@@Not4Talent_AI i can definitely tell you had at least some form of prior art experience just because you understand perspective, proportions, some color theory, and definitely values for shading. Good luck broski.
@@AerysBat hahahah idk about that. since I still couldnt make super specific changes or create on-model images fast enough to be competitive in the space.
@@qirmi2030 hahhaha some, yeah. I've drawn for a bit. But I almost never finished my drawings cuz I only really enjoy the ideation process. AI is helping a lot with the other parts. Hopefully I'll be able to do what I want to do eventually. tyty btw!
6:04 This is the moment that I give up on using AI and fix it manually. But the images are so different from what I want that I'm using more as sketches before a real piece, like testing ideas. I tried to use it to enhance the painting, but there were SO many errors with lighting, shadow, and rendering that I had to ditch AI. But, it's very good to test ideas. With generated AI images, it's clear when something won't work before doing it "for real".
Yeah, If you actually know how to draw Im sure you'll see tons of little (or not so little) errors that I cant see atm. Or at least cant identify and know how to fix them. And at that point its probably easier to just start from skratch. I agree that it can be great with iterating or just getting to a "final" version of the idea fast, to see if its worth finishing or what needs to be cahneged before drawing the final image
I'd like to know how I can keep my character design consistent when using AI to create compositions. When I use lineart in XL, it often doesn't stay true to my design, especially if the character has more cartoon-like proportions than those typically found in anime or video games.
Hmmm cartoon proportions are super hard bc AI hasnt seen as many of those. And they dont follow any super easy rules. I do plan on experimenting to find a way of making stuff like fan art and such. Which would requiere to mantain a character consistent. If I find a good way Ill make the tutorial eventually¡
If your carton proportion have sufficient picture you Can do a Lora out of that. If not IPadpter is the best thing you Can have to bring a bit of style from a picture sample. (It influence the network rather than constraining the generation with pixels)
This is really cool, it is pretty similar to my own workflow, but some good tips in there, I have also had good success with the cleaner extension to quickly remove objects.
Sinceramente, estoy un poco harto de que todas las imágenes que veo en diferentes aplicaciones de generación de imágenes y, concretamente, en Stable Diffusion, son de poses para revistas. Retratos en los que vemos al sujeto/a/e 'mirando' a la cámara. Mi imaginación va más allá y está relacionada con el dinamismo de las películas o series. Y no he encontrado mucho, por no decir nada, en ningún sitio. Así que te felicito por este canal y por las explicaciones pormenorizadas que das sobre cada uno de los temas que tratas en tus videos. Te animo a seguir adelante explorando este magnífico mundo de las imágenes dinámicas o de acción. Un saludo de un seguidor desde Canarias
La verdad es que te entiendo. Uno de los motivos por los que empecé el canal era este. Así que me alegra mucho ver que llega a la gente adecuada. Muchas gracias por tu feedback, ánimos y por tu comentario positivo!! Saludos desde Barcelona
At this point it would be a great investment to learn how to draw plus some anatomy lessons and then use the line art control net. that's what I use any way (plus some photo bashing), you can get much better results than the one you posted. great video btw
Im trying tbh, I do want to make a vid on going directly from a drawing, even one with some base colors. But I lack the artistics skills to make something better than what AI can help me with. Id be super interested in seeing how you do it tho. Ive never had much succes using the lineart model. Less so with extreme foreshortening like in this video. Simpler images maybe. But these ones idk. If you can share some it would be super appriciated. Maybe we could see it on discord if you want. Ofc no pressure tho, I'll experiment on my own when I make the video for a workflow like that. Even if my artistics skills aint there yet, the process could be followed with someone better than me anyway. Thanks for the feedback and for sharing!!
@@marcozolo3536 hi!! Sorry, Not atm but I do plan on testing both krita and flux. So even if I might not make it in time for a tutorial on flux integrstion. There could be videos related to that on the future
Ill send you the full name later. Atm I dont have electricity to check the full link 😂. I think you can se the name in the vid, but Ill try to send you the link as well
Just mouse. If I think I'll need a graphic tablet I go draw directly in photoshop to inpaint regularly later. Like in the pirare image when making the depth map
LOL, your workflow is absolutely on point, but basically shows that what SD is... is a style transfer colorist... it's basically doing nothing else for you.
hahahhaa tyty! that and iterating on the contents shapes, lighting etc 😂 its doing a lot of the parts I dont really enjoy about drawing, or Im to bad to do myself still 😅
I studied 3D and got into drawing there. So I've been drawing as a hobby for a bit. not too seriously, but I enjoy it from time to time. mainly sketching 😂
tyty!! hahahah little by little. I talk about a lot of topics at once xD I'm a 3D modeler and texture artist atm, as a full time job. Ai is something I'm interested in and experiment with on my spare time
I got the feeling your perspectives are too extreme with too much deformations. I suspects there isn't that much training data. Have you tried with less extreme camera deformations? And perhaps it could worth doing a lora with more perspectives to help the model ?
The less deformation and less extreme de perspective the easier it is. Thats why I chose hard ones to make sure it was viable on any idea. I do think that training a lora couls help yeah 😂
bro is just becoming an artist at this point
haahhahhaha sure would like to xD ty btw!
@@Not4Talent_AI IMO you could already be applying to art studios and simply show them your RUclips channel as your portfolio
@@Not4Talent_AI i can definitely tell you had at least some form of prior art experience just because you understand perspective, proportions, some color theory, and definitely values for shading. Good luck broski.
@@AerysBat hahahah idk about that. since I still couldnt make super specific changes or create on-model images fast enough to be competitive in the space.
@@qirmi2030 hahhaha some, yeah. I've drawn for a bit. But I almost never finished my drawings cuz I only really enjoy the ideation process. AI is helping a lot with the other parts. Hopefully I'll be able to do what I want to do eventually.
tyty btw!
my favorite SD youtuber. Love that you dont repeat the same basic tips every channel does. You experiment and explain everything very well.
I try my best hahaha. Happy to hear you like it. Thank you!!
You have brought SD near to story boarding. Excellent work
hahaha tyty!
Perspective and dynamism is one of the few limitations Stable Diffusion has to making it's images look actually legitimate. Great tutorial my man.
Thank you! Hope it helps!
Your conintued explorations into this new technology is the best. Thank you for all the content you make.
thank you for your continued support!! helps a lot! hope to keep providing 😄
Wow, he's producing stuff artists already done before, but using corporate service subscriotion based on stolen data. So brave!
@@milionST It's locally installed, so not service subscription. The rest you are right on
6:04 This is the moment that I give up on using AI and fix it manually. But the images are so different from what I want that I'm using more as sketches before a real piece, like testing ideas. I tried to use it to enhance the painting, but there were SO many errors with lighting, shadow, and rendering that I had to ditch AI. But, it's very good to test ideas. With generated AI images, it's clear when something won't work before doing it "for real".
Yeah, If you actually know how to draw Im sure you'll see tons of little (or not so little) errors that I cant see atm. Or at least cant identify and know how to fix them.
And at that point its probably easier to just start from skratch.
I agree that it can be great with iterating or just getting to a "final" version of the idea fast, to see if its worth finishing or what needs to be cahneged before drawing the final image
I really love bashing multiple good parts together, a skill i learned from you. :D
Hahahha its so nice to do and doesnt take much tbh. Tyty btw, glad it helped!
Your content is an absolute brilliant for SD users. Thanks a lot
happy to hear that! Thanks for your positive comments!!
Awesome video! You make the best SD videos. I learned a lot of stuff from you!
thank you so much!!
I render the first pass with a solid plain background and use ic-light to iterate the background in a second pass after the pose has been established.
Thats a good idea too yeah. Ty for sharing!
Espectacular como siempre. Tus videos están a otro nivel!
Muchas gracias!! Espero que sea útil jajajja
Need more of you bro! The last great teacher not focused on StinkyUI
Thank you!! Hahahhaha comfiUI is great but Im too lazy for that yet 😂
I'd like to know how I can keep my character design consistent when using AI to create compositions. When I use lineart in XL, it often doesn't stay true to my design, especially if the character has more cartoon-like proportions than those typically found in anime or video games.
Hmmm cartoon proportions are super hard bc AI hasnt seen as many of those. And they dont follow any super easy rules.
I do plan on experimenting to find a way of making stuff like fan art and such. Which would requiere to mantain a character consistent.
If I find a good way Ill make the tutorial eventually¡
If your carton proportion have sufficient picture you Can do a Lora out of that.
If not
IPadpter is the best thing you Can have to bring a bit of style from a picture sample. (It influence the network rather than constraining the generation with pixels)
been waiting for this for a long time
hahhahaha hope it helps then!!
@@Not4Talent_AIMe to😊
This is really cool, it is pretty similar to my own workflow, but some good tips in there, I have also had good success with the cleaner extension to quickly remove objects.
oh yeah that's a pretty common one. Never used it tho idk why tbh hahahaha I've seen it used and it's cool. I'll see if I get into it xD
Tyty btw!
THE GOAT IS BACK!
😂😂 tyty its been a while
Super high quality tutorial, as always! Thank you for sharing it! :D
Ty for watching and leaving nice feedback!! Hope it helps!
I always learn so much from you bro. Thank you so much. I realise I need to up my artistry skill to get the most out of SD
Happy to help!! I do think that leveling up artistic skills is always a good thing tbh hahahaha tyty!!
bro is giving me hopes that my control freak ass really may be able to do that comic I want to make someday
hahahhaha I hope so xD That'd be kind of my goal as well xD
Absolutely beautiful. You really are using the pose-depth-map a lot. Might need to implement that after all... Hmm.
Ty!! I do use that a lot. For most of the images I make tbh 😂😂 very very usefull
The tutorial we all needed. Thanks a bunch, man!
Thank you for the positive comment!!
Sinceramente, estoy un poco harto de que todas las imágenes que veo en diferentes aplicaciones de generación de imágenes y, concretamente, en Stable Diffusion, son de poses para revistas. Retratos en los que vemos al sujeto/a/e 'mirando' a la cámara. Mi imaginación va más allá y está relacionada con el dinamismo de las películas o series. Y no he encontrado mucho, por no decir nada, en ningún sitio.
Así que te felicito por este canal y por las explicaciones pormenorizadas que das sobre cada uno de los temas que tratas en tus videos. Te animo a seguir adelante explorando este magnífico mundo de las imágenes dinámicas o de acción.
Un saludo de un seguidor desde Canarias
La verdad es que te entiendo. Uno de los motivos por los que empecé el canal era este. Así que me alegra mucho ver que llega a la gente adecuada. Muchas gracias por tu feedback, ánimos y por tu comentario positivo!!
Saludos desde Barcelona
Awesome video, i will love a video on character consistency while able to have all this control.
ty!! I think the only main diff would be to inpaint the body using a previously created lora
One day you will make an anime with just ai tools.love this tutorial. Please make more
hahahahha that would be so dope
tyty!! I will!
Just outstanding stuff!
thank you very much!
At this point it would be a great investment to learn how to draw plus some anatomy lessons and then use the line art control net. that's what I use any way (plus some photo bashing), you can get much better results than the one you posted. great video btw
Im trying tbh, I do want to make a vid on going directly from a drawing, even one with some base colors.
But I lack the artistics skills to make something better than what AI can help me with.
Id be super interested in seeing how you do it tho. Ive never had much succes using the lineart model. Less so with extreme foreshortening like in this video.
Simpler images maybe. But these ones idk.
If you can share some it would be super appriciated. Maybe we could see it on discord if you want.
Ofc no pressure tho, I'll experiment on my own when I make the video for a workflow like that. Even if my artistics skills aint there yet, the process could be followed with someone better than me anyway.
Thanks for the feedback and for sharing!!
Any tutorials on how to integrate Flux with Krita AI?
@@marcozolo3536 hi!! Sorry, Not atm but I do plan on testing both krita and flux. So even if I might not make it in time for a tutorial on flux integrstion. There could be videos related to that on the future
May I ask, which depth and openpose model that you used with PonyXL?
I never found one that work well.
Ill send you the full name later. Atm I dont have electricity to check the full link 😂. I think you can se the name in the vid, but Ill try to send you the link as well
Diffuser controlNet works for me i use the FULL variant
oh, shit, sorry totally forgot XD
As lefourbe says, that's the good one:
huggingface.co/lllyasviel/sd_control_collection/tree/main
@@Not4Talent_AI Thank you!
You use a graphic tablet in inpaint sketch or just mouse?
Just mouse. If I think I'll need a graphic tablet I go draw directly in photoshop to inpaint regularly later.
Like in the pirare image when making the depth map
Awesome video, ty.
hope it helps!! tyty
LOL, your workflow is absolutely on point, but basically shows that what SD is... is a style transfer colorist... it's basically doing nothing else for you.
hahahhaa tyty! that and iterating on the contents shapes, lighting etc 😂 its doing a lot of the parts I dont really enjoy about drawing, or Im to bad to do myself still 😅
Respect you ❤ thx a lot always
thanks to you for the support and encouraging comments!!
Well, AI is just another tool after all. If you are familiar with it, and you are skilled enough, your image will be great eventually. 😉
100% agree! It helps build other skills in the process too, which is fun hahaha
it more about how much general knowledge you have about fundamental of art.
@@You_are_wrong99 Yep even with just purely prompting a real artist is going to have better results than someone with no art skills.
You are a pro😅 better than me
Jahhahahaha idk If Im better than anyone tbh, but tysm!
what shade lipstick is you wearing?
none, but I have noticed that it seems like I do xD no idea why tho. probably the lighting
buen video bro
Ty tio
i'm just wondering, but is your hobby drawing or you studied art before you started with ai art?
I studied 3D and got into drawing there. So I've been drawing as a hobby for a bit. not too seriously, but I enjoy it from time to time. mainly sketching 😂
@@Not4Talent_AI I can tell, but wanted to be sure.
Subscribed, but you're too advanced for me rn. 😅 Were you an 'actual modeller/ artist/ painter' before you got into AI?
tyty!! hahahah little by little. I talk about a lot of topics at once xD
I'm a 3D modeler and texture artist atm, as a full time job. Ai is something I'm interested in and experiment with on my spare time
Wow
I got the feeling your perspectives are too extreme with too much deformations. I suspects there isn't that much training data. Have you tried with less extreme camera deformations? And perhaps it could worth doing a lora with more perspectives to help the model ?
The less deformation and less extreme de perspective the easier it is. Thats why I chose hard ones to make sure it was viable on any idea.
I do think that training a lora couls help yeah 😂
do they allow nsfw positions???
yes, there are no limits with the local install
👋
👋👋