Why isn't everything upscaled, or at the very least denoised?
Why is the internet overrun with turd quality overcompressed low-res jpeg garbage?
Why are you ok with this?
>>134487990
There is no point in denoising unless the cause of the noise is degradation due to compression, which means somebody fucked it up when saving the source in first place.. Upscale is placebo, it doesn't give you anything.
Using JPEG for low-frequency images is retarded.
>>134487990
Sometimes its better to add noise. For example to remove mach bands. Also its not really possible to "denoise" something.
>>134487990
Because we're not all autistic techies like you, anon.
>>134489209
>>134489527
>>134489894
so, using the above image as an example, are you saying you appreciate the one on the left as much as the right one or that they are equal in quality?
>>134490108
That reminds me, I'm due for an eye exam. Haven't had one since 2005.
>>134490108
There are slight differences between the two, but neither one seems to detract from my ability to appreciate it. Nor do I think either looks worse than the other.
>>134490158
as long as you're not diabetic or something
>>134490183
wouldn't there be a difference between your ability to appreciate the art and an objective comparison of the images' quality?
>>134490278
That doesn't make the objective comparison somehow "better" than subjective. Beauty is in the eye of the beholder, after all. One-size-fits-all crap like you're spouting really limits your ability to appreciate things, even if it's in the name of "good" taste.
>>134490183
>>134490108
the combination of blur, noise, and artifacts is an assault to my eyes
>>134490351
so the fact that one image is of objectively higher quality while being otherwise identical holds no value whatsoever?
>>134490351
better > worse
pretty one-size-fits-all right there
>>134490375
But not to mine.
>>134490406
Not for me; either one would serve its purpose well enough. Any more than that is just autism.
>>134490108
>Poorly save a .png as a .jpg to create artifacts
>uses it as an argument for upscaling and DNR
I don't follow.
>>134490470
>>134490351
so in that case why does anyone bother restoring old paintings and frescoes? if the detail and quality don't matter as long as the general image is still there. It's an effort to preserve the quality of the original work as best we can without having access to the actual source
>>134490108
No, I think the one on the right looks better. I just meant that less noise does not always look better.
It's also true that there could be less compression artifcats on the internet, for example by using jpeg2000 instead of jpeg-1. As the name suggests jpeg2000 was developed 15 years ago, but everyone was already using jpeg-1 at that time and nobody wanted to switch. Even now jpeg-1 is the standard and there is (almost) no browser that supports jpeg2000.
>>134490514
This is proof of how well it works, the picture on the left is the original. It's a random jpeg image someone else posted earlier
>>134490615
Old paintings and frescoes have actual physical materials they are composed of, so naturally, restoring them is a good idea so that they are viewable, period.
OP's example isn't that.
OP's example is just two different file formats/filters/whatever applied to the original source, neither one of which appears to me to be better than the other. This debate thread is just autism.
>>134490653
well that's true if noise was meant to be there, or if it helps mask other flaws as was mentioned earlier
>>134490726
>>134490675
if autism speaks, you shout
>>134490826
Cute, anon.
Way to sidestep my point.
>>134490858
>restoring them is a good idea so that they are viewable, period
how exactly does this not support the OP?
file degradation is comparable to physical damage in its effect
>>134490926
If either one of OP's example looks unviewable to you, then your eyes work differently than mine. The differences are subtle enough that anyone who doesn't know what to look for (which is most folks) might not see them, let alone care.
It's like looking at two images of a carrot, with one being the next shade of orange different from the other. It's barely perceptible to the human eye, and not a big enough deal to go all autistic over like this. Most folks don't care about this stuff.
>>134491062
which is exactly where the problem comes from.
it wouldn't be an issue if people weren't this complacent. What is the point of having high resolution desktop monitors and TVs? a picture can be conveyed just as well with a color CRT television from 1960.
Without a push for higher quality and detail in visual content we'd be stuck with mediocre blurbs of pixels rather than the high-res images that are now available.
This is the reason blu-ray is a thing, and DVD before it, and VHS before that. There is no reason to limit ourselves with visible imperfection
>>134491294
That's a fair point, but the upper limit to what the eye can see kind of limits the demand for anything of greater resolution. It's like there's no point to going past that limit 'cause nobody can even see the difference anymore. That's where this just turns to autism IMO.
TV marketers and the like are gonna have to do harder and harder sales jobs, 'cause demand for stuff beyond 4K resolution likely won't be there.
As for techniques to sharpen images and expose more detail, most folks will leave it to people in the know about such things. A number of people would look at products using these techniques just for the novelty, but eventually just go back to the default "if it just works, that's all we need" attitude.
>>134490675
I'm pretty ignorant when it comes to these sorts of things, but how did upscaling and denoising restore the depth of color you see on the right? I thought upscaling was just blowing up a low resolution image and that denoising essentially amounted to blurring out noise and then sharpening the image.
>>134491294
This draws a parallel with audiophiles. Plebs just don't give a fuck and it's kind of depressing.
>>134491814
It's been a long time since I've seen that Tidus. I missed him.
>>134491516
that's very true
considering how the TV industry has been the past few years they'll always find a way to raise the bar. Depending how VR sells over the next few years that might be it.
but the issue is mostly that a lot of content online is automatically deteriorated with each upload, resulting in a swamp of compromised videos and images. I guess you could describe it as a preservation effort or something.
>>134491724
The colors shouldn't be too different in the right picture. You may be getting this impression because the color is more uniform thanks to artifact reduction
>>134491876
>a lot of content online is automatically deteriorated with each upload
Hmm, I wasn't aware of this, so I can see how images and videos would degrade over time with many uploads across the net.
What about image/video files that someone just downloads once and never updates? Since it's not being uploaded anywhere else, it shouldn't degrade, right?
>>134491814
I like to think it's more reasonable to reduce the amount of compression images are subjected to than buying silver-infused power cables with stands to keep them from touching the floor on a $50 000 speaker setup to enjoy a two minutes 5Gb 32-bit linear PCM recording of an orchestra that was made using $50 microphones with no preamp
>>134491947
true but so many people share content online that the source can sometimes be lost permanently. Have you ever reverse-searched a very low resolution image and found nothing helpful?
>>134492039
>>134491947
for example this random reaction image I shamelessly pulled from another thread
>>134492039
Yeah, a number of times; I figured it was 'cause of the ant-sized resolution.
>>134492091
It was beyond saving. Even the processed image still looks messed up and wavy everywhere
>>134492091
>>134492125
Though at least it does look sharper; I'd think the right-hand image would be adequate for most folks' purposes, and even untrained eyes would probably notice the difference in the left image.
>>134492118
even search algorithms can't see everything that is on the internet. Once the OC is too old, hosted somewhere too obscure, or even taken down entirely, it can no longer be accessed by the vast majority of people
Does moot's jpeg bullshit actually reencode the images or is it just a metadata wipe?
I'd be slightly disappointed in 4chan if it actually lowered the quality of jpeg images uploaded. Seems against the whole point of being an imageboard.
This thread's already full of bullshit, so I'll just say I would rather watch a show in a window at its correct resolution than cover it with an ugly smear so it looks "cleaner" when I blow it up way too big.
>>134492252
well, i have no clue but i suppose uploading a jpeg image right now will answer that question. original for this one was 1.5 MB exactly
>>134492283
idk mang that first pic doesn't look too much like an ugly smear to me
>>134491987
An amp with decent speaker or headphone setup with a flac is night and day compared to laptop speakers, and an uncompressed image file is much better than an overcompressed/perpetually degenerating jpeg. Despite that, low quality mp3s and jpeg/gifs are still the standard on the majority of the internet. That's all I'm saying.
>>134492330
>>134492252
well, looks like nothing changed here
>>134492361
Did you even look at it at full res? Fucking lines are a mess. Literal vaseline.
>>134492381
>gifs
APNG support when, moot?
>>134492413
at 1:1 ratio i can't tell at all. once i zoom in far more than anyone ever should i can see wobble around the eyes
>>134492413
>>134492283
lesser of two evils?
obviously we would all like to have the original vector file the image was rendered from after being drawn but as was mentioned before sometimes the source is simply unobtainable
>>134492091
The right is missing detail. Her left eye no loner has a pupil.
>>134487990
>left side
Because some faggots like to eat industrial tier garbage all day long.
>>134492381
definitely true. what i meant is that it's easy to go overboard/full placebo mode in the audiophile world
Most people wouldnt notice it unless you placed the image side by side and most people dont fucking care and dont go into an autistic fit over it.
>>134492683
most people would notice and then not care because we are all worthless pieces of shit
>>134492091
Why don't find the fucking original pic and save it?
>>134492715
Im pretty sure most people dont equate their self worth to being able to spot blurr on an image in an online imageboard. But if you want an excuse to feel like a piece of shit, more power to you sempai.
>>134492413
>>134492283
>not appreciating a free upgrade
I thought this was the hyper-autism scanlation thread for a second
>>134492920
who says there's no hyper-autism here
>>134487990
>Why isn't everything upscaled
Are you stupid?
>or at the very least denoised
Why denoise shit you can find on google with no artifacts? We are talking about anime pics here.
>Why is the internet overrun with turd quality overcompressed low-res jpeg garbage?
Are you seriously asking this?
>Why are you ok with this?
Because no one cares about people getting eye cancer unconsciously.
>>134493001
illuminating
>>134492885
Now that looks nice, what did you use?
>>134493243
the exact same thing as the other ones. It's a custom program that uses cuda to retrieve as much data as possible from an original image and recreate a higher resolution version of it. It's a pain in the dick to set up but there's a reduced version available online here if you just want to try it
http://waifu2x.udp.jp/
i'm fairly certain it can also be adapted to restore the frames in a video but I've never tried. madvr filters on mpc-hc are ok for anime if it's already in 1080p or 720p.
>>134492683
>>134492715
>>134492884
There was a time when /a/ would.
>>134493382
>most people
>/a/
>>134493382
how is that even possible
>>134493412
"most people" that aren't on /a/ aren't relevant to the discussion since they aren't on /a/.
>>134493382
actually there's a pretty big difference in file size and resolution. that's probably what caused the difference
>>134493382
If you post the same image in diferent qualities then yeah maybe some people will but unless theres some weird hivemind posting its very unlikely you will get that situation and even then its unlikely someone will care.
>>134493421
See the aspect ratio, colors and filesize.
>>134493446
You have people who unironically watch download HS when there's an alternative because of speed (same with camrips) but you also have BDfags who wait months, but in general people still like their shit like nice and clean.
>>134493367
Pretty sure there's a MACROSS OP upscaled with waifux2 somewhere, it looked meh.
>>134493582
kinda want to see ergo proxy or trigun processed with it. DVD-only releases suck
>>134493582
HS fags physically repulse me.
I thought upscaling was bad. People told me not to buy the FLCL blu-rays because they are an upscale.
>>134494028
>tfw reading this while listening to OK Computer
Also not happening since Manglove, you better start working on upscaling that yourself.
>>134494189
aaaaaaaaaa i replaced my gtx 660 with a 290x from amd
no cuda
the full version of waifu2x is incompatible
>>134494183
it's not inherently bad but most forms of upscaling cause more issues than they resolve
>>134494312
Get that 660 back then. I'd do it if I knew how to.
>>134494341
but muh vidya
>>134494183
>>134494323
It's bad on your hard drive for sure.
You also won't get any more details so it'll only be a better looking verson (unless it's a really shitty upscale) of what your video player can do on playback at the cost of a much bigger filesize.
>>134494419
true, madvr filters can definitely help but they require some pretty beefy hardware to run at full capacity. The one advantage presented by waifu2x specifically is that is looks a lot more natural than filters alone and doesn't require any processing at playback. I personally wouldn't worry about storage since 2TB drives sell for ~ USD 60
>>134493421
I bet it's just OP samefagging since IP count wasn't a thing yet.
I compared 720p to 1080 One Punch episode 1 files in mpc-hc and found the 1080 to be slightly more choppy. Is this normal or did I configure madVR incorrectly?
>>134495195
some settings on madvr need a lot of processing power, such as luma/chroma doubling. Some of the numerous upscaling filters are also a bit power-hungry. Try using the ones that rely on your GPU if your system has a relatively decent one installed
>>134495195
>slightly more choppy
You mean your player was dropping frames?
Update your video player. If that doesn't work throw out your toaster and buy a real computer.
>>134495281
I have an intel cpu overclocked to 4.4GHz and a 290x. Even with that setup I still can't max out all the options on madvr. SVP on the other hand is a breeze for the few occasions where i choose to enable it
>>134495319
I actually had to use svp while watching Knights of Sidonia. Sitting through the choppy 3D animation was excruciating
>>134495319
I'm not on windows so I use mpv. opengl-hq all the way.
My computer's not even that good and it runs pretty well. Maybe ten frame drops per episode, and I usually only know about them because it keeps a counter visible in the CLI.
I'm not sure how it compares to maxed madvr in quality, though.
>>134495281
>You mean your player was dropping frames?
No, it just somehow didn't look as fluid. Maybe my autism is just kicking in.
>Update your video player. If that doesn't work throw out your toaster and buy a real computer.
Using the latest version of KCP and my computer is pretty good. I guess I'll take a look at it again.
>>134487990
What the fuck does this have to do with anime?
>>134490653
Because jpeg2000 isn't free.
