Looks like I'm going to need more RAM. Waifu2x uses 14 GB of RAM at these settings

Looks like I'm going to need more RAM. Waifu2x uses 14 GB of RAM at these settings.

Attached: 1543461731289.png (990x1034, 79K)

>he fell for the 16GB meme

Attached: 1533832227205.gif (1000x1000, 818K)

RAMlets, when will they learn?

16gb of ram is turning into 8gb in #currentyear. with ram dropping in price now there is absolutely no reason to not have 32gb.

Attached: poointhemultithreadpajeet.png (2560x1440, 3.7M)

Wintoddlers BTFO

Attached: 1407400455791.gif (700x704, 295K)

>he doesnt know about superfetch

>Waifu2x
Just use the web version faggot

you must be doing something wrong, waifu2x actually use very little ram to run

use it at these settings and it'll eat more than 16 GB of RAM and take hours if you don't have a Nvidia GPU

Attached: 1535159143144.png (1004x477, 62K)

Also this is very high quality settings if you want the best output so its going to be demanding. I have image comparisions using the different models, UpRestNet10 seems the best quality image by far, Split Size, the higher the value the more memory it will use but the faster the processing.
TTA Mode also increases quality of the image but makes the processing take longer and is more demanding on your CPU and I think RAM too.

quad cores in a year 2o19

Attached: snapshot_14.23_[2016.09.27_04.27.38].png (1920x1080, 2.64M)

I hate how waifu2x and esrgan are pretty much nvidia only. fucking retarded chinks

What does this do?

this is not a quad core tho

You can also use your CPU. Its slower though.

You SHOULD use your CPU as it gives better results (just like for video encoding)

Is it worth it? Post some results.

Image quality: best to worst
2-D illust (CUnet Model) > 2-D illust (UpResNet10) > 2-D illust (UpRGB Model) > 2-D illust (RGB Model) > Photo (UpPhoto Model) > Photography, Anime

Using 2-D illust (CUnet Model) with max split size and batch will require shit tons of RAM. If you use a smaller split size it'll use more CPU and much less RAM but will be slower. Split size and batch don't affect image quality.

>2-D illust (CUnet Model) > 2-D illust (UpResNet10
Can somebody confirm?

Might try upscaling an anime to 1080p if it already isn't using waifu2x at cranked settings. Will need a powerful computer or it'll take months.

>still using waifu-2x when we have A.I assisted upscaling now
How's it living in 2012 gramps?

What exactly do you think waifu2x is you brainlet?
Do you even know what a DCNN is? Or does something have to be a GAN to be "AI"?

>What exactly do you think waifu2x is you brainlet?
Weeb shit that will never come close to any paying solution like Let's Enhance

Why the fuck are you using your cpu for this? Use your fucking graphics card, it uses your vram then
No, waifu2x caffe's the one that's nvidia only, and there's a fork that gives amd support for your dumb little card basically nobody uses and bothers with.

>found the gooey fork
>but didn't find the amd fork
Are you retarded?

When will the AMD fork get a proper GUI to change settings like Caffe?

> all this back-peddling

>Disk 0 (D)
>Disk 1 (C)
>Disk 2 (E)
My autism is triggered.

COPE