Setting "Photo Real"
6. Prompt: pixar character of a man in the woods
I was able to get some good results with those settings and prompt. Good luck!
Super tutorial loved your training style. I have downloaded the video for my training. Keep the good work going. All the best. Cheers
Thanks a ton! I really appreciate your comment. I'm glad you found the tutorial helpful! Thanks again!
Thank you for this. It was so easy to follow-up. I am a free subscriber and could not use the trained model aspect but was able to get good results with the seed approach. Thank you
Thanks for letting me know! I'm glad you were still able to get some good results. Thanks for watching. :)
Fantastic - the best explanations on the Net to date
Wow, thanks! Glad it was helpful! Thanks again for watching!
All what I need. Thanks for this clear tuto
Happy to help! Thanks so much for using the tutorial!
Brilliant video thanks I’m going to try this !
Thanks so much! I hope you enjoy using LeonardoAI as much as I do! :)
You are brilliant, thanks a lot! Subscribed
You are far too kind! Thanks so much for the sub. I really appreciate it.
Great, thanks for sharing.
You're very welcome! Thanks for watching!
Awesome video, After going around in circles trying to find out what they are and how to use fixed seeds, I finally stumbled across your video
On a side note, I sometimes make whiteboard animation videos and I've been looking for a way of generating the exact same style of artwork Maybe I've been looking in all the wrong places but I'm yet to find the right method. if you could point me in the direction of a decent video, that would be brilliant
Hey! Thanks so much! Have you tried the "style reference feature"? Here is a good breakdown on how to use it in Leonardo.AI. leonardo.ai/wiki/applying-style-reference-to-an-existing-image/ I hope that helps!
All characters looking directly camera. What should we do if we want to see the character looking into the distance, or at a book, or at a picture on the wall, or talking to someone?
Hi! Thanks for watching! I would add that text to the prompt. "Woman, looking away from camera, reading a book, etc." Hope that helps and good luck!
@@blogwithben Thanks for your answer. Before watching your video, I tried similar prompts as you said but they didn't work. I was actually looking for a solution to this problem so I watched your video.
what the use of model and training if need to stick to the prompt it seems midjourney are more stable in chracter reference
Thanks for watching! I have yet to test Midjourney's consistent character reference, but you could be right. I've just found Loenardo's to be easier to manipulate. I need to test Midjourney though. I'll do that this weekend! Great suggestion!
For me this is the best tute on the subject. Detailed yet concise and precise, all steps clearly demonstrated. Thanks Ben for sharing. 👍👍
You're very welcome! That's so nice of you to say! Thanks for watching and I hope you have fun with Leonardo.ai!
Very informative! Thank you so much mate
No problem 👍 You're very welcome. Thanks for watching!
Thank you Ben, for tutorial!
You're very welcome! Thanks so much for using the tutorial and watching. I really appreciate it.
Thank you for this tutorial ❤
You're very welcome! Hope it was helpful!
Thanks for this however I noticed my model looks different in full body poses. It's like the facial features deviate away from the original trained model the further the face is away from the "camera"
Hi! Thanks so much for watching! I've found that there is some variation in the body types as well. I have faith that it will improve over time, but for now it's kind of test and experiment as you go. Thanks again for watching!
how to create consistent characters on multiple characters, when using leonardo ai for multiple characters to show consistent we are getting wrong output. Is there a way to have two or more character who are having multiple poses
That is a great question and I haven't tested out 2 characters yet. You could try writing extremely descriptive prompts for both characters and naming them. Then use that seed image and use their names in the future prompts to see if they carry over to the new prompt.
Thank you for your wonderful education. By the way, which screenshot program do you use?
Hi! Thanks so much! I really appreciate it. I use ScreenFlow to record my screen.
I have only seen this one video of yours and learned so much. Have you discussed the upgrade feature? It takes 30 credits per image, and the results are amazing.
Thanks for watching! I have not created a new yet, but I plan to. LeonardoAI has come out with some really cool new features since this video and I plan to cover them in an updated tutorial soon. :)
using seed number method seems to create a better consistency model than using the trained model
Thanks for watching! I totally agree. :)
I m totally new to Leonardo but boy am I glad I was introduced to it and also found your video. This is lightyears ahead of Firefly, Copilot image generator or even MJ (in terms of user friendliness).
Two questions: after watching this vid - using the seed method produces MUCH better results. Why would you train a model? I thought it would be better but it was WAY worse.
2nd - Can u update this vid as Leonardo introduced the image guidance feature which apparently makes this whole process easier and better (?)
Hi! Thanks for watching! I just wanted to present both options, seeds and data sets, for the purpose of the tutorial. Additionally, I am working on updated videos. Coming soon! :)
Very helpful!
Glad you think so! Thanks for watching!
Thanks. Great tutorial. For using characters in storytelling we need consistency in outfits as well. So, presumably, if you prompted her wearing a white T-shirt, denim jeans and white sneakers, you could generate enough images of her in that outfit to pretty much guarantee consistency of both face and outfit, right?
Yes, exactly. However, it won't be perfect. I've found that you have to really drill down your prompts to get the outfits consistent. There was always a little variation in the clothes when I tried that. Great question though. I think the tech is getting there and we will be able to do what you're needing very soon. Thanks so much for watching and using the tutorial. I really appreciate it!
Hi Ben! Great video, really useful. But I'm looking to tell a story with images. Somehow like storyboarding in a comicbook style. Whats could be the process? Should I train the dataset with comicbook style images or more like a human-like model?
Thanks for watching! Why not try both? You could generate two separate but consistent image styles by using the image generator and then train two models with those two separate datasets to see which one would work for your story boards. Do you think you could do that? Great idea though. I love the idea of using these characters for a storyboard!
great video! i want to use this tool to advertise clothing. is there a way to upload a specific t-shirt design or something and use it on the models I create?
Hey! Thanks for watching! Not that I know of. I personally would create the models and then add the t-shirt design in Photoshop.
how can we add video (move ) to this model ?
Thanks for watching! Leonardo has an AI video generator included with the paid plans. This allows you to add movement to your images. leonardo.ai/ai-video-generator/
Great video. What I don’t like about Leonardo is that to delete any ai image it has to be the premium plan.
Thanks for watching! Yeah, a lot of companies do that with their subscriptions plans. They add features to entice you to purchase a subscription. However, I think it's worth it. :)
Change to legacy mode in upper right corner, you can delete your images without premium plan
How would you have the character using or holding a product you are selling or advertising?
That's a great question. You would more than likely have to put the "type of product" they are holding in the prompt and then add the branding via Photoshop or some other graphic design tool. Thanks so much for watching!
My biggest issue with using Leonardo for character models which I have done for over a year.Is that if you change the actual generation model then it will lose the actual features for the character and adapt it to the style of that model. When Leonardo updated their photo real the diversion 2 it screwed up the output of all my characters.So I still use version 1 for them. I like the output better.Anyway after you run it through universal upscale post edit.
Great point. If you find a version that works best for you then by all means, use that one. Thanks so much for watching!
Great video! Do you know if it is possible to achieve this consistent characters through their API as well?
Oh, great question! I'm not 100% sure, but I believe so. I'll have to do some research though. I haven't gone that route yet. Thanks for watching!
Question............... How can I get the motion pictures to have good movement? I noticed that when I create a motion picture it looks weird are deformed. Is there a way to get that to stop and get good motion pictures to create a moving story?
Hey! Thanks for watching! It's definitely not perfect yet. I think it will just take some time for the technology to catch up to our expectations. It will get there soon though. Just look at how far this tech has come in one year. Keep experimenting and trying different prompts. That's what helps me get better results.
Thanx man, I really learnt something...
You're very welcome! Thanks so much for watching!
The content you share is very good, thank you for sharing.
Thank you too!
It was great, thanks
But how about if we want to create consistent family? For example, parents and three children and have them together or individually in different conditions and scenes, it is possible?
Thanks for watching! I've actually gotten that question quite a bit and I haven't had a chance to test it. I don't think it's possible at the moment, but I suppose you could try to create a consistent image with four main characters and use that seed as the base for future prompts. I don't think it's possible with the trained model though. At least not yet. :)
@@blogwithben THANKS
I think I'm getting the hang of creating a more consistent character, but can't get her to change her expression no matter what I prompt. Is this something I should worry about during character creation or is this done in model training?
Great tutorial! I just want to ask, is there a way to use 2 or more custom characters in a scene?
Hey! Thanks for watching and great question. At the moment, I don't think so. However, that is a great test opportunity.
Can you please advise how i can use my face to generate myself using Leonardo website. Kind regards
Hi! Thanks for watching! I would create a dataset with images of your face. I tried it, but it wasn't perfect. Experiment and see what works best for you, but that's how I would go about it now. :)
Really great tutorial! Question ... If I am building a longer story about characters in a workplace setting, is it possible to use two or three trained models in one scene? Can I just use the models names and get a consistent look?
Hi! Thanks so much for watching! I've actually gotten that question quite a bit and I haven't had a chance to test it. I don't think it's possible at the moment, but I suppose you could try to create a consistent image with three main characters and use that seed image as the base for future prompts. I don't think it's possible with the trained model though. At least not yet. :)
Thank you for the tutorial.
I was wondering - is it possible to train a model that is full length? What about clothing? Do I have to train that as well with my model?
I'm not that into modern clothing, tho sometimes modern clothing is more appropriate - like for weddings or baby showers.
I like fantasy clothes and armor - adventurer leather gear, warrior metall armors, veiled Goddess / God clothing, armors looking like game outfits (like Skyrim and so on). And a bit of Goth, Steampunk, sci-fi.
You're welcome! Yes, you can train full-length models and include clothing styles during the training process. For fantasy, Goth, Steampunk, and other unique styles, you can curate specific images to guide the model. Let me know if that helps!
@@blogwithben This will be a very interesting challenge for me. Hopefully not too hard. 😅😊 Thank you!
What a great video!!
I have a very important question. How do I create one image with two characters? For example, I am telling a story about a elderly man and a young traveler.
First I generate the two characters walking on a path but I can't create a picture of them both talking to each other.
Then I get other characters every single time. So my question is; can you generate two characters and make scenes with them? For example walking on a path, sitting on a bench.
Thanks in advance I really love your content
Hmmmmm that's a really good question. I've never tried that, but it would be interesting to test. I would try to make them their own unique characters in the prompt and then use that seed for the origitnal image of both of the characters and see the new images make both characters consistent in the new image generations. Thanks for watching and good luck!
@@blogwithben I tried that already but is doesn't work in my opinion. Thanks for responding, I really appreciate it, just like your content!
@@PersonalGrowthHubb @blogwithben I'm in a similar situation. A possible solution is to create each character (transparaent backgrounds), create a scene (no characters) and then use Canva layers etc to create the "whole" image. A lot of work, but at this stage it might work.
hi, pls help, i couldn't edit model settings, after creating a model and I can't change the category
Nice. Habe es ausprobiert und ein Referenz Foto als Vorlage .trotz prompt würden nur Portrait Fotos generiert. Obwohl ganauer prompt war gegeben das sie in ganzen zeigen sollte .
Best video on Leonardo AI. THANKS mATE
No problem 👍 Thanks so much for watching!
How do you get the consistent characters wearing the same clothes for a book?
Thanks for watching! That can be somewhat tricky, but it would rely on your prompt. Try to be as descriptive as possible when describing the clothing and keep it the exact same in every prompt.
If i try to use it for my story characters, should i training model for each character separately?
Thanks for watching and great question. Yes, I would train the model separately for each character. I don't think we can create consistent characters in the same setting at the moment. Hopefully in the near future though!
What are the dimensions to use for 16 :9 you tube video
Hey! Thanks for watching! The resolution for this aspect ratio is 2560x1440. TH-cam automatically adapts to the size of each video, and may add padding for optimal viewing for some video and device aspect ratios.
Thanks for your videos, can you tell me if in Leonardo it is possible to put clothes and accessories on my model by passing a cut-out photograph? I would tell him put on this shirt, put on these pants and pass them with a picture. I have to make a lookbook for a client and I would like to wear the brand's clothes and accessories. A thousand thanks
Hi! Thanks for watching! I'm not sure we are there yet in terms of the technology and the output of the images. That is a great idea for future ideation, but I don't think you can do that now. At least the results probably wouldn't be ideal at the moment.
How about consistent outfits?
How about consistent hairstyles (and hair length)?
How about consistent body types (“athletic build” seems a little too broad of a term to me that doesn’t consider things like height, bone structure & fat distribution)?
Hi! Those are all great ideas. I think you would need to be explicit in your prompts when it comes to those details. So, after you create your seed image, just add those specific details to the prompt to keep it consistent throughout,. At least that's how I'd go about it. Thanks for watcing!
I have a question: when i generate for example 8 Images. And take thad Seed. Is that Seed for all the 8 Images? It is possible to take the Seed from a chosen Image, only one?
Hey! Thanks for watching. Great question. As far as I know, all we can do now is get a seed from a set of image generations.
@@blogwithben
Yes and that is not good
This is a great tutorial, however (at least in my case) training and data sets are available only under a paid subscription.
Hi! Sorry about that. Yeah, you'll need a paid plan to use all the features.
I tried to do a Silkie chicken and it didnt train it at all. Each of the them just look like the same photos (30) I posted to do the training. But were blurry. Very disappointing to say the least.
That's a bummer. This could be due to insufficient image quality or variety in the dataset. For better results:
1. Use high-resolution images with diverse angles and lighting.
2. Ensure Leonardo AI's training settings are optimized for the model type.
If the problem persists, let me know!
Hi Ben. How about two consistent characters in one scene. Can we do it using Leonardo ?
Hi! Thanks for watching! I've gotten that question a lot and I don't think we are there yet. Maybe next year! Keep testing and see if your prompts/seeds can get your desired result though. :)
Thank you Ben
You're very welcome, my friend!
thank you!
You're welcome! Thanks for using the tutorial! I really appreciate it!
Can i use 2 trained models together?
Hi! Thanks for watching! I don't think that is possible at the moment. Great idea though. I'll have to play around and see if you can use two models together.
I tried using the seed method but it didn't generate the same character.
Thanks for watching! Sometimes it takes a few tries to get the consistency you want. Experiment with different prompts and seed images. :)
i used the seed as u u said its not working
Great walk through, can't wait to watch your other videos
Awesome, thank you!
Ben, I am frustrated. I cant get full body characters. all i ever get is mugshots, how did you get almost full body like that?
Hi! Thanks or watching! I would check the image dimensions. Try using 2:3 for full body shots.
You can mass download if you go to personal feed and are a paid sub.
Really!?! I'm going to have to try that today! Thanks for the insight! I really appreciate it!
Very good dude....
Thanks a lot!
For me it world for 3-5 pictures tjen he mixes up stuff. I am really frustrated, u can see it in my videos. They are mixing everything up and then I see other videos and think wow that could be me but boo no chance because it’s not working
Hi! Thanks for watching! Leonardo has made some updated since your comment. I would try your prompts again and see how they turn out. Sometimes it takes experimenting with different words and the way you phrase them in your prompts. Keep at it! :)
Great ❤
Thanks so much!
is it possible to upload a pic of a woman's face (ie head & shoulders) and apply a tanned style to her without changing her facial features ? Thank you for excellent teaching.
Thanks for watching! You would probably need to create a data set of that particular woman to keep her characteristics consistent and then you could generate a prompt using that data set but make her tan in your prompt. Does that make sense?
@@blogwithben Yes thanks. God bless your channel.
Great tutorial, looks so easy, but doesn't work for me at all. Each picture I get new face, kinda annoying.
Sorry to hear that. Thanks for watching though. Maybe keep experimenting with different prompts.
Same with me. I am creating a children's book but generating inconsistent characters is wasting my time and coins 🥴🥴
Why not "Character" in the category of the fine-tuned model?
Thanks for watching! Not sure. That could be some good feedback for Leonardo to include in their next update. :)
@@blogwithben no i mean there already is and it's visible in your video (it's the second item in the list of categories), i was wondering why you didn't chose it (but given the answer i guess you missed it 😂)
Ahhhh I see now! Hahaha! Good eye. I'll try that next time!
in your tutorial when you create your own model it appear the flag " this model is NSFW " but when i do it doesn't appear
Hey! Thanks for watching. I'll have to revisit this, but it may be due to an update.
Wait. Did he say ctrl+p for paste!?
Hey! Thanks for watching! Hahaha! I did say Command P. Whoops! My bad!
I don't exactly agree with your definition of a seed.
A seed is just a number used to create a noise pattern. So it's not actually an image itself.
So if you create and image you like with that seed then you can use the exact same prompt along with other settings to recreate that image. Then you can change colors or other slight modifications and get slightly different results.
I was confused when I first started because people keep saying "seed image". So in my mind I thought it was some random image of a bird or landscape or person.
Thanks for watching. I get what you're saying. Your seed image set in Leonardo AI uses a fixed set of numbers, but it then represent the seed from what all other image generations are based off of. That's how I interpret it and teach it. I hope that helps and thanks again for watching!
This is good. Thanks
Thank you too! Appreciate it!
Still its good enough there are other apps that can do this like Sea art AI what we need is conisistent clothing and detailed designing
Thanks for watching! You're right, there are other options out there, but I like Leonardo.ai. This will only get better over time too! Thanks again for watching!
I know of a third way to create consistent images, would anyone like to know ? and if you go in to your personal feed you can download numerous pictures at once, so go to your personal feed and on the top right select "select images" , select your images, and to the top right, now click where it says download and it will download a folder with the images you selected..
Hey! Thanks for watching and great tip! I'm going to try that!
Good job
Thanks a lot!
thankyou for no bull just making us buy paid course or affiliate marketing fluffs
Thanks so much! I'm glad you appreciate that. :)
this is so fantastic! thank you.
Glad you like it! Thanks for watching!
Pretty helpful
Glad you think so! Thanks for watching and using the tutorial.
Great job but I don't understand how the training model is any more helpful then simply using a seed.
Hi! Thanks for watching! The training model just let's you have a "set" model instead of having to use the seed value every time. However, I see what you're saying. :)
I need non-human consistent characters
This is a tutorial for creating models on the old version of Leonardo, how about showing us how to do it on the new version. Oh, that's right... the new version doesn't do models.
Thanks for pointing that out! The new version of Leonardo focuses on different features, but I’ll make a note to create content highlighting its current capabilities. If there’s a specific feature you'd like me to cover, feel free to let me know!
Nice
Thanks