[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/ic/ - Artwork/Critique


View post   

File: 1.62 MB, 1400x955, 107879694_p0.png [View same] [iqdb] [saucenao] [google]
6823316 No.6823316 [Reply] [Original]

Hey guys, I'm just looking for a step by step guide on how some anime artists render. Sort of like pic related. Specifically I have a hard time figuring out how they add gradients to shadows. Can never tell if they use blending modes or airbrush/gradient tool.

>> No.6823318
File: 261 KB, 1071x934, Screenshot 2023-09-01 165914.jpg [View same] [iqdb] [saucenao] [google]
6823318

or like in the neck shadows here.

>> No.6823335
File: 1.31 MB, 875x2500, Taylor Pillow1comp.jpg [View same] [iqdb] [saucenao] [google]
6823335

>>6823316
fairly simple method
>multiply layer
>cellshade
>smudge blur where needed
>air brush where needed
>lock layer
>overlay layer clipped to multiply
>do the saturated shadow edge meme
>airbrush highlights

or you can just used the painterly smudge brushes and do everything in one layer like I did on picrel. I do this because I prefer having some brush work show

>> No.6823338

>>6823316
Huh, kind of reminds me of BBCChan's skin rendering, or Jirag-
>>6823335
Ah, speaking of which lmao

>> No.6823347

>>6823316
looks completely ai generated

>> No.6823361

>>6823347
Explain your thought process. Do an analysis of the image, show us why you think that. To me, it does not look ai generated. (I have lots of experience in both ai art and digital painting)

>> No.6823372

>>6823335
Op here. So I had an attempt on the neck area on this (ignore the thick lines it was done quickly) >>6823318
I locked the cellshaded layer and erased, then tried locking it and air brushing. each shadow is on a separate layer. I'm not sure this is what the artist here did?>>6823318

>> No.6823373
File: 15 KB, 442x482, Screenshot 2023-09-01 175909.jpg [View same] [iqdb] [saucenao] [google]
6823373

>>6823372
forgot the image lol

>> No.6823375
File: 3.27 MB, 1200x5144, 1612369391383.jpg [View same] [iqdb] [saucenao] [google]
6823375

>>6823316

>> No.6823381

>>6823361
it has the ai generated phenomenological impact, that's what matters

>> No.6823385

>>/6823347
It does not look AI and this anon is simply retarded

>> No.6823413

>>6823385
AI made a massive damage for anime coom
It's basically what >>6823381 said
>Verification not required.

>> No.6823433

>>6823381
If you are that anon who said it looks "completely" ai generated, back up your statement with visuals. Or simply conceed that no, it does not look completely ai generated.

>> No.6824085

>>6823373
op here. Not sure this is correct, still feels off

>> No.6824090
File: 663 KB, 574x841, el monstro del feminina..png [View same] [iqdb] [saucenao] [google]
6824090

>>6823433
>"completely" ai generated, back up your statement with visuals.
not the anon but I do have an example an entire folder to be exact here is one demonstration of his point.

>> No.6824094
File: 576 KB, 1536x982, It's over.png [View same] [iqdb] [saucenao] [google]
6824094

>>6823433
rendersisters.....
on a side note when i saw this result i couldn't stop laughing coomslop is literally indistinguishable from AI slop now .

>> No.6824097

>>6824094
as some anon said before,ai ruined generic coom,which is laughable
at least ai will filter those hacks

>> No.6824109

>>6824097
>at least ai will filter those hacks
I agree anon still after some introspection i think i have a theory as to why the bot detector can't distinguish the two anymore all anime pinupfags and smut peddlers essentially use the exact same rendering techniques as anon showed here
>>6823375
with some minor deviations here and there they are all just copying and making derivatives of each other without understanding or creativity becoming the human equivalent of a generative AI.

>> No.6824121

>>6824109
one of the reasons i didn't immediately accuse it of being done by ai was the consistency between drawings
but yeah,the theory sounds about right

>> No.6824139

>>6824121
>was the consistency between drawings
prompter trannies can also promt "consistent" characters I forgot what they used to do it exactly but it's possible.
usually AI slop has 3 to 4 glaring things that set it apart from human art.
>blur and plastic sheen
>machine artifacting
>hands and feet
>no intent/direction or understanding which leads to things like "is the placement of the light source actually sensable? ect.

>> No.6824183

>>6824139
there is always going to be something that is going to give it away

>> No.6824285

>>6824183
>there is always going to be something that is going to give it away
yeah but those are the most common and there is no real way to actually solve them because there hard baked in the machine process all NN systems suffer from this same ailment: They don't have what human mind would call a "concept" or self awereness, the way human mind learns is not by mindlessly storing data about random things, we create a hierarchy of similar things in regard to why and how they look, work behave and mean, so that whenever we encounter something that we have never seen, we can at least more or less deduct how that things should work or what it should do, we know that things should follow the laws of physics, etc.
let me give you an example A self driving car does not even know that it is driving or that it is a self driving car, a 10 year human child can learn how to drive after a few lessons with gramps those Teslas have perfect memory can process information hunderds of times faster than a meatbag and have more driving hours that all humankind combined in it's dataset yet it still sucks ass at driving.
why? because it's a supercharged pattern recognition bot it can only do things that sound reasonable in the limited context it is given through the data it's being fed.
untill they manige to develop something like an AGI that is self aware and possesed the ability to experience they will never overcome this limitation.

>> No.6824299

>>6824094
>literally indistinguishable
those detectors are worthless. use your eyes, or ask a pro.

>> No.6824837

OP here. idk why this thread devolved into ai talk. Regardless of if it is or not, I provided a 2nd image, which is what I really wanted to decipher rather than the 1st but left it just incase.

>> No.6824845

>>6823316
Add a multiply gradient layer of grey to darker gray and clip to the shading.
Also yeah you can use airbrush or blending blur brush. For the color you could eyedrop the skin base color and then increase the saturation a bit for the midtones

>> No.6824846

>>6824837
The resident AIfag in this board is actually just some permabeg that hates twitter anime pinups so whenever he sees them he had to remind the AI slop is going to take over their job or something

>> No.6824864

>>6824846
ah ok, noted
>>6824845 thanks anon