Continuing from the last thread.
for %%f IN (*.jpg) do (
ffmpeg -loop 1 -i "%%~nf.jpg" -an -c:v libvpx -qmin 16 -qmax 16 -quality best -threads 4 -t 2 -r 1 "%%~nf.webm"
)
Save as .bat and run in the directory of JPG images.
Example Webm is 306KB and is transparent. Original PNG was 1,190KB. File size reduction: ~4X.
>Let's transform our images into a lossy format
Fuck off.
Intentionally reducing the quality of your images is shitposting.
Proof of webm transparency support in Opera 37.0
Here is the original PNG. A full fat 1,190KB
>>142645669
>image file size reduction: ~4X
So, it's pointless.
>>142645754
You can change the -qmin and -qmax to 4 to have very little quality loss.
I forgot to mention that is what what my webm was encoded with.
>>142645980
It can be more depending on source. VP8 compression is just that good.
Here is another example. Webm was encoded with -qmin and -qmax of 4 (highest vp8 quality) and has a file size of 279KB. PNG used was 1,363KB. Total file size reduction: ~5X
>>142646189
Here is the original PNG. A full fat 1,363KB
Here is another example. Webm was encoded with -qmin and -qmax of 4 (highest vp8 quality) and has a file size of 193KB. PNG used was 1,177KB. Total file size reduction: ~6X
>>142646420
Here is the original PNG. A full fat 1,177KB
wtf
Just use jpg.
>>142646558
Does not support transparency, needs to be 2X as big in file size to compete with VP8 used in these static webms.
Gonna try this out.
png is 1.17mb
>>142646558
Better yet who cares about filesize anymore
>>142646844
qmin and qmax set to 4 for webm
file size is 261kb
whoa
>>142645669
>noticeable quality loss
>doesn't work on all browsers
>load time is longer than the original png
Give this man a Nobel prize.
>>142646894
This. Let's all demand 4chan accept TIFF images and upload raw 10-20MB images. Everyone has 1GB/s internet and 4chan has unlimited bandwidth for all users amirite? mobile user are a myth.
>>142647037
>mobile user are a myth.
Mobile users aren't a myth. But i sure wish they'd leave.
>>142646995
>>noticeable quality loss
I don't see it
>>doesn't work on all browsers
Why are you not using Opera or Chrome?
>>load time is longer than the original png
Are you on dial-up internet? Sorry to hear than senpai.
>>142647086
>I don't see it
Compare the text in >>142646420 and >>142646463.
Just for an example.
>Chrome
Because botnet.
>>142647086
Look closer
>Why don't you change your browser to what I tell you because I want to change what images you can post on this site!
Are you Hitler maybe?
Yeah a png loads faster than your gay webms even though it's 4x bigger, I surely have slow Internet.
>>142647157
I don't get it, text looks the same. Maybe it's my laptop. It's an old HP from like 2010.
>>142647249
>Maybe it's my laptop.
The text is very obviously corrupted in the webm.
If your laptop has such a shitty panel that you can't tell the difference then you are not in a position to make assessments of image quality.
If you just want to reduce filesize just churn the png through a few filters.
For anyone wondering what >>142647157 is talking about, look at the red text. Though apart from that the quality loss isn't that bad. I'm impressed.
Here is a link to compare them better:
screenshotcomparison.com/comparison/175284
>converting from lossy to lossy
>converting images to videos
Or you could just use mozjpeg. While the results might be noisier when zoomed in close, at least you can have full chroma resolution with it. Just compare picture related to >>142646189 and the source in >>142646244 and notice how the VP8 version looks notably blurrier thanks to 4:2:0 chroma subsampling.
>>142647666
You can see the general blurriness all over the edges, but nice try Satan.
>>142647693
>>converting from lossy to lossy
He's using png sources senpai, which I assume have the best quality.
>>converting images to videos
4chan refuses to allow webp
>Or you could just use mozjpeg. While the results might be noisier when zoomed in close, at least you can have full chroma resolution with it.
I honestly would rather have bleeding colors than the shitstain noisy artifacts JPG produces
>>142647819
I'd also like to point it out that it transparency doesn't work on all browsers since you like to ignore it.
>>142647807
Holy shit I must be going blind because the only difference I can see is the red text looks more blocky on the webm. That's it.
>>142647870
Who cares about transparency?
If you want to make reaction images for 4chan you can just put the appropriate blue-tone into the background. That used to be extremely common a few years ago.
If you want the transparency for other purposes (further editing) then use the original fucking png.
>>142647870
WHY THE FUCK ARE YOU NOT USING CHROME OR OPERA?
Who uses firecuck or internet exploder lmao?
>>142647883
Move your mouse over the two images rapidly, so basically you switch between them instantly and you'll see it.
>>142647928
>supporting chrome
>cuck
>lmao
Fuck off.
>>142647957
Yeah because supporting firefox which is infested with feminists and SJWs is a much better option.
>>142647997
Seriously, get the fuck back to >>>/v/.
>>142647819
Or you could just save the JPGs at a decent quality.
The chroma subsampling issues can be very much visible at 100% zoom already whereas with decent JPG quality you'll have to zoom in and stick your nose to the screen to see the noise clearly enough to be bothered by it.
>>142647883
Here's another mouseover comparison between >>142646189 (resaved as PNG) and the attached picture:
http://screenshotcomparison.com/comparison/175289
Notice how the VP8 version looks considerably blurrier, especially in red areas, thanks to the 4:2:0 chroma subsampling.
>>142647946
I can't see the difference family. Jesus christ, could I have eye cancer or something?
>>142648079
You probably just need a better monitor.
>>142648069
>using a JPG in a comparison
Are you retarded?
Here is a correct comparison everyone, webm was converted to PNG using ffmpeg.
http://screenshotcomparison.com/comparison/175291
>>142648449
>>using a JPG in a comparison
>Are you retarded?
Anon's point isn't to prove that lossy formats are lossy.
His point is to prove that jpg is better than webm.
>>142647924
It was OP's point you fagistani. Tell it to him.
lmao, what the fuck is going on here?
>>142648449
The comparison was between VP8 and JPG compression, not between source and VP8. Read the post and the labels next time.
>>142648755
Then use the same file size ya knobsocket.
Here I did a comparison for you. PNG is extracted from 279KB Webm. JPG is 282KB, so it even get's a slight advantage in file size.
http://screenshotcomparison.com/comparison/175292
JPG obviously looks like shit.
>>142648940
They're both shit.
Going that far down in quality is pointless. Are we facebook?
/g/ here. This is from a webp thread we had a while ago.
I think that in all honesty we should adopt Webp as soon as possible on 4chan. OP's method of image storage is shit but a clue as to how good WebP is. Compression is actually better in WebP than VP8 now I think.
Anyway pic related is a scaled down version of a test an anon did on /g/.
Full res one can be found here:
https://my.mixtape.moe/nqvdrz.png
It's shit
>>142646519
>wtf
>>142647086
>Why are you not using Opera or Chrome?
>>142647037
Who the fuck cares about mobile users?
>Are you on dial-up internet? Sorry to hear than senpai.
>This. Let's all demand 4chan accept TIFF images and upload raw 10-20MB images. Everyone has 1GB/s internet and 4chan has unlimited bandwidth for all users amirite?
This fucking nigger.
>>142647224
Not our fault if you use a shitty browser
>>142649905
You are just plain shitposting at this point.
Do people really use browsers other than chrome and opera? Do they like constant crashes and non-support for cool things like transparent webms?
Honest question here, I used firefox for about 4 years but gave up with all the instability issues it has.
>>142648940
>Then use the same file size ya knobsocket.
Look at >>142647693 in comparison to >>142646189 then. The JPG's got more noise if you look up close, but because of the full chroma resolution it still looks much nicer at 100% zoom. This was my main point anyway - full chroma resolution is a much nicer perk as far as quality is concerned, especially since you can just bump up the JPG quality to reduce the noise even further without ending up with that much bigger files.
Again, try actually reading the posts next time.
>>142649989
>>>/g/et yourself fucked.
>>142650216
>cuck
Quality response.
>>142650086
Not him but what do you think of Webp? It looks very promising especially what >>142649483 posted.
Also is it true webp now has better image compression than VP8 which it was based on?
>>142649483
Does webp kills the iqdb?
>>142650384
WebP suffers from the same issue of being limited to 4:2:0 chroma that I've been going on about in this thread, which is a pretty substantial issue considering that you can have full resolution chroma with JPG.
On the whole I don't think WebP offers good enough improvements for the massive loss of universal compatibility you get from using it. As far as new image formats are considered, something like FLIF and BPG are much more interesting.
>>142650250
Cut you deep enough
>>142650605
Webp is not supported on iqdb (for now). However you can easily convert the webp file to a png in ffmpeg (-i in.webp out.png) to use it on iqdb until it does.
>>142650690
See? You can do better when trying anon.
>>142650724
>no excuses for saucefags
Nice.
>>142650724
>>142650779
>implying you can't just search for the thumbnail like you're doing for most webms already
>>142650649
But webp has the most support in terms of being a meme format. It can replace png, jpg, and gif right? As I understand, it has lossless encoding, transparency, and animation support as well.
Looks like a great format to replace the clusterfuck of image formats we currently have.
>>142650860
>it has lossless encoding, transparency, and animation support as well.
So does apng.
>>142650860
GIF is better replaced with actual video files (which can just be webms or mp4s), and again, I don't see it offering enough over PNG and JPG to warrant the enormous loss in compatibility. Especially when it only supports 4:2:0 chroma in lossy mode whereas JPG can do 4:4:4.
>>142651072
>GIF is better replaced with actual video files (which can just be webms or mp4s)
Kinda seems like a chore to press play on a 2-3 second animation. WebP would just automatically play, no video player required. This seems great for webpages.
>and again, I don't see it offering enough over PNG and JPG to warrant the enormous loss in compatibility.
Well it basically replaces both formats and adds the ability to encode images with lossy compression and still maintain transparency. Seems like a big deal especially for web devs.
>Especially when it only supports 4:2:0 chroma in lossy mode whereas JPG can do 4:4:4.
Yeah but JPG will still look like shit despite that. See >>142649483
>>142650954
apng is a dead format unfortunately.
>>142650086
I'm only doing this because I love you Daiz. The JPG looks like shit.
http://screenshotcomparison.com/comparison/175298
>>142651419
>WebP would just automatically play, no video player required.
Protip: You need just as much video playing capabilities to play VP8 video inside a WebP container as you do to play VP8 video inside a WebM container.
And you can achieve "GIF-like" video on the web much the same with <video autoplay loop muted>
>Well it basically replaces both formats
Knowing people that'd probably just mean they'll fuck up whether to go lossless or lossy even more than they do with PNG and JPG right now.
>Yeah but JPG will still look like shit despite that.
Look at >>142646189 and >>142647693 at 100% zoom. The most notable difference between the two is the chroma subsampling issues in the VP8 version. The JPG is noisier up close, but you can reduce that by simply bumping up the quality, which doesn't increase the filesize THAT much. I'd take >>142648069 over the VP8 version in >>142646189 pretty much any day of the week.
>>142651727
>285KB
That's not my JPG, the ones I posted were 274KB and 353KB, encoded with mozjpeg.
>>142651808
Hmm you're right. The one I downloaded from 4chin is of different filesize? What the fuck...
>>142645669
I don't see anything transparent in that image
>>142652008
That's because you're using an obsolete web browser. Upgrade to chrome.
Oh boy I sure can't wait to have to open a video player to look at static images
>>142652089
You're supposed to convert them into jpgs after saving them.
>>142651754
You're okay with JPGs that have noise and destroy a lot of detail?
Also I'm not talking about a hacked webm to display an image. I'm talking about webp which probably has better image compression than VP8. Webp was based off VP8 I frame encoding but they are not 100% the same right?
>>142652066
>Upgrade to chrome.
Did not read a word after that
>>142652066
Also there was a hidden Carlos in my post
>>142652089
>what is image hover in 4chin settings
>>142652175
kek
>>142652175
What about rotational velocidensity?
>>142652175
Goddamn, that sounds way easier than what I've been doing (printing them)
>>142652219
>Webp was based off VP8 I frame encoding but they are not 100% the same right?
It's the same VP8 bitstream. If they've made any still image encoding improvements there it should have made its way back to the video encoding side as well. The lossless mode is slightly different since the image is stored as ARGB there, but for lossy compression it really shouldn't be any different than with this webm trickery here.
And yes, if we're going lossy then I'll take some noise with full-resolution chroma over less noise with half-resolution chroma. And filesize isn't honestly that critical that you couldn't just bump up the JPG quality some. You're not going to be destroying a lot of detail unless you set the quality to really low.
>>142650860
>meme format
Fuck off with this bullshit.
>>142652225
That was the end of his post. Noone did.
>>142652066
>>142652568
>not reading right to left
You are not weeaboo enough for /a/.
>>142652476
Now I see why the other anons love you so much. You're really fun to talk to and know your shit.
Anyway yeah for typical images WebP won't improve much. However what about high res photos? As phone displays start to exceed 1440p (see sony Z5 premium) and people start to adopt 4K res screen they'll want to look at nice 8-16MP hig res photos of stuff. Those will be fuckhuge and shaving them down to less than 10MB file sizes would help with loading times across all connections. Some users may have 1Gbit/s connections but it doesn't mean the website owner can serve them images at 1Gbit/s.
>>142649483
JPG on suicide watch?
>>142653015
jpg will never die. jpg is the strongest.
>>142653723
Probably this. JPG is the undead cockroach from 1992 that haunts us to this very day. We'll probably still be using JPG when we have 16K res monitors.
