You can now use Segment Anything instead of having to select components of the image in PS, in order to select specific parts of an image to change them.
The Segment Anything model is great. But not precise enough yet. So again, I do what is fastest for me. In reality right now ai is still slower in many of the processes than a good illustrator or graphic designer. So I always try to work the fastest way. 🙌
thx for this amazing video, Could you pz make a Video for how to train a specific item like a custom Type of clothes, for example Kurdish / Moroccan Traditional clothes are not familiar with Ai tools, But i want to custom train it, is there a way to train clothes? would be amazing if you can make a video about it, Thx so much
@@LevendeStreg you know, bode based Comfy Ui. It's an alternative to Auto1111. Not many deep dive videos on it. Would love your fresh and instructive workflow, if you get into it
@@___x__x_r___xa__x_____f______ Hi again @rafpaff - thanks for pointing me in that direction. Nodebased UI are cool, I agree. I'll look into it. I've actually tried it out before (just didn't know what it was called). ON RodPod they have a Comfy UI set up already for you to just try out and use when you build a pod!
Damn, i really ,issed the boat with SD :( . Hey just took a look at that Adobe Color site, its brilliant!!! Looking forward to the MJ vid. Great Stuff.
If I wanted to simply change the color of something and describe using the text prompt, would that be difficult? Example: if I want a white piano to be changed into a pink piano (nothing changes except for the color of the piano)
Well, you can pull in the image into ControlNet and use the same seed (if you have it) and then prompt again. Is that what you mean? Keep the scale low if you don’t want to change very much.
Is there anything similar for Midjourney please I had this exact problem today as soon as I tried adding another character in the scene to whisper to the other character and everything went weird!
Thank you for your comment, @thelibrarian3734 - There isn't a similar way to do it in Midjourney. Sorry to say. But there's a very good chance the reason it all goes strange in Stable Diffusion is because your graphics card and lack of vRam. Using Multi ControlNet is super hard on the computer. It takes a really strong graphics card.
9:19 => Isn't it better just to use gradient maps to change the colors than taking it back to SD? Same thing with the first photo. Isn't just faster to just create a mask and change the hair color in photoshop? It seems like doing extra work for no reason.
Yes if you’re happy with the light and shadows. I probably should have explained that. But in the pictures I’ve combined - the shadows and lights don’t fit. His light source comes from beneath the canvas - her light source is from the top. So they don’t match. That’s why I pull them into Photoshop - this way the light source is changes to match…
thanks for the video, but the 2.5 minute, double intro is a little much for a 10 minute video. i know, everyone and their dog does double intros now, but it is soooo tedious.
Hahahahhaha... Thank you kindly for the feedback. It's much appreciated. I learn as I go. So I try to make each new video better than the last. This will help. Thanks!
The constant exploding thumbnail and moving hands are super distracting and add very little to the tutorial. This would be a great tutorial if it wasn't so heavily, "I am trying to be an influencer," and focused on, "I am actually trying to give you a good tutorial." Also the final image her forearm is as long as her entire torso, so this isn't actually that great a tutorial for making something that looks solid.
And I agree - her forearm is too long, but this was just to show the process - and not create a masterpiece. It takes an enormous amount of time to do these tutorials…🎉
Hi Maria, I have a question, can Stable diffusion AI add color to an existing comic lineart? For example i.pinimg.com/originals/72/c6/f4/72c6f433cd20a40b82a5116a25999466.jpg Thank you for your good video as always.
What a great question. And thank you for the praise, @Qjojotaro. Yes, it should be, but you need to add colors in an image sketch - and the use the input you showed me in ControlNet. I’ll try to do a tutorial on it and see how it works!🙌
You can now use Segment Anything instead of having to select components of the image in PS, in order to select specific parts of an image to change them.
The Segment Anything model is great. But not precise enough yet. So again, I do what is fastest for me. In reality right now ai is still slower in many of the processes than a good illustrator or graphic designer. So I always try to work the fastest way. 🙌
The best tutorial for SD I've ever seen. Please do more like these! Congratz
Oh thank you so much, it really means a lot to me. 🙌
Wow! What a workflow! ...BIG FANX! ...
BTW:
I very often put just one color in the neg. prompts to get more harmonic results....
Glad you like it and thanks for watching and commenting!
Các video của bạn thật sự rất rất tuyệt, nó là những video hay nhất mà tôi được xem về Stable Diffusion. Thank you so much!!
Oh thank you so much, I really appreciate it. It means a lot for me. Cảm ơn bạn rất nhiều🙏
SUBSCRIBED. thank you so much, this is some really great information!:)
thx for this amazing video, Could you pz make a Video for how to train a specific item like a custom Type of clothes, for example Kurdish / Moroccan Traditional clothes are not familiar with Ai tools, But i want to custom train it, is there a way to train clothes? would be amazing if you can make a video about it, Thx so much
Thank you @a9x for that great idea. I do have an upcoming video on how to train - and I will see if I can put it in there. Thank you🙌
@@LevendeStreg Thank you so so much 🙏
please do some Comfy Ui tuts? That's where it is at, isn't it?
Haha… What Do you mean with “comfy”? Not so high powered? 🕺
@@LevendeStreg you know, bode based Comfy Ui. It's an alternative to Auto1111. Not many deep dive videos on it. Would love your fresh and instructive workflow, if you get into it
@@___x__x_r___xa__x_____f______ Hi again @rafpaff - thanks for pointing me in that direction. Nodebased UI are cool, I agree. I'll look into it. I've actually tried it out before (just didn't know what it was called). ON RodPod they have a Comfy UI set up already for you to just try out and use when you build a pod!
@@LevendeStreg thanks! There's a bit of development in this space which I think is a good sign.
Damn, i really ,issed the boat with SD :( . Hey just took a look at that Adobe Color site, its brilliant!!!
Looking forward to the MJ vid.
Great Stuff.
Thank you so much - I really appreciate it. Glad you liked it. And it’s never too late to start with Stable Diffusion.🙌
If I wanted to simply change the color of something and describe using the text prompt, would that be difficult? Example: if I want a white piano to be changed into a pink piano (nothing changes except for the color of the piano)
Yeah, that's doable. Just like I show in the video🙌
If i lost the prompts, how I can do it only with the image?
Well, you can pull in the image into ControlNet and use the same seed (if you have it) and then prompt again. Is that what you mean? Keep the scale low if you don’t want to change very much.
Is there anything similar for Midjourney please I had this exact problem today as soon as I tried adding another character in the scene to whisper to the other character and everything went weird!
Thank you for your comment, @thelibrarian3734 - There isn't a similar way to do it in Midjourney. Sorry to say. But there's a very good chance the reason it all goes strange in Stable Diffusion is because your graphics card and lack of vRam. Using Multi ControlNet is super hard on the computer. It takes a really strong graphics card.
10 minute video, 7 minute of talking about yourself, 3 minutes of the actual relevant content xD
9:19 => Isn't it better just to use gradient maps to change the colors than taking it back to SD? Same thing with the first photo. Isn't just faster to just create a mask and change the hair color in photoshop? It seems like doing extra work for no reason.
Yes if you’re happy with the light and shadows. I probably should have explained that. But in the pictures I’ve combined - the shadows and lights don’t fit. His light source comes from beneath the canvas - her light source is from the top. So they don’t match. That’s why I pull them into Photoshop - this way the light source is changes to match…
@@LevendeStreg makes sense. good point about the light source.
👋
Hi Louis. Thank you for watching and commenting. It really helps. I appreciate it 🙌
thanks for the video, but the 2.5 minute, double intro is a little much for a 10 minute video. i know, everyone and their dog does double intros now, but it is soooo tedious.
ok, it's a good 5 minute tutorial that starts at about 4:30, but still has an embedded commercial planted about every minute
Hahahahhaha... Thank you kindly for the feedback. It's much appreciated. I learn as I go. So I try to make each new video better than the last. This will help. Thanks!
The constant exploding thumbnail and moving hands are super distracting and add very little to the tutorial. This would be a great tutorial if it wasn't so heavily, "I am trying to be an influencer," and focused on, "I am actually trying to give you a good tutorial."
Also the final image her forearm is as long as her entire torso, so this isn't actually that great a tutorial for making something that looks solid.
Thank you kindly for watching and commenting. I learn all the time and try to make better content. I will take for comments into consideration! 🙌
Just curios - what do you mean with constant exploding thumbnail?
And I agree - her forearm is too long, but this was just to show the process - and not create a masterpiece. It takes an enormous amount of time to do these tutorials…🎉
@@abtrapp oh great Anton! Thanks for explaining. Really good point about the speed. I hadn’t thought of that. 🙌
Hi Maria, I have a question, can Stable diffusion AI add color to an existing comic lineart? For example i.pinimg.com/originals/72/c6/f4/72c6f433cd20a40b82a5116a25999466.jpg
Thank you for your good video as always.
What a great question. And thank you for the praise, @Qjojotaro. Yes, it should be, but you need to add colors in an image sketch - and the use the input you showed me in ControlNet. I’ll try to do a tutorial on it and see how it works!🙌
@@LevendeStreg Thank you very much. I will wait for the tutorial. No pressure though.
@@qjojotaro2695 The list of videos I want to do is so long, so thanks for the patience.
@@LevendeStreg Thank you for your contribution. So far I have not seen anyone doing this coloring tutorial.
@@qjojotaro2695 Me neither. So it's top priority for me!