Consistent Characters are now much easier to get in Leonardo Ai with their new character reference feature! I made a comprehensive tutorial on that here: th-cam.com/video/7ordFNVhmzo/w-d-xo.html
Hey, it's awesome to hear these vids have helped you! It is hard to stay up to date with all the new features & tools being released all the time. I'll try to make more guides like this though
I admire your teaching style, you make it easy to follow along and you are very through. When you can, please create a tutorial on hacks for achieving truly consistent characters with same clothing using Leonardo Ai. Thanks
Love your channel man!! I have learned a lot. I have been using LEO for a while, but thanks to you video I better understand it. Could you maybe do one that explains the settings on the bottom left? The Guidance Scale and Seed and Scheduler. I don't really completely understand those.
Hey, I'm glad this video helped! 1. Guidance Scale: In theory this trades off prompt adherence and variety. So the higher guidance scale you use the more closely the ai model will follow your prompt, but it will have less variety in the generated images. 2. Seed: This can be used to replicate results. So if you use the same exact seed for different image generations, as long as your prompt and other settings are the same, the ai model will generate the exact same image. This is useful for replication and debugging purposes. 3. Scheduler: This is kind of an advanced feature for diffusion models which is what Leonardo's Ai is. Diffusions models start from a random noisy image (kind of like TV static), and slowly remove the noisy pixels to generate nice looking pictures. The scheduler slightly alters the rate that the random noise is removed, which can produce some different results. I haven't tried them all out myself though. I think Alchemy will get you the best results overall, but it's worth trying out the different samplers.
Great video! Btw, would you like to make a video about "stable artisan" which Stable Diffusion in Discord, that seems very attractive. It would be great if you can share a professional tutorial.
Leonardo made great still images for my video but when motion was added, it could not keep the face of the main image character stable or even decent. What should one do to create better motion facial features? Thanks.
I haven't done that much testing with the motion tool in Leonardo Ai yet. Maybe you could try to turn down the motion slider? Or try a slightly different image for your character. I will look more into the motion tool in the future.
The concepts created in the images may be similar, but the actual results will not be the same. For photorealism, Midjourney produces higher quality than Leonardo.
Leonardo is more flexible and gives you more options for what you can do. Midjourney overall produces higher quality images and is a bit easier to use.
I will see what's out there 👍. It is hard to find good free upscalers though. Upscayl is an option you can try to run locally. th-cam.com/video/BRx9OiBOUWE/w-d-xo.html
Does anyone know if there is a similar posing feature or parameter i midjourney? It can use image references to get similar poses, but always takes the style from the ref and leaves the output image with artefacts that make it look grainy or otherwise pixelated
I'm also having trouble getting it to work correctly. For now, a workaround is to just directly use the character reference image url, instead of prefer option set inside the vary region box. --cref image_url
So giid tutorial with many details, thank you for sharing it ! Also I'm using Stylar now, hope you cake make some tutorial about it ,cuz it's free and have lots of interesting styles
Hey Tao, great video tutorial. Just keen to know which image generator tool you recommend people to learn first? I've just been using DALL-E via ChatGPT and learning the fundamentals of prompt engineering to craft the output. Not coming from a design or photography background the simplicity is what appeals with that process, coupled with not having an exact image in mind. I'd like to create stylised images of designer clothes, to compliment a book about streetwear brands, rather than having to get permission from the brands/photographers to use their actual photos. Do you have any recommendations for this the best way to do this? I've tried using the brand's actual images as style guides but the results were a bit whack. Ideally I'd like to keep the outfits and footwear as close to the original as possible.
DALL-E and ChatGPT is honestly pretty good. It relatively easy to use and the quality is decent. If you want photorealism Midjourney is probably you're best bet. Leonardo Ai will give you a bit more control over the images but the quality won't be exactly as high. If you're using Leonardo Ai and want to keep the styles as closely as possible, you could try using the Edge reference along with style reference to preserve as much of the original image shape as possible. I have an older video guide for Fashion with some different streetwear brands, in Midjourney here: th-cam.com/video/zg3opvRjm6A/w-d-xo.html It's made for version 5.2 but it would probably work fine for version 6.
![ILLSLICK - WINTER IS COMING [Official Video]](http://i.ytimg.com/vi/dwTMeM1zIhQ/mqdefault.jpg)
Consistent Characters are now much easier to get in Leonardo Ai with their new character reference feature! I made a comprehensive tutorial on that here: th-cam.com/video/7ordFNVhmzo/w-d-xo.html
You really are a great teacher. You have no idea how much you are changing my life. Thank you loads
Hey, it's awesome to hear these vids have helped you! It is hard to stay up to date with all the new features & tools being released all the time. I'll try to make more guides like this though
Another impressive video, well done! The difference with your video: You are not in a hurry, and save time by being clear and straight to the point.
Thank you! I try to take my time and explain things well. I appreciate that you like these tutorials.
Thankyou for heeding to my request on Leonardo Tutorials. Great tutorial. Thanks once again 🙏
You're welcome! I'm working on another Leonardo Tutorial right now 👍
@@taoprompts Great 👍
Wow, wow, the control net and style images are awesome!
Yeah, it's a really flexible tool. They've also added a new character reference tool I'm making a tutorial on.
thank you Tao. Leonardo Ai is consistently improving.
Absolutely, it's so much better than 6 months ago.
Great teacher with greatest skills and manners God bless you ❤...
Thank you for the kind words 🙏
Fantastic tips, finally got me playing with Leonardo again 🎉
It's a great software and definitely has it's place in anyone's toolkit.
I admire your teaching style, you make it easy to follow along and you are very through. When you can, please create a tutorial on hacks for achieving truly consistent characters with same clothing using Leonardo Ai. Thanks
Thank you 👍. I do plan on looking into consistent characters with Leonardo Ai, it seems pretty involved but it will work on your own images.
@@taoprompts Great! Looking forward to it.
Love your channel man!! I have learned a lot. I have been using LEO for a while, but thanks to you video I better understand it. Could you maybe do one that explains the settings on the bottom left? The Guidance Scale and Seed and Scheduler. I don't really completely understand those.
Hey, I'm glad this video helped!
1. Guidance Scale: In theory this trades off prompt adherence and variety. So the higher guidance scale you use the more closely the ai model will follow your prompt, but it will have less variety in the generated images.
2. Seed: This can be used to replicate results. So if you use the same exact seed for different image generations, as long as your prompt and other settings are the same, the ai model will generate the exact same image. This is useful for replication and debugging purposes.
3. Scheduler: This is kind of an advanced feature for diffusion models which is what Leonardo's Ai is. Diffusions models start from a random noisy image (kind of like TV static), and slowly remove the noisy pixels to generate nice looking pictures. The scheduler slightly alters the rate that the random noise is removed, which can produce some different results. I haven't tried them all out myself though.
I think Alchemy will get you the best results overall, but it's worth trying out the different samplers.
Leonardo Ai has made some huge updates recently, it's one of the best Ai Image generation tools!
Its not free if we wanna use photo real we need subscription
Great video! Looking forward to more videos on Leonardo.
I will keep up updated 👍. There's a lot that can be done with this tool.
Love your teaching style, could you teach stylar? another ai tool
Thank you. I will take a look at it and see what it can do.
Great video!
Btw, would you like to make a video about "stable artisan" which Stable Diffusion in Discord, that seems very attractive.
It would be great if you can share a professional tutorial.
I didn't know that stable diffusion had a discord, I will have to take a look.
Leonardo made great still images for my video but when motion was added, it could not keep the face of the main image character stable or even decent. What should one do to create better motion facial features? Thanks.
I haven't done that much testing with the motion tool in Leonardo Ai yet. Maybe you could try to turn down the motion slider? Or try a slightly different image for your character.
I will look more into the motion tool in the future.
@@taoprompts This stuff is still in its infancy, it won't be perfect for awhile.
Question: Can the prompts from Midjourney be used in Leonardo AI or it won't be as good as in Midjourney.
The concepts created in the images may be similar, but the actual results will not be the same. For photorealism, Midjourney produces higher quality than Leonardo.
@@taoprompts I see, ty.
how to generate consistent image & create group of character in leonardo
Do you mean consistent characters/ multiple consistent characters? I will try it out and see what's possible 👍
@@taoprompts yes consistent multiple characters, thank in advice
@@taoprompts Do you have a a place where we can all meet up to discuss Leonardo?
@@innervisionjj I don't have a discord or anything like that, but Leonardo Ai does have a discord channel that anybody can join.
What’s best Leonardo or mid journey?
Leonardo is more flexible and gives you more options for what you can do. Midjourney overall produces higher quality images and is a bit easier to use.
Please teach character mapping in different scenes using Leonardo ai
I have a guide for Ai comics showing different character scenes in Leonardo here: th-cam.com/video/rIsus4VLVTw/w-d-xo.html
@@taoprompts thanks ❤️
Hello Tao. We are waiting for your videos on upscaling images for free. ❤
I will see what's out there 👍. It is hard to find good free upscalers though. Upscayl is an option you can try to run locally. th-cam.com/video/BRx9OiBOUWE/w-d-xo.html
Does anyone know if there is a similar posing feature or parameter i midjourney? It can use image references to get similar poses, but always takes the style from the ref and leaves the output image with artefacts that make it look grainy or otherwise pixelated
I don't think that's possible in Midjourney at the moment. It would require controlnet to be added.
@@taoprompts They got to get on that ASAP
Would you please help me midjourney's very region tool doesn't working i have to design consistent characters please
I'm also having trouble getting it to work correctly.
For now, a workaround is to just directly use the character reference image url, instead of prefer option set inside the vary region box.
--cref image_url
Its also not working @@taoprompts
@@taopromptsPlease help me😢
@@taopromptsVery region isn't working i have to make designs im facing difficulty😢
@@taopromptsI have to make ordr pleaseee response 😢
Like u said Leonardo is more friendly than MJ so why not making more tutorials video about Leonardo, I just purchased tokens for a month to try on…
There are many possible topics for me to cover, I try to pick the ones that are most interesting.
Why do Leonardo AI generated images still look cartoonish and or airbrushed clip art?
Try using the style reference with a more "cinematic" photo. Also if you use the PhotoReal feature the images look pretty realistic to me.
Please next new photoshop beta
I will take a look and see what's up
good for portrait photo pictures, other than that meh
The image guidance with pose, depth, style control is a really useful tool, for getting the exact composition you want in a photo.
Leonardo AI is better and better - improves faster than Midjourney.
They are adding many more features and experiment a lot more
So giid tutorial with many details, thank you for sharing it ! Also I'm using Stylar now, hope you cake make some tutorial about it ,cuz it's free and have lots of interesting styles
I'm glad you enjoyed the tutorial 👍. Thanks for the recommendation.
Hey Tao, great video tutorial. Just keen to know which image generator tool you recommend people to learn first?
I've just been using DALL-E via ChatGPT and learning the fundamentals of prompt engineering to craft the output.
Not coming from a design or photography background the simplicity is what appeals with that process, coupled with not having an exact image in mind.
I'd like to create stylised images of designer clothes, to compliment a book about streetwear brands, rather than having to get permission from the brands/photographers to use their actual photos.
Do you have any recommendations for this the best way to do this?
I've tried using the brand's actual images as style guides but the results were a bit whack.
Ideally I'd like to keep the outfits and footwear as close to the original as possible.
DALL-E and ChatGPT is honestly pretty good. It relatively easy to use and the quality is decent.
If you want photorealism Midjourney is probably you're best bet. Leonardo Ai will give you a bit more control over the images but the quality won't be exactly as high.
If you're using Leonardo Ai and want to keep the styles as closely as possible, you could try using the Edge reference along with style reference to preserve as much of the original image shape as possible.
I have an older video guide for Fashion with some different streetwear brands, in Midjourney here: th-cam.com/video/zg3opvRjm6A/w-d-xo.html
It's made for version 5.2 but it would probably work fine for version 6.