#but overall... generic process

[hehearse]

hi! would love to hear about ur art process <3

hi!! there's not much to my process really x'3 it's pretty simple and short!

first is stickmen stule sketch just to have something in front of my eyes, then colors, cause it's the most exciting part!

then it's fussing over colors a bit, sometimes with gradient maps (i figured out how they work and now i WILL use them everywhere), then when the colors look nice, a proper sketch time, and that sketch will serve as a guide for lineless !

Silly comics are even simpler cause there're only two-three steps. Possible thumbnails, then rough sketch, then lines?

(usually text (if there's any) goes under thumbnails but i outlined the comics about yjh keeping bodyparts safe before i drew it so no text was needed)

And i do have a few timelapses here and there but they... make my process out to be way more unhinged than it actually is.

Video ID: A sped up process of drawing an illustration for Lee Hyunsung and Yoo Joonghyuk in front of a crumbling wall.

[anemonet]

hey rememberthe reverse iterators au. are you ever going to post about that again because i think it was pretty cool!

pebbles when he didnt even get to try

[More of my reverse iterator au if anyone is wondering what is going on!]

[sideblogdotjpeg]

caldwell is my number 1 enemy "this is my frog character he is so silly and goofy" then why does he keep saying things that make me sad when i think about them

[mx-nada]

on my hands and knees BEGGING YOU on info for how to draw dreadlocks

i don't know if i could make a good tutorial so what i can say is that references are ur BESTEST friend— via pinterest or just any image search engine

i do have these below as examples of how i draw locs, if that's anything.. ^_^ it's best to think of them as simple shapes, and then just adding more details and textures from there

[st4rstudent]

tips on how to draw cogs?

truth be told, I haven't drawn many cogs (mainly just one or two), but I will try to answer this as generally as I can!

This is a long post so I'll post this under the cut.

my biggest tip, as well as for anything, is to break down the shapes. If you try to look at everything at once it becomes really easy to overwhelm yourself. Examples will be drawn sketchy, but hopefully the point gets across.

Our first test subject is mouthpiece which I've never actually drawn before [marketable plushies post doesnt count]. as you can see, I tried to break down a lot of the shapes into much more simpler ones. A lot of cogs seem to share some very obvious shapes, probably because a lot of them are object-related (normal cogs included- first that comes to mind are pencil pushers).

You can do this with really ANY character. Here's our second test subject DOPA for another example.

now these are mainly references from the pictures themselves, which are great as a starting point. my other big tip is just to practice. you don't have to do big full pieces every day, but even just little doodles will help you. i think its very rare for anyone do to super good their first time and this goes with anything, even if you're a highly skilled artist.

you've probably also noticed but I've left out details in my examples, such as the little dots on mouthpiece's eye borders. If you feel that adding all the small details would overcrowd your drawing or just not help it, then I think it's fine to leave them out (of course there are always exceptions to this but as a general rule, its ok).

along with removing details, i think adding details also cannot hurt [to an extent]. if there's something you think would look nice then add it. who cares its your drawing have fun. As an example I'll use prethinker. I always add the little line in the middle, which his model doesn't have. But this is also based off of the 1.3 promotional art as well. Another example could be that when I draw misty, I add a line that connects their eyebrows to their eye. idk just feels more robotic to me and i like it

and again as you notice, my drawings are not model accurate. your style is most likely not going to be exactly like the game and that's completely fine. adjust to your liking, interpret as you will. enjoy the process of creation.

And finally, I'll add my little tip for metal coloring. One thing about metal is that usually the lightest highlight is next to the darkest shadow

The gif goes as base -> shadow 1 -> shadow 2 -> highlight 1 -> highlight 2 -> cleanup + extra

I like to add a few light marks to indicate scratches. although if you wanted more depth, you'd take the darkest shadow and lightest highlight and put them together like this

often the metallic parts are not the focus of my works so I leave it without. As for the random black marks on the lineart, that just also gives it more depth.

And finally, look at other peoples art. Look at the way they draw and pick out bits and pieces of what you like and try to interpret that in your own style.

ok I think that's all. Hopefully this helped!

[greywooden]

my review of maxton hall

[poison-raika]

Слабкі сторони створення зображень ШІ

Найбільша вада програмного забезпечення для створення мистецтва штучного інтелекту полягає в тому, що воно повністю ігнорує основи та переходить безпосередньо до розфарбовування самого зображення. Ось чому він такий нестабільний і виникає так багато помилок обробки. Зазвичай, малюючи малюнок, ви вибираєте композицію, малюєте приблизний ескіз, вносите необхідні виправлення, малюєте штриховий малюнок і, нарешті, розфарбовуєте його. Не починайте малювати з нуля. Якщо у вас немає міцної основи, ви не зможете правильно пофарбувати. Отже, які основи ШІ? Спочатку створіть загальну композицію. Обчислення точного положення, координат і взаємного розташування кожного елемента. Розрахунок відносного розташування об’єкта, фону та переднього плану. Далі створюється та аналізується чернетка, щоб перевірити та виправити будь-які помилки обробки чи розрахунків. Нарешті, генерується сама картинка. Звичайно, якість навчальних даних і метод навчання також дуже важливі. Зрозуміло, що традиційний метод навчання показу учням великої кількості зображень і фотографій уже досяг своєї межі. Зайве говорити, що неможливо подолати всі проблеми, з якими стикається програмне забезпечення для створення зображень ШІ. Важливо створити систему, яка генерує зображення, прикріплює виправлені зображення до областей, де є помилки, і дозволяє машині навчатися, щоб вона могла розпізнавати та виправляти власні помилки. Крім того, також необхідна модель, подібна до тієї, що використовується для розпізнавання та ідентифікації обличчя, співвідношення позицій, співвідношення та форма кожної частини (елемента), наприклад, два очі, розташовані вгорі, ніс у центрі та рот. нижче, що теж потрібно вчитися. Реальність поточних (традиційних) методів навчання та генерації полягає в тому, що вони структуровані як система для копіювання підробки. Також важливо додати 3D-моделі (шари дроту) і дані руху до навчальних зображень, щоб допомогти машині вивчити їх форми та рухи. Якщо штучний інтелект не має міцної основи, воно не може сподіватися на подальший розвиток. Зрозуміло, що міцна основа необхідна не лише для створення іміджу, але й у всіх сферах. Автоматизація продовжить розвиватися в майбутньому, і її неможливо зупинити. Однак нинішня ситуація полягає в тому, що і інженери, і широка громадськість зосереджуються лише на швидкості обробки та потужності та нехтують базовим навчанням. Простіше кажучи, ми настільки зосереджені на результатах, що повністю забуваємо про основи, джерело всього.

<>

[1o1percentmilk]

the issue with AI chatbots is that they should NEVER be your first choice if you are building something to handle easily automated forms.... consider an algorithmic "choose your own adventure" style chatbot first

it really seems to me that the air canada chatbot was intended to be smth that could automatically handle customer service issues but honestly... if you do not need any sort of "human touch" then i would recommend a "fancier google form"... like a more advanced flowchart of issues. If you NEED AI to be part of your chatbot I would incorporate it as part of the input parsing - you should not be using it to generate new information!

[shannonpurdyjones]

Plot outlines are vital to my writing process. They help me hone in on the trajectory of the story, giving my brain something to ignore while taking the draft in an entirely different direction.

[brackenferns]

A question did you plot your Taryn X ghost fic aka outline it or just wrote it and let the story come to you?

I'm a methodological pantser (thanks ellen brock for helping me realize my writing process is not crazy ✨) so I started with a very loose idea of: (1) where I wanted the characters to start, (2) where I wanted them to end, and (3) big developments that needed to happen along the way. all the little details came together in the process of writing, revising, re-outlining, restructuring, then rewriting. that might sound wild haha but it's just how my brain works—so yes I had a plan, but my outline was very bare bones to begin with & became more fleshed out as I wrote!

[friendofthecrows]

I went through this phase where all of my main character sets had a structure similar to the following: two besties living their consensually codependent dreams, the world's most awful bastard (affectionate but also derogatory), and an irresponsible adult who sort of accidentally adopted these teenagers and now has to try to be the responsible adult.

And honestly? That was banger.

[berryblu-soda]

Seedling is such an interesting little story that i love, but i´d be lying if it wasnt almost completely sprouted (hehe) from the visual of those character duos that are a small child + big scary looking creature who acts as their guard, top tier trope frfr

(also huge shoutout to @ashrayus for hearing me out and adding onto the rambles so long ago, bestie the robot story wasnt working, we got plants now 🌱)

[cannibalismyuri]

hey sara when you read rwrb did you also assume alex knew he was bi up until his sexuality crisis revealed the truth to you or was it just me who labored under delusions for the first 25 percent of the book, a whole quarter of the way through, before getting the shocking surprise that alex claremont-diaz, main character of nyt bestselling debut novel red white and royal blue, by author casey mcquiston, was NOT an out and proud bisexual man?

no actually, because im not delulu like u are . anyway .

[rinisdrawing]

so ruin, huh…

[orcelito]

Fixing my dialog up

Blurgh

[poison-raika]

Weaknesses of AI image generation

The biggest flaw in AI image generation software is that it completely ignores the basics and just starts coloring the picture itself. That's why it's so unstable and makes so many processing errors. Normally, when drawing a picture, you decide on the composition, draw a rough sketch, make corrections as necessary, draw the line art, and finally color it. Don't just start drawing from scratch. If you don't have a solid foundation, you can't draw a picture properly. So what are the basics of AI? First, create the overall composition. Calculate the exact position, coordinates, and positional relationship of each element. Calculate the positional relationship between the subject and the background and foreground. Next, generate and analyze the rough sketch, and verify and correct any processing and calculation errors. Finally, generate the picture itself. Of course, the quality of the learning data and the learning method are also very important. It is clear that the traditional learning method of showing a lot of images and photos to learn has already reached its limit. It goes without saying that AI image generation software cannot overcome all of the challenges it faces. It is essential to create a system that can generate images, learn from corrected images, and recognize and correct mistakes. Furthermore, a model like that used for face recognition and identification is also necessary, and it is necessary to learn the position, ratio, and shape of each part (element), such as two eyes lined up at the top, a nose in the center, and a mouth below that. The current (conventional) learning and generation method is a system that is like copying a fake. It is also important to add 3D models (wire layers) and motion data to the learning images to learn the shape and movement. If AI images do not have a solid foundation, they cannot be expected to develop any further. It is clear that a proper foundation is essential not only in image generation, but in all fields. Automation will continue to advance in the future, and it cannot be stopped. However, the current situation is that engineers and ordinary people are only focusing on processing speed and processing power, and neglecting basic learning. Simply put, we are chasing results and completely forgetting the fundamentals, the origins.

<>