How much technological advancements is due to weebs?
How much technological advancements is due to weebs?
Other urls found in this thread:
make.girls.moe
github.com
arxiv.org
github.com
twitter.com
Too much
A lot more than you'd think.
Not enough
anime was a mistake
more from china as there's no social life so people can only relate to drawings
it's also a plague in USA but not so much in EU luckily
this
>shooman
technology is a subset of anime
I wonder what technique they used. The technical paper at make.girls.moe described that they had difficulty in making a GAN based upscaler for GAN generated animu faces.
Not much.
Usually Math/compute students in Asia(china,south Korea,Taiwan,Japan) are weebs,but don't means NEET weeb is research.
Acceptable answers.
Waifus are the true patrician motivational muses
>"photoshop"
Pls
Of course it will look like pixelated shit, probably just upscaled with linear/cubic. Any file viewer can do the same.
For line art and images with few colors, there are far better algorithms. both vector and raster.
The right image is better but it's not perfect. Lines get curvy/weird, some get blurry or completely blend with color.
Try GMIC for GIMP. Specifically upscale/smoothing algorithms. It takes some knowledge to use but it's a very powerful tool.
>oilpaintings are technological advancements
Most video format progress, most all subtitles and a lot of programming.
High 10 Profile
I want to lick Elis pussy and grab her big boobs if you know what I mean.
Like Jow Forums only exists because of weebs
Moot was a normie, though.
Weebs are the bottom of the bottom of society. They basically suck and can't invent anything.
As far as video encoding standards goes, a lot.
greater-than sign shad((((person))))
I'd say it's more gamers in general but weebs own the cell-shading/"abstract graphics" front.
nice site; now I can make my own waifu
>what technique they used
github.com
arxiv.org
The right image is probably for now the best you can get. But I won't mind a comparison with GMIC by someone knowing how to use it.
Make your favorite doujins colorful by PaintsChainer.
github.com
moot was only a normie by weeb standards. By normie standards he was a weeb.
This
It's pretty interesting that the technique boils down to denoising an upscaled image with a 3 layer convnet. I thought they would've used a deconvolutional network.
It's crazy good, thanks. I've been struggling with coloring anything for a long time, hope this will help me, if not to get a sense of how to color things, maybe then to just make them colored at all desu
All of it.
Wrong.
no he wasn't he didn't even watch anime
No one knows what it's like
to be shadman
to draw bad, man
behind blue mice
most developers are furry or weeb.
lmao this isn't remotely true
>Japanese people
>weebs
from termcolor import colored
print colored('>Shadman', 'green')
you are not a true developer, then
cel shading is going to replace traditional 2D animation sooner than we think
I'm absolutely not a developer by profession, but I know plenty.
That doesn't mean to imply that I don't have development experience though; I have plenty. Mostly low level.
I'm neither a furry nor a weeb.
2d boys are cute though
It's probably not a GAN but just a regular NN predicting the pixels. GANs are notoriously hard to work with, but there's no reason they couldn't do it.
>I am not a developer
>but I know plenty
kek, java workers are not developers, user
I don't know anyone who develops Java professionally and didn't suggest that I do, so I'm not sure where that assumption came from.
we got bitcoin
Simpsons upscale when?
hopefully never, it's a shit show that's only good for reaction images
is it possible to add additional scalign algorithms to Photoshop for purposes like this?
>python
Not from a jedi
Still better than when FX did it
isnt this a bit of a fucking waste of time? you have to color it anyways, the hardest part of coloring is just getting the rgb codes, then you finally scribble all the colors and it looks worse than the line art. you might as well just open it in gimp and paintcan
what does that even mean
Stuff like this exists but it's paid plugins afaik. And Adobe is incredibly lazy with features.
I tried some stuff with GMIC. It's very hard to understand how it works so this image is not as good as waifu2x, but still better than default option.
I think if you combine the two resulting images (something like bilinear upscale and waifu2x) one could have an even better result.
Waifu2x doesn't seem to preserve something I can't quite put my finger on all too well, so maybe do chroma separately?
the way these algorithms work is blend/flatten everything out. but they are only that sensitive to detail and thin lines. in result you're left with either thick lines or no lines at all.
it probably works better on simpler images, not something with a painted feel