How to Make Character Concept Art with AI (Free and Easy) - Stable Diffusion Tutorial 2022
Vložit
- čas přidán 8. 09. 2022
- How to Install and Use Stable Diffusion (June 2023) - Basic Tutorial
• How to Install and Use...
Other options for using Stable Diffusion: / dreamers_guide_to_gett...
----------------------------------------------
Did you like this vid? Like & Subscribe to this Channel!
Follow me on Twitter: / albertbozesan
![Eminem - Houdini [Official Music Video]](http://i.ytimg.com/vi/22tVWwmTie8/mqdefault.jpg)
Keep making these. I love the workflow that you're using. It's extremely helpful.
Agree man, keep them coming. AI will only improve process, we’re the creators!
Watched your previous video with the chest, the potion and the flail (sounds like an inn name, or the start of a joke) which made me delve further into your channel and here I am. Subscribed! I am almost 60, been playing D&D since the 70s and, I've got to tell you, this is a dream come true!
Thanks so much! I bet this is a great tool for dungeon masters, too 😄😄
@@albertbozesan I would say the possibilities are endless!
Yesssss. Exactly what I was hoping for. Thank you kindly. Got yourself a sub
Really cool !! long concept art videos are definitely the most interesting !
Great video. You explain it very well. Keep the videos coming on this and related topics
Another banger video once again ❤️🔥🙏
Very cool what you are doing. Thank you!
Prompts like full page and front and back or concept art yeild some really impressive stuff, throw “symmetrical or symmetry” in and youll get some great basic character art. Started modeling some and am impress how seamless *some* of them come out!
At 15:40 where you clone over an unwanted area I found a useful method for this. Select the unwanted area, press Shift + F5, make sure the dialog that pops up has 'context aware' selected, and click OK. Not better, just an interesting alternative that I use a lot.
OMG. I love you for these videos.
Very relaxing voice and greatly explained
coz you did great work on the video, thank u for your help o7
Nice! Finally been waiting for your video! As an artist this type of workflow make life a whole lot easier and saves time too.. I have a question though.. Should you be worried about copyrights infringement? Like when freelancing and doing commissions ?
Depends on where you are from, but a quick google search came up with: "Under US copyright law, these images are technically not subject to copyright protection. Only "original works of authorship" are considered. "To qualify as a work of 'authorship' a work must be created by a human being," according to a US Copyright Office's report"
This is still a grey area, so I don’t think anyone can give you a definitive answer.
I’d say using the AI commercially is probably fine, as the AI doesn’t take from any one photo specifically but learns from a combination of multiple photos.
However, I would be careful about using an initial image that doesn’t belong to you (unless it’s public domain or under a similar license), as the results can be similar enough to be recognizable. Plus, it’s not very fair to the artist to just rip their image and stick it in an AI then call it your own.
Thank you for your videos Albert, great stuff. By the way, what GPU is on your PC?
I’m happy you like the vids! I run SD on an RTX 2070S.
the spikes are pretty smart if you wanna sever your carotid haha
I would love to see if you could draw a character in a different pose using AI. Say, take Batman and put him in an action pose.
Depending on how well that works (how consistently you can reproduce an OC), it might be of help with drawing manga panels.
That would be really cool. A big weakness of stable diffusion is making reliably similar looking images.
Thanks for the tutorial! I watched your last 2 stable Diffusion videos too, but i have a doubt. How much VRAM do you think i would need to run Stable Diffusion? I followed the simple instructions guide on reddit and looks like my gpu was out of memory (GTX 1650 on a Lenovo Ideapad Gaming, i know laptops aren't good for this but it's a necessity).
You should be able run it locally with around 6GB of vram - and more optimizations are coming. If that doesn’t work for you, have you tried running it online on a Google Colab Notebook? The Web UI works there, too. github.com/altryne/sd-webui-colab
I have 1060ti 6gb , its ok for 512x and 20steps
Also exist improvements which allows to create even larger images on same video card.
My suggestion just use google colab.
I have the same laptop with a 1650 and 16gb of ram and i'm able to run the optimized version.
In general latops do tend to have weaker versions of most GPUs; the rare exception are some laptop 3080s that come with 16GB of VRAM
@@pastuh I have the same video card. Have you tried creating images larger than 512? What are some ways to do it?
I started to put everywhere "vibrant" feels like adds not only more colors, but also makes more sharper objects..
how can I find the artist data? I dont know artists and their art? PLease help
Would it be possible to use the AI to create scenes for comics?
Can you reuse this armored character that you created here, and ask the AI to show him in different scenes, according to the verbal prompts that you give it?
In comics, it’s important to be able to reuse the same characters & backgrounds throughout the whole comic. The characters & backgrounds need to be depicted in different ways, according to the scene, such as from different camera angles, and under different lighting conditions. Also, the characters need to be portrayed in different poses, and their facial expressions also need to be changed. A lot of the time, characters will also need to be shown interacting with each other (e.g., sharing a hug). Comic characters should be placed within the backgrounds from the story, instead of being generated separately, and then having to be photobashed into the background. It would be difficult to rely on photobashing when it comes to comics, since it’s not easy to match up the lights & shadows of different images, and it would be tedious to do this since comics have a lot of panels to work with.
I really hope it will become possible to someday use AI to create comic books 😊
I have seen someone generate a unique face with Artbreeder, then use reface to put that face into different poses and situations, then use textual inversion/dream booth to train stable on that face. Then they could get the same face into different poses 😄
I’m not sure how to do the same for costumes/armor, but that idea was already pretty genius!
Such a novel workflow
im both not a coder and a dyslexic, is there a video tutorial on how to install this stuff cuz text-based guides make me feel like the dumb ass I really am.
I get it…I haven’t made a video because the methods change too often, making the vid obsolete in days. And there’s a few commands that should be copy-pasted rather than read off a screen.
Do you have a specific step that you’re getting stuck on? Maybe I can help here.
So you photobash and use SD as a render pass...
I suppose that’s a way to describe it 😄
Unfortunately all your Videos are in English. Impossible to translate them with subtitle in french 😢 and you are speaking so fast !
Sorry about that. I try to be as clear as possible and will keep my speed in mind in future videos!
to bad i dont have photoshop
Use gimp or similar. All the same methods apply.
Just as with digital illustration, AI only sheds a light on the actual fundamental art knowledge of person using it, if it's good it will show, it it's bad, it will REALLY show. Seriously, that hand, those proportions, the light... I can't believe serious illustrators and concept artists are worried about AI taking their jobs away.
If you think I'm just being mean, remember you called this "concept art" in the first place. I mean, for hobbyist, fair game, have fun, but concept art... please, do look up what actually entails.
19:40 Don't get your hopes too high on 1.5 "fixing hands", search for: "NEW 1.5 Stable Diffusion Available To Try Now! Is This New Ai Model Better Than 1.4?"
Yep, I mentioned 1.5 in this video, too :) looking forward to it being released open source!