|
|
afed
|
2024-02-08 07:14:48
|
https://caniuse.com/hevc
```Enabled by default in Nightly and can be enabled via the media.wmf.hevc.enabled pref in about:config. 10-bit or higher colors are not supported.```
|
|
|
Quackdoc
|
2024-02-08 07:15:39
|
> 10-bit or higher colors are not supported.
|
|
2024-02-08 07:15:47
|
ouch, does chrome suffer the same limitations?
|
|
|
username
|
2024-02-08 07:15:59
|
why would it
|
|
|
lonjil
|
|
lonjil
oh, I guess I might not have the defaults
|
|
2024-02-08 07:16:15
|
oh, the default is color management for tagged images only. So HDR images will be tone mapped, presumably to sRGB?
|
|
|
Quackdoc
|
|
username
why would it
|
|
2024-02-08 07:16:32
|
no idea, I dont see why this has the limitations either
|
|
|
|
afed
|
2024-02-08 07:16:48
|
not supporting 10-bit is strange, so hdr hevc doesn't work
and it's only for firefox, other browsers have it
|
|
|
lonjil
|
2024-02-08 07:17:09
|
We should all just use 10-bit h264 instead
|
|
|
Oleksii Matiash
|
2024-02-08 07:17:18
|
Btw, anyone knows why ff does not support hevc on system with hw decoding support? License issues or just nobody did it?
|
|
|
username
|
|
Quackdoc
no idea, I dont see why this has the limitations either
|
|
2024-02-08 07:17:19
|
likely because Firefox doesn't have a proper pipeline for handling HDR or high bit depth content
|
|
|
Quackdoc
|
2024-02-08 07:17:43
|
wait, do they just not work properly with 10bit images? thats... bad.
|
|
|
lonjil
We should all just use 10-bit h264 instead
|
|
2024-02-08 07:17:55
|
https://tenor.com/view/meme-man-panik-moving-head-gif-17318955
|
|
|
Oleksii Matiash
Btw, anyone knows why ff does not support hevc on system with hw decoding support? License issues or just nobody did it?
|
|
2024-02-08 07:18:13
|
this is what we are talking about, it just hasnt landed yet
|
|
|
|
afed
|
|
lonjil
We should all just use 10-bit h264 instead
|
|
2024-02-08 07:18:28
|
https://bugzilla.mozilla.org/show_bug.cgi?id=1711812
|
|
|
lonjil
|
|
afed
https://bugzilla.mozilla.org/show_bug.cgi?id=1711812
|
|
2024-02-08 07:18:37
|
lol
|
|
|
Quackdoc
wait, do they just not work properly with 10bit images? thats... bad.
|
|
2024-02-08 07:18:49
|
10 bit AVIF seems to decode just fine
|
|
|
Oleksii Matiash
|
|
Quackdoc
this is what we are talking about, it just hasnt landed yet
|
|
2024-02-08 07:18:50
|
Thank you, missed this point
|
|
|
lonjil
|
2024-02-08 07:19:20
|
The default is to only color manage *images*
|
|
|
username
|
2024-02-08 07:20:04
|
what OS are you on? last I checked HDR AVIFs showed up very dark and wrong in Firefox
|
|
|
Quackdoc
|
2024-02-08 07:20:05
|
well either way, can't wait for servo T.T seems like they might be looking at jxl support soonish too
|
|
|
|
afed
|
|
lonjil
10 bit AVIF seems to decode just fine
|
|
2024-02-08 07:20:44
|
10-bit av1/avif should work, but, hdr probably doesn't work properly, so just 10-bit sdr
|
|
|
lonjil
|
|
username
what OS are you on? last I checked HDR AVIFs showed up very dark and wrong in Firefox
|
|
2024-02-08 07:20:50
|
Linux. Do you have an AVIF file that is both 10 bit and HDR tagged I could test? The test image I found just says 10 bit, it might not be HDR.
|
|
|
Oleksii Matiash
|
|
lonjil
The default is to only color manage *images*
|
|
2024-02-08 07:21:22
|
And OOB it is set to wrong behavior, iirc. Like use color management for images with color profile, and do nothing without. For wide gamut monitors it is incorrect
|
|
|
lonjil
|
2024-02-08 07:21:33
|
indeed
|
|
|
|
afed
|
|
afed
10-bit av1/avif should work, but, hdr probably doesn't work properly, so just 10-bit sdr
|
|
2024-02-08 07:21:51
|
because 10-bit for av1 is mandatory and it's the main profile
|
|
|
username
|
|
lonjil
Linux. Do you have an AVIF file that is both 10 bit and HDR tagged I could test? The test image I found just says 10 bit, it might not be HDR.
|
|
2024-02-08 07:22:41
|
I didn't really do any proper testing I just went here https://people.csail.mit.edu/ericchan/hdr/hdr.html
|
|
|
lonjil
|
2024-02-08 07:23:34
|
that's kinda funny
|
|
2024-02-08 07:24:04
|
HDR JPEGs tonemap just fine, but these AVIF images are dark. How can they color manage JPEGs correctly but not AVIFs?
|
|
2024-02-08 07:24:43
|
Also I wanted to see how they are supposed to look like so I opened one in MPV, but it gave me totally incorrect colors instead 😄
|
|
|
Traneptora
|
2024-02-08 07:24:59
|
which one?
|
|
2024-02-08 07:25:02
|
now I'm interested
|
|
|
lonjil
|
|
Traneptora
which one?
|
|
2024-02-08 07:25:16
|
first one on that page
|
|
2024-02-08 07:25:38
|
looks like this
|
|
|
username
|
|
lonjil
HDR JPEGs tonemap just fine, but these AVIF images are dark. How can they color manage JPEGs correctly but not AVIFs?
|
|
2024-02-08 07:25:40
|
I assume the JPEGs are 8-bit and well those just tonemap themselves from what I know so Firefox isn't doing any work in that case
|
|
|
lonjil
|
2024-02-08 07:26:11
|
Firefox needs to read and properly apply the embedded ICC profile
|
|
2024-02-08 07:26:20
|
Otherwise it looks washed out
|
|
|
Traneptora
|
|
lonjil
first one on that page
|
|
2024-02-08 07:26:23
|
it's an AVIF file tagged as yuv444p10le with a GBR matrix
|
|
2024-02-08 07:26:31
|
which is incorrect and doesn't work with ffmpeg
|
|
|
lonjil
|
2024-02-08 07:26:47
|
ah! broken images that just happen to work sometimes, love it.
|
|
|
|
afed
|
2024-02-08 07:26:53
|
Firefox - Chrome
|
|
|
lonjil
|
2024-02-08 07:27:12
|
Someone should tell the Chrome people to amend the AVIF spec to allow incorrectly tagged AVIFs
|
|
|
Traneptora
|
2024-02-08 07:27:37
|
ye, yuv444 shouldn't ever have a matrix tagged as identity
|
|
2024-02-08 07:27:41
|
it should be tagged as RGB instead
|
|
|
lonjil
ah! broken images that just happen to work sometimes, love it.
|
|
2024-02-08 07:31:02
|
do note that re-interpret-casting the pixel data as gbrp10le works
|
|
|
lonjil
|
|
Traneptora
|
2024-02-08 07:31:15
|
`ffmpeg -i hdr_cosmos01000_cicp9-16-0_lossless.avif -f rawvideo - | ffplay -f rawvideo -video_size 2048x858 -pixel_format gbrp10le -i -`
|
|
2024-02-08 07:31:45
|
the colors won't be correct here because there's no color management in ffplay
|
|
2024-02-08 07:31:48
|
but they won't be mangled
|
|
|
lonjil
|
2024-02-08 07:34:04
|
actually
|
|
|
username
|
|
Quackdoc
has this actually shipped? last I checked was about a half month ago and it didn't work on stable
|
|
2024-02-08 07:34:34
|
put `9N4WGH0Z6VHQ` into https://store.rg-adguard.net/ with the dropdown changed to "ProductId" and download the .appxbundle and then hopefully `media.wmf.hevc.enabled` should allow HEVC to work in Firefox
|
|
|
lonjil
|
2024-02-08 07:34:41
|
the way it looks is a near perfect match for how it looks in the original SDR rendering, while afed's screenshot is incredibly overaturated.
|
|
|
Quackdoc
|
|
username
put `9N4WGH0Z6VHQ` into https://store.rg-adguard.net/ with the dropdown changed to "ProductId" and download the .appxbundle and then hopefully `media.wmf.hevc.enabled` should allow HEVC to work in Firefox
|
|
2024-02-08 07:35:15
|
I already have the HEVC extensions
|
|
|
username
|
|
Quackdoc
I already have the HEVC extensions
|
|
2024-02-08 07:38:53
|
ah alright then \👍 wasn't sure since Microsoft has tried hiding the the free version as much as they can to the point of downright removing the store page
|
|
|
Traneptora
|
|
lonjil
the way it looks is a near perfect match for how it looks in the original SDR rendering, while afed's screenshot is incredibly overaturated.
|
|
2024-02-08 08:10:31
|
it's definitely wrong
|
|
2024-02-08 08:10:59
|
ffplay doesn't color manage so it uses your monitor's color which is probably sRGB
|
|
|
lonjil
|
2024-02-08 08:12:33
|
maybe they made the image "HDR" by just giving it a crummy profile that makes it way too saturated
|
|
|
Traneptora
|
|
lonjil
maybe they made the image "HDR" by just giving it a crummy profile that makes it way too saturated
|
|
2024-02-08 08:13:31
|
for comparison
|
|
2024-02-08 08:13:45
|
second one is mpv, which understands `cICP`
|
|
2024-02-08 08:13:50
|
first one is eom which doesn't
|
|
2024-02-08 08:14:09
|
first one reinterpret-casts the BT2020/PQ pixel data as sRGB
|
|
|
lonjil
|
2024-02-08 08:15:58
|
the eom one is a bit duller than the original SDR version:
|
|
2024-02-08 08:16:06
|
I guess it's just a cosmic coincidence
|
|
|
Traneptora
|
2024-02-08 08:16:27
|
it's also notably less blue
|
|
2024-02-08 08:17:21
|
I also don't believe it's oversaturated, considering that mpv renders it the same way
|
|
|
lonjil
|
2024-02-08 08:17:58
|
I didn't mean that Chrome renders it in an oversaturated way
|
|
2024-02-08 08:18:08
|
I meant that that HDR image is too saturated
|
|
|
Traneptora
|
2024-02-08 08:18:24
|
oh you mean the actual image itself is not ideal
|
|
|
lonjil
|
|
Traneptora
|
2024-02-08 08:19:31
|
interestingly the image has no CLLI or MDVC data
|
|
|
VcSaJen
|
2024-02-09 07:11:15
|
People often say "if you complain about Chromium monopoly, why you're not using Firefox, huh?". Like it's a some sort of gotcha. Except I'm not using Chromium-based browser...
|
|
|
Quackdoc
|
2024-02-09 07:11:46
|
im waiting for servo T.T
|
|
|
Cacodemon345
|
2024-02-09 07:12:17
|
Maybe a port of Epiphany to Windows would be better.
|
|
|
Quackdoc
|
2024-02-09 07:15:52
|
I was never the biggest fan of epiphany
|
|
2024-02-09 07:16:19
|
I didnt mind the KDE one before they killed it however
|
|
|
spider-mario
|
2024-02-09 08:13:30
|
Konqueror used to be my main browser
|
|
2024-02-09 08:13:37
|
it actually lasted for quite a while
|
|
|
Traneptora
|
|
VcSaJen
People often say "if you complain about Chromium monopoly, why you're not using Firefox, huh?". Like it's a some sort of gotcha. Except I'm not using Chromium-based browser...
|
|
2024-02-09 08:28:08
|
I was just pointing out that switching from firefox to edge *because* of the chromium monopoly seemed like backward logic
|
|
2024-02-09 08:28:24
|
I don't think one's choice of browser as an end user should be based on fighting a monopoly
|
|
2024-02-09 08:28:42
|
but it seems strange to jump ship from firefox to chromium-based edge because you don't like the chromium monopoly
|
|
|
yoochan
|
2024-02-09 08:55:12
|
Firefox works well. I don't understand the grudge about performance. Using both browsers at work every days (both open all day long) both are perfectly responsive, the only which may seems sluggish sometimes is chrome...
|
|
|
|
paperboyo
|
2024-02-09 09:13:59
|
FWIW, Firefox is my main browser on OSX&Win desktops (Chrome on mobile). I also use Chrome (and Opera). Performance isn’t an issue, but Firefox is behind on some stuff (ICC, HDR, font cleverness…).
|
|
|
Jarek Duda
|
2024-02-09 09:16:44
|
https://www.phoronix.com/news/Mozilla-New-CEO-2024 maybe will revisit ignored trending (?)
|
|
|
|
paperboyo
|
|
Jarek Duda
https://www.phoronix.com/news/Mozilla-New-CEO-2024 maybe will revisit ignored trending (?)
|
|
2024-02-09 09:23:24
|
This obsession with the “future”… I would first focus on the past. And fix issues opened 16 years ago. 😜
|
|
|
yoochan
|
|
paperboyo
This obsession with the “future”… I would first focus on the past. And fix issues opened 16 years ago. 😜
|
|
2024-02-09 09:26:52
|
Or the _present_ with firefox having the worst score on interop tests for standards support 😅 https://wpt.fyi/interop-2023
|
|
|
Oleksii Matiash
|
|
paperboyo
FWIW, Firefox is my main browser on OSX&Win desktops (Chrome on mobile). I also use Chrome (and Opera). Performance isn’t an issue, but Firefox is behind on some stuff (ICC, HDR, font cleverness…).
|
|
2024-02-09 09:43:11
|
What is the issue with ICC in FF?
|
|
|
|
paperboyo
|
|
Oleksii Matiash
What is the issue with ICC in FF?
|
|
2024-02-09 09:45:27
|
https://bugzilla.mozilla.org/show_bug.cgi?id=455077, but plenty more (v4 buggy, some more exotic ones too, and HDR, up to a point, is ICC issue too)
Biggest one is that by default, Firefox isn’t colour managed at all (you need to change `gfx.color_management.mode`).
|
|
|
Oleksii Matiash
|
|
paperboyo
https://bugzilla.mozilla.org/show_bug.cgi?id=455077, but plenty more (v4 buggy, some more exotic ones too, and HDR, up to a point, is ICC issue too)
Biggest one is that by default, Firefox isn’t colour managed at all (you need to change `gfx.color_management.mode`).
|
|
2024-02-09 09:53:38
|
Well, yes, it is a shame that CM is not on by default (btw, I thought it is, just with 1 in that setting instead of correct 2), but when enabled it works correctly. Can say nothing regarding HDR, but for SDR display it works nice even with wide gamut monitor. So in theory it may be not ideal, but for 'just use' it is fine
|
|
|
lonjil
|
2024-02-09 09:56:56
|
It does not work correctly
|
|
2024-02-09 09:57:24
|
Even with the option to fully enable color management, some things are not color managed.
|
|
|
Oleksii Matiash
|
|
lonjil
Even with the option to fully enable color management, some things are not color managed.
|
|
2024-02-09 10:09:45
|
I meant images only, sure
|
|
|
lonjil
|
2024-02-09 10:16:24
|
Unfortunately I found that it doesn't work for at least some AVIF images.
|
|
|
Oleksii Matiash
|
2024-02-09 10:26:05
|
We do not use avif here 😅 Joking, but I believe I did not see avif images in the wild, fortunately
|
|
|
diskorduser
|
2024-02-09 10:28:34
|
Firefox is slow and sometimes unresponsive. There are several webapps don't work well with Firefox. I have the clear cache every week in firefox to make them work. Edge works flawlessly.
|
|
|
novomesk
|
2024-02-09 10:31:20
|
Many people observe AVIF only after they download it from website. For example this site uses AVIF: https://opensea.io/assets/ethereum/0x495f947276749ce646f68ac8c248420045cb7b5e/73660949971956018117082564691530062244240304751348656613467082491606652157953
|
|
|
Oleksii Matiash
|
2024-02-09 10:34:16
|
Well, it is visible, however it shows the point <@167023260574154752> was talking about: CM is on when image is inside page, and off when "open image in new tab". For jpeg it works correctly in both cases
|
|
|
yoochan
|
|
diskorduser
Firefox is slow and sometimes unresponsive. There are several webapps don't work well with Firefox. I have the clear cache every week in firefox to make them work. Edge works flawlessly.
|
|
2024-02-09 10:49:15
|
Never had to clear cache, what kind of app fails to work for you? If you can tell
|
|
|
Traneptora
|
2024-02-09 10:49:22
|
Interesting, the six digit nuke code site serves jxl?
|
|
|
yoochan
|
2024-02-09 10:49:25
|
Nice!
|
|
|
diskorduser
|
|
yoochan
Never had to clear cache, what kind of app fails to work for you? If you can tell
|
|
2024-02-09 10:52:46
|
The app is internal use only. Not available on internet
|
|
|
yoochan
|
2024-02-09 10:56:11
|
Porn has always been an early adopter of new techs 😅
|
|
|
sklwmp
|
2024-02-09 10:58:18
|
it's always either gaming or porn that adopts bleeding-edge tech for some reason
|
|
2024-02-09 10:58:29
|
or piracy, for that matter (AV1)
|
|
|
Traneptora
|
2024-02-09 10:59:17
|
I was thinking of a different site
|
|
2024-02-09 10:59:21
|
but lmao
|
|
|
lonjil
|
2024-02-09 10:59:46
|
Does porn actually lead much at all? I keep hearing that claim but I've never seen anything that seemed to actually be true.
|
|
|
sklwmp
|
|
Traneptora
I was thinking of a different site
|
|
2024-02-09 11:01:10
|
(2), no idea what that one is
|
|
|
VcSaJen
|
|
sklwmp
or piracy, for that matter (AV1)
|
|
2024-02-09 12:02:35
|
Yeah, like DJVU
|
|
|
|
okydooky_original
|
|
Dexrn ZacAttack
What are people using now?
|
|
2024-02-09 07:32:27
|
On desktop, I use Firedragon, which is a fork of Librewolf for Garuda Linux. On Android, I use either Cromite or SmartCookieWeb-Preview (GeckoView based, supports extensions).
|
|
|
username
|
2024-02-09 07:33:33
|
I think Firedragon is switching to be a fork of Floorp
|
|
|
|
okydooky_original
|
2024-02-09 07:50:16
|
I don't know if you saw my conversation in the main(?) channel (the one bridged with Matrix) about having an "album" function for JXL, where you save multiple, usually similar, images in one file to get compression benefits and ease of sharing. But, Hitomi is exactly the kind of use case for that, in terms of web stuff. They host visual novel game CG and they organize the images into groups based on what scene they depict. Each one of those groups could be a JXL using this hypothetical "album" function, which would compress in much further (since each individual image is so similar) and keep things organized and convenient.
If this function gets implemented, then it'd be cool to see it used in cases like this, or on photography and stock image sites that host image sets and variations.
|
|
|
Traneptora
|
|
I don't know if you saw my conversation in the main(?) channel (the one bridged with Matrix) about having an "album" function for JXL, where you save multiple, usually similar, images in one file to get compression benefits and ease of sharing. But, Hitomi is exactly the kind of use case for that, in terms of web stuff. They host visual novel game CG and they organize the images into groups based on what scene they depict. Each one of those groups could be a JXL using this hypothetical "album" function, which would compress in much further (since each individual image is so similar) and keep things organized and convenient.
If this function gets implemented, then it'd be cool to see it used in cases like this, or on photography and stock image sites that host image sets and variations.
|
|
2024-02-09 07:51:06
|
multipage images are specified as animated JXLs with negative frame duration
|
|
2024-02-09 07:52:20
|
That said a set of images is probably best left as individual images
|
|
|
jonnyawsom3
|
2024-02-09 07:52:34
|
They mean having it reuse content between frames
|
|
|
Traneptora
|
2024-02-09 07:53:12
|
ye, animated JXL can do that
|
|
2024-02-09 07:53:59
|
VN cg assets typically only contain one background and then multiple small overlays, so I believe they are suggesting that the overlays be baked into the frames
|
|
2024-02-09 07:54:09
|
which animated jxl could accomplish
|
|
2024-02-09 07:54:19
|
without recoding the common regions
|
|
|
|
okydooky_original
|
|
lonjil
Does porn actually lead much at all? I keep hearing that claim but I've never seen anything that seemed to actually be true.
|
|
2024-02-09 07:55:27
|
Eh. I think maybe only sometimes? I think it's more a case of rabid/aggressive adoption rather than actual innovation. Kind of like when a bunch of "artists" and animators made the switch from Source Filmmaker to Blender and starting utilizing more advanced rigs for character models. I don't think any of them really advanced 3D or Blender tech more than any other enthusiast. Like, Blender's developed just fine without them, and I'm talking about add-ons and everything.
|
|
|
Traneptora
VN cg assets typically only contain one background and then multiple small overlays, so I believe they are suggesting that the overlays be baked into the frames
|
|
2024-02-09 07:59:45
|
This was clarified in the older conversation. I do remember that.
However, any site that shares the outputs of those scenes would benefit from that. Or any user who wants to view scenes from any device's gallery without having to open the game (esp. if it only works on Windows and they're using iOS or Android).
Actually, I did a content rip of a couple VNs many, many years ago and the scenes were individual images. When I say scenes, I mean the "full CG" ones and not the ones where a characters are overlayed.
|
|
|
username
|
|
username
I think Firedragon is switching to be a fork of Floorp
|
|
2024-02-09 08:00:11
|
I probably could have worded this message better originally but I was talking about this https://forum.garudalinux.org/t/new-firedragon-version/33729
<@324721113324191747>
|
|
|
|
okydooky_original
|
|
username
I probably could have worded this message better originally but I was talking about this https://forum.garudalinux.org/t/new-firedragon-version/33729
<@324721113324191747>
|
|
2024-02-09 08:01:07
|
No, it wasn't you. I just had no idea what Floorp was. 😮
Thanks for the link. I hadn't checked Garuda news in a while.
|
|
|
Traneptora
|
|
lonjil
ah! broken images that just happen to work sometimes, love it.
|
|
2024-02-09 08:17:25
|
update: I checked the AV1 spec, and there's no field for pixel format
|
|
2024-02-09 08:17:41
|
instead RGB is signaled specifically by setting `matrix_coefficients = MC_IDENTITY`
|
|
2024-02-09 08:18:20
|
this, of course, raises the question "what happens if you set matrix_coefficients to identity, but the image is chroma-subsampled?"
|
|
2024-02-09 08:18:29
|
the AV1 spec forbids this
|
|
2024-02-09 08:18:39
|
> If matrix_coefficients is equal to MC_IDENTITY, it is a requirement of bitstream conformance that subsampling_x is equal
> to 0 and subsampling_y is equal to 0.
|
|
2024-02-09 08:19:14
|
so ffmpeg is incorrectly exposing yuv444p10le instead of gbrp10le here
|
|
|
lonjil
|
|
Eh. I think maybe only sometimes? I think it's more a case of rabid/aggressive adoption rather than actual innovation. Kind of like when a bunch of "artists" and animators made the switch from Source Filmmaker to Blender and starting utilizing more advanced rigs for character models. I don't think any of them really advanced 3D or Blender tech more than any other enthusiast. Like, Blender's developed just fine without them, and I'm talking about add-ons and everything.
|
|
2024-02-09 08:33:32
|
yeah. actually, I recently watched a video about that meme, and they reached out to some of the biggest artists in the old SFM and current Blender animated porn scenes, and asked them about it, and they said that the idea that they advanced anything is BS.
|
|
|
Traneptora
update: I checked the AV1 spec, and there's no field for pixel format
|
|
2024-02-09 08:34:04
|
aha
|
|
|
I don't know if you saw my conversation in the main(?) channel (the one bridged with Matrix) about having an "album" function for JXL, where you save multiple, usually similar, images in one file to get compression benefits and ease of sharing. But, Hitomi is exactly the kind of use case for that, in terms of web stuff. They host visual novel game CG and they organize the images into groups based on what scene they depict. Each one of those groups could be a JXL using this hypothetical "album" function, which would compress in much further (since each individual image is so similar) and keep things organized and convenient.
If this function gets implemented, then it'd be cool to see it used in cases like this, or on photography and stock image sites that host image sets and variations.
|
|
2024-02-09 08:34:38
|
> I don't know if you saw my conversation in the main(?) channel (the one bridged with Matrix)
matrix bridge has been dead for a while
|
|
|
|
okydooky_original
|
|
lonjil
yeah. actually, I recently watched a video about that meme, and they reached out to some of the biggest artists in the old SFM and current Blender animated porn scenes, and asked them about it, and they said that the idea that they advanced anything is BS.
|
|
2024-02-09 08:36:17
|
If you have a link, I'd like to watch that. I only just inferred that observation.
|
|
|
lonjil
> I don't know if you saw my conversation in the main(?) channel (the one bridged with Matrix)
matrix bridge has been dead for a while
|
|
2024-02-09 08:37:31
|
Yeah, it was, like, in November that the "album" idea I had was discussed. I remember something about adding a flag to the file could let a gallery app know to handle it as such (compared to an animation or something)?
|
|
|
lonjil
|
|
If you have a link, I'd like to watch that. I only just inferred that observation.
|
|
2024-02-09 08:40:23
|
https://www.youtube.com/watch?v=BYDr8jHYc4Y
|
|
|
|
okydooky_original
|
2024-02-09 09:00:53
|
Thanks.
|
|
|
lonjil
https://www.youtube.com/watch?v=BYDr8jHYc4Y
|
|
2024-02-09 09:07:07
|
I kind of doubt VHS, digital cameras, video streaming, and online credit card payments being a thing are *because* of smut stuff, but okay. Lol
|
|
|
Traneptora
|
2024-02-09 10:29:20
|
Online credit card payments actually have negative synergy with porn sites
|
|
2024-02-09 10:31:55
|
there's been a history of issues in that regard
|
|
|
|
okydooky_original
|
2024-02-10 06:02:43
|
Yes! Thanks for finding it. "Patches" was the other term that was used for compressing across frames/layers. I forgot about that.
I don't have any at the moment, but I'll send some over when I stumble on some. I'll also make sure they're SFW, so anybody can use them.
|
|
|
sklwmp
|
2024-02-10 06:28:42
|
i guess we can add hitomi to the list of websites supporting jxl... though not sure we'd wanna advertise that 😅
|
|
2024-02-10 07:13:25
|
too compute-intensive and not worth it to convert existing galleries, ig...
|
|
|
username
|
2024-02-10 07:28:30
|
youtube does something kinda similar with thumbnails, new videos will have both a WebP and JPEG encode of the thumbnails while older videos will just have JPEG
|
|
|
HCrikki
|
2024-02-10 07:57:54
|
Such galleries can be mirorred more or less partially, like for a sfw demonstrator
|
|
2024-02-10 08:00:29
|
Theres also the option to generate galleries from lossless ai prompts wich wouldnt have any copyright issues and have jxl served primarly to clients with direct download links in case anyone isnt viewing the format they want
|
|
|
jonnyawsom3
|
2024-02-10 12:52:26
|
I can't see how AI images could be called lossless...
|
|
|
w
|
2024-02-10 12:53:39
|
the same as how anything can be called lossless
|
|
|
Jarek Duda
|
2024-02-10 04:09:31
|
https://www.reddit.com/r/GalaxyS23Ultra/comments/1ang39u/samsung_yet_again_downgraded_the_quality_of/ they complain that with S24 RAW started using lossy with jxl?
|
|
|
_wb_
|
2024-02-10 04:21:24
|
They should have done that as an option imo. Making lossy obligatory is something people don't like...
|
|
|
diskorduser
|
2024-02-10 05:04:04
|
I think iphones use lossy compression for dngs/raw photos. Samsung is trying to copy them.
|
|
|
Kejchi
|
|
Demiurge
|
2024-02-13 12:20:51
|
Lossy might not even be the issue
|
|
2024-02-13 12:21:04
|
They are complaining about banding
|
|
2024-02-13 12:21:35
|
Banding is a decoding issue. Ideally you should never get banding when decoding a JXL...
|
|
2024-02-13 12:23:28
|
It probably is being decoded incorrectly, without dither, and to a color profile that poorly represents the colors that have bad banding
|
|
|
jonnyawsom3
|
2024-02-13 12:27:19
|
For a second I thought that was Windows, the joys of simplified icons
|
|
|
Kejchi
|
2024-02-13 04:06:07
|
https://f-droid.org/packages/com.dot.gallery/
|
|
|
Quackdoc
|
|
Kejchi
https://f-droid.org/packages/com.dot.gallery/
|
|
2024-02-13 08:29:39
|
>A11 or newer
|
|
2024-02-13 08:29:45
|
T.T
|
|
2024-02-13 08:29:59
|
thats a really odd target
|
|
|
VcSaJen
|
2024-02-15 12:35:18
|
wpclipart have JPEG XL among supported formats for clipart download
|
|
|
username
|
|
VcSaJen
wpclipart have JPEG XL among supported formats for clipart download
|
|
2024-02-15 12:37:40
|
this looks so unnatural and weird
|
|
|
VcSaJen
|
|
username
|
2024-02-15 12:39:42
|
- All the options are lowercase except "JPEG-XL"
- All the options are shorthand except "JPEG-XL"
|
|
2024-02-15 12:40:10
|
also the first two options look off center in a weird way
|
|
2024-02-15 12:40:29
|
what a strange little website
|
|
2024-02-15 12:40:44
|
these are just nitpicks but still
|
|
|
monad
|
2024-02-15 07:20:08
|
What's JPEG-XL?
|
|
|
Oleksii Matiash
|
|
username
- All the options are lowercase except "JPEG-XL"
- All the options are shorthand except "JPEG-XL"
|
|
2024-02-15 07:32:18
|
Jxl is too new abbreviation, people are not used to it?
|
|
|
yurume
|
|
monad
What's JPEG-XL?
|
|
2024-02-15 07:57:05
|
all JPEG standards minus JPEG XL? *grin*
|
|
|
Tirr
|
2024-02-15 07:57:47
|
or JPEG without extra large things
|
|
|
_wb_
|
2024-02-15 07:58:32
|
It's "JPEG XL" and "JPEG" (or "JPEG 1") in full, and "jxl" or "jpeg" if you want something shorter
|
|
|
Traneptora
|
2024-02-15 04:34:30
|
I wish JPEG XL stood for "JPEG eXtended Life"
|
|
2024-02-15 04:34:39
|
it doesn't, but it's not too late to backronym it
|
|
2024-02-15 04:34:40
|
:D
|
|
|
_wb_
|
2024-02-15 04:37:30
|
nobody really knows what it stands for anyway
|
|
|
Traneptora
|
2024-02-15 04:39:04
|
~~probably cause it doesn't stand for anything tbh~~
|
|
|
lonjil
|
2024-02-15 04:39:10
|
eXtreme Lulz
|
|
|
Traneptora
|
2024-02-15 04:39:17
|
iirc X is just `JPEG X*` and L is long-term
|
|
2024-02-15 04:39:25
|
but that's not really something you can spell out
|
|
|
_wb_
|
2024-02-15 04:42:02
|
JPEG XT was for eXTensions, I think
JPEG XS was for Speed (and "extra-small" in terms of complexity)
JPEG XR was for eXtended Range, since it can do HDR (not really a distinguishing feature since literally any JPEG codec can do that, but whatever)
JPEG XL is not clear but it was started shortly after JPEG XS, so there is probably some geeky "extra-large" element in the name (in terms of complexity/features). The L has also been said to stand for "long-term".
|
|
|
Traneptora
|
2024-02-15 04:44:24
|
tbf, OG JPEG might be able to do 12-bit and HDR but it's not widely supported
|
|
2024-02-15 04:44:34
|
so you shouldn't expect it to work
|
|
|
_wb_
|
2024-02-15 04:44:36
|
but yeah, JPEG is not particularly good at naming. I mean, they literally called their flagship codec (until JXL became the new flagship codec) "JPEG 2000", which may have sounded futuristic and cool when they were designing it but...
|
|
|
Traneptora
|
2024-02-15 04:44:56
|
tbf video codecs come up with silly ambiguous names too
|
|
|
lonjil
|
|
Traneptora
tbf, OG JPEG might be able to do 12-bit and HDR but it's not widely supported
|
|
2024-02-15 04:45:01
|
OG JPEG can do 10-bit and HDR even with just the parts that are commonly supported, though!
|
|
|
_wb_
|
|
Traneptora
tbf, OG JPEG might be able to do 12-bit and HDR but it's not widely supported
|
|
2024-02-15 04:45:11
|
fair enough, but this was several years after J2K, which can even do 38-bit for some weird reason
|
|
|
Traneptora
|
2024-02-15 04:45:13
|
"Advanced Video Coding" "High Efficiency Video Coding" "Versatile Video Coding"
|
|
|
w
|
2024-02-15 04:45:41
|
jpeg 2024
|
|
|
_wb_
|
|
lonjil
OG JPEG can do 10-bit and HDR even with just the parts that are commonly supported, though!
|
|
2024-02-15 04:45:41
|
it can, but not with the common decoders which are going to produce an 8-bit output buffer 🙂
|
|
|
Traneptora
|
2024-02-15 04:45:43
|
it's not like H.264 is that much sexier than MPEG-4 AVC
|
|
2024-02-15 04:46:03
|
we all call it H.264 but that's like calling JXL "18181"
|
|
2024-02-15 04:46:11
|
not exactly exciting
|
|
|
lonjil
|
|
_wb_
it can, but not with the common decoders which are going to produce an 8-bit output buffer 🙂
|
|
2024-02-15 04:46:29
|
yeah, but no difference in format. And though it'll be more bandy that you'd want, the HDR ICC profiles will still work.
|
|
|
_wb_
|
2024-02-15 04:47:26
|
fun fact: JPEG AIC was originally going to become some new codec, it stands for Advanced Image Coding (this was in the era of AVC). It then turned into a standard for image quality assessment and we are now referring to AIC as meaning Assessment of Image Coding.
|
|
|
Oleksii Matiash
|
|
Traneptora
"Advanced Video Coding" "High Efficiency Video Coding" "Versatile Video Coding"
|
|
2024-02-15 05:37:01
|
SD card standards - hold my beer
|
|
2024-02-15 05:38:43
|
Never understood why not just SD1\2\3.. R:xx W:xx IO:xx
|
|
|
CrushedAsian255
|
2024-02-15 09:06:36
|
“USB 3.2 Gen 2x2” anyone?
|
|
2024-02-16 02:28:06
|
Also USB-C is just a mess as well
|
|
2024-02-16 02:28:21
|
Cable 1: 2.0 speeds, 60W power
|
|
2024-02-16 02:28:39
|
Cable 2: 3.2(2x2) speeds, 15W power
|
|
2024-02-16 02:28:51
|
Cable 3: 3.0 speeds, 100W power
|
|
2024-02-16 02:28:56
|
Look identical
|
|
|
Quackdoc
|
2024-02-16 02:38:55
|
certified cables will have marking on them that denotes their speed, and sometimes usb-pd, unmarked cables have a specific "speed and power" rating that should be expected but I cant remeber what it is
|
|
2024-02-16 02:39:46
|
also you can blame the media for this, USB-IF had "marketing" terms availible for ages that no one followed because they wanted to be special goobers, the media refused to use the marketing terms, and companies obviously wanted to use what they consumers got force fed, but all certfied USB products will have "USB 10gbps" "USB 5gbps" on the packaging, I can't reber when they started making certified devices put the markings on cable
|
|
|
190n
|
2024-02-16 03:14:04
|
also even 20 gbps flash drives will often run at 10 gbps in practice cuz so few devices support gen2x2
|
|
2024-02-16 03:14:22
|
even most devices which support 40 gbps operation with thunderbolt 3 or usb4, don't support usb 3.2 gen2x2
|
|
|
Quackdoc
|
2024-02-16 03:15:12
|
[Hmmm](https://cdn.discordapp.com/emojis/654081052108652544.webp?size=48&quality=lossless&name=Hmmm)
|
|
|
190n
|
2024-02-16 03:20:33
|
it's so silly
|
|
2024-02-16 03:20:40
|
and usb4 even has an explicit 20gbps mode
|
|
2024-02-16 03:20:49
|
but that's different from 3.2 gen2x2
|
|
2024-02-16 03:21:20
|
hopefully usb4 controllers get cheap enough that drives can start supporting usb4 at whatever the speed of the flash is, 20 or 40 or something else
|
|
|
w
|
2024-02-16 03:24:40
|
amd displayport 2.1 uhbr 13.5 (54gbps)
|
|
2024-02-16 03:25:14
|
similar to HDMI 2.1 but gamers believed it was better
|
|
2024-02-16 03:26:55
|
this number marketing works
|
|
|
lonjil
|
|
CrushedAsian255
Cable 2: 3.2(2x2) speeds, 15W power
|
|
2024-02-16 10:04:11
|
So, USB-C cables can either be marked with their PD capabilities, or unmarked. Unmarked cables are assumed to work with 3 A and 20 V, i.e. 60 W. If you're only getting 15 W, your cable is broken.
|
|
|
CrushedAsian255
|
|
lonjil
So, USB-C cables can either be marked with their PD capabilities, or unmarked. Unmarked cables are assumed to work with 3 A and 20 V, i.e. 60 W. If you're only getting 15 W, your cable is broken.
|
|
2024-02-16 10:15:20
|
i think somethings going wrong with the PD processor, as it is stuck at 5V/3A
|
|
|
lonjil
|
2024-02-16 10:16:17
|
Maybe the aux data wires got pinched and aren't connected anymore
|
|
2024-02-16 10:16:59
|
Then it'd never negotiate a higher voltage
|
|
|
jonnyawsom3
|
2024-02-16 10:24:14
|
Maybe <#806898911091753051>
|
|
|
lonjil
|
|
Traneptora
we all call it H.264 but that's like calling JXL "18181"
|
|
2024-02-16 11:19:32
|
Hm, I don't think that's right. Under ISO/IEC, MPEG-4 AVC is 14496 part 10, while H.264 is the ITU-T designation. JPEG1 is T.81, but it seems JXL doesn't have an ITU-T recommendation yet.
|
|
|
Traneptora
|
|
lonjil
Hm, I don't think that's right. Under ISO/IEC, MPEG-4 AVC is 14496 part 10, while H.264 is the ITU-T designation. JPEG1 is T.81, but it seems JXL doesn't have an ITU-T recommendation yet.
|
|
2024-02-16 11:27:11
|
it was an analogy, not a perfect one tho
|
|
|
MSLP
|
|
Traneptora
we all call it H.264 but that's like calling JXL "18181"
|
|
2024-02-16 11:32:21
|
h.264, h.265, h.266 is easier to remember for me which is newer than AVC, HEVC, VVC.
But you're right, the acronyms are more exciting, everytime I see "AVC" without any others codec of MPEG family mentioned i keep thinking it's h.265 not h.264, so this acronym has some excitement in it.
|
|
|
CrushedAsian255
|
|
MSLP
h.264, h.265, h.266 is easier to remember for me which is newer than AVC, HEVC, VVC.
But you're right, the acronyms are more exciting, everytime I see "AVC" without any others codec of MPEG family mentioned i keep thinking it's h.265 not h.264, so this acronym has some excitement in it.
|
|
2024-02-17 12:03:18
|
wait isn't HEVC h265?
|
|
|
MSLP
|
2024-02-17 12:34:02
|
yep, but I'm always wrong with "AVC"
|
|
|
CrushedAsian255
|
2024-02-17 12:41:21
|
oh
|
|
2024-02-17 01:28:07
|
i don't think apple support animated JXL
|
|
|
Kremzli
|
2024-02-17 01:29:42
|
My friend tested it and it did not animate
|
|
|
username
|
|
Kremzli
My friend tested it and it did not animate
|
|
2024-02-17 01:32:04
|
what device/OS? because it seems like JXL support differs between desktop/MacOS and mobile/iOS for some reason
|
|
|
Kremzli
|
2024-02-17 01:32:42
|
iphone 15 pro max latest user beta
|
|
2024-02-17 01:54:33
|
I'm gonna hopefully make them test on all their devices
|
|
|
username
|
2024-02-17 01:55:56
|
probably only need to test between mobile and desktop
|
|
|
Kremzli
|
2024-02-17 02:01:00
|
Might as well see if the tv can do it
|
|
2024-02-17 02:01:17
|
Idk how they will get the file but maybe it works
|
|
|
|
gbetter
|
2024-02-17 03:11:52
|
I had an older Intel iMac running macOS Ventura and *that* version of Safari (and ONLY Safari - no OS level support) would show animated JXL while my M2 laptop and iPhone running current OSs shows animated JXL as static images. Something about backporting the latest Safari with all new features to run on earlier macOSs must be different code. I no longer have old iMac so I can't test or prove this.
|
|
|
CrushedAsian255
|
2024-02-17 05:18:57
|
maybe the old Safari uses libjxl where the new Safari uses a custom thing
|
|
|
_wb_
|
|
lonjil
Hm, I don't think that's right. Under ISO/IEC, MPEG-4 AVC is 14496 part 10, while H.264 is the ITU-T designation. JPEG1 is T.81, but it seems JXL doesn't have an ITU-T recommendation yet.
|
|
2024-02-17 07:39:35
|
For some reason, JPEG never bothered to create an ITU version of jxl. Historically the ITU versions were useful because ITU makes their specs publicly available, but they stopped doing that when it's also an ISO standard, so it has become a bit of a pointless exercise to walk through the process twice...
|
|
|
Kremzli
|
2024-02-17 12:39:34
|
No animation on MacBook Pro M1 Pro 14 inch macOS Sonoma 14.3.1 or iPhone 15 Pro Max iOS 17.4 beta 3
|
|
2024-02-17 06:54:19
|
Did windows get native jxl support yet? (or ever)
|
|
|
lonjil
|
2024-02-17 07:06:25
|
It'll probably show up on the store as an installable codec in 2025
|
|
|
Traneptora
|
2024-02-18 10:30:27
|
https://github.com/Alex313031/thorium-libjxl/issues/18
|
|
|
jonnyawsom3
|
2024-02-18 11:03:10
|
> <@853026420792360980> @o-l-a-v @gz83 We fixed it, as they fixed it upstream. Now we are using libhighway 1.0.7, and libjxl 0.9.2.
|
|
2024-02-18 11:03:23
|
Not sure why I didn't expect it to ping
|
|
|
Traneptora
|
2024-02-18 11:03:35
|
ye
|
|
2024-02-18 11:03:41
|
that's what I was pointing out was that last comment
|
|
2024-02-18 11:03:52
|
didn't make that clear, sorry
|
|
|
Kremzli
|
|
Kremzli
No animation on MacBook Pro M1 Pro 14 inch macOS Sonoma 14.3.1 or iPhone 15 Pro Max iOS 17.4 beta 3
|
|
2024-02-19 06:08:27
|
Apple TV also is just no animation
|
|
|
fab
|
2024-02-19 06:50:43
|
Is this content suited for TikTok?
|
|
|
0
|
|
fab
Is this content suited for TikTok?
|
|
2024-02-19 10:31:23
|
ѕоlо glі zіngаrі dеmmеrdа ѕuоnаnо lа fіѕаrmоnіса
роі сhе саzzо tі ѕраmmі іn <#803574970180829194> bооmеr rіnсоglіоnіtо с'hаі l'аlzhеіmеr? nоn ѕі саріѕсе mаі сhе mіnсhіа ѕсrіvі
|
|
|
CrushedAsian255
|
|
Kremzli
No animation on MacBook Pro M1 Pro 14 inch macOS Sonoma 14.3.1 or iPhone 15 Pro Max iOS 17.4 beta 3
|
|
2024-02-19 10:39:50
|
should i submit feeback assistant apple jpeg xl?
|
|
|
Kremzli
|
2024-02-19 10:40:27
|
Idk sure
|
|
|
spider-mario
|
|
0
ѕоlо glі zіngаrі dеmmеrdа ѕuоnаnо lа fіѕаrmоnіса
роі сhе саzzо tі ѕраmmі іn <#803574970180829194> bооmеr rіnсоglіоnіtо с'hаі l'аlzhеіmеr? nоn ѕі саріѕсе mаі сhе mіnсhіа ѕсrіvі
|
|
2024-02-19 10:50:18
|
does this message also show as being from no one to anyone else?
|
|
2024-02-19 10:50:21
|
(weird sentence)
|
|
|
|
veluca
|
|
lonjil
|
2024-02-19 10:52:07
|
joined 2022, no messages since then
|
|
|
CrushedAsian255
|
2024-02-19 10:55:36
|
```json
{
"id": "448983994760167434",
"username": "",
"discriminator": "6560",
"avatar": "880f0f64e729da4a3a1e797ed093b3be",
}
```
seems like they had an empty username, and didn't switch to the new username system
|
|
|
username
|
|
CrushedAsian255
```json
{
"id": "448983994760167434",
"username": "",
"discriminator": "6560",
"avatar": "880f0f64e729da4a3a1e797ed093b3be",
}
```
seems like they had an empty username, and didn't switch to the new username system
|
|
2024-02-19 10:58:16
|
wait like not even a blank unicode char? just flat out nothing?
also I have a mutual server with them and they only have a single message in said other server
|
|
|
CrushedAsian255
|
2024-02-19 10:59:32
|
some weird character `U+E00AA`
|
|
2024-02-19 11:00:02
|
in an unallocated part of unicode??
|
|
2024-02-19 11:00:08
|
https://upload.wikimedia.org/wikipedia/commons/thumb/4/49/Roadmap_to_Unicode_SSP_multilingual.svg/1100px-Roadmap_to_Unicode_SSP_multilingual.svg.png
|
|
|
Olav
|
|
> <@853026420792360980> @o-l-a-v @gz83 We fixed it, as they fixed it upstream. Now we are using libhighway 1.0.7, and libjxl 0.9.2.
|
|
2024-02-20 03:29:14
|
It still doesn't work though ( https://github.com/Alex313031/thorium-libjxl/issues/18#issuecomment-1954423712 ), so it's not fixed upstream either?
|
|
|
username
|
2024-02-24 02:00:16
|
https://discord.com/channels/794206087879852103/805722506517807104/1210669016352428104
https://www.reddit.com/r/jpegxl/comments/1ay98yo/open_jpeg_xl_in_photopea/
|
|
|
HCrikki
|
2024-02-24 11:08:36
|
lots of webp and avif are passively served by large sites and cdns, and its natural many users want to convert em to another usable format.
in compareason, jxl is in a 'generate your own images' situation
|
|
2024-02-24 11:11:01
|
imo galleries small and large should offer jxl versions of images as downloadable files. its not necessary to extend that to an entire site, just the top images or curated galleries would suffice
|
|
|
monad
|
2024-02-25 06:41:56
|
'many' as in 'imperceptibly few'
|
|
|
VcSaJen
|
2024-02-26 01:05:06
|
webp support for uploads is abysmal, and that's a format that's been out for a decade and already deprecated by Google. If new formats keep getting deprecated that fast, JPEG/PNG will never die.
|
|
|
fab
|
|
VcSaJen
webp support for uploads is abysmal, and that's a format that's been out for a decade and already deprecated by Google. If new formats keep getting deprecated that fast, JPEG/PNG will never die.
|
|
2024-02-26 09:23:02
|
HDblog still uses,check HDblog.android and gives 90% reduction compared to SSIMULACRA 73
|
|
2024-02-26 09:23:21
|
I hope to test JPEG XL Next 2 wkks
|
|
|
Nova Aurora
|
|
VcSaJen
webp support for uploads is abysmal, and that's a format that's been out for a decade and already deprecated by Google. If new formats keep getting deprecated that fast, JPEG/PNG will never die.
|
|
2024-02-27 07:23:39
|
Deprecated?
|
|
|
fab
HDblog still uses,check HDblog.android and gives 90% reduction compared to SSIMULACRA 73
|
|
2024-02-27 07:23:47
|
Hdblog?
|
|
|
Kremzli
|
2024-02-28 06:04:04
|
https://community.signalusers.org/t/jpeg-xl-support/50331/
|
|
|
HCrikki
|
2024-02-28 09:20:32
|
jxl would be an easy win for any web service that primarly or exclusively serves data to their own apps (mobile or desktop)
|
|
|
Kremzli
|
2024-02-28 09:24:48
|
well jxl seems like an easy win for any service because of that transcoding feature
|
|
|
HCrikki
|
2024-02-28 09:35:48
|
imo jpegli too deserves more spotlight. itd be a way to win holdouts by sneakin improvements into their workflows that easen them into upgrading to jxl
|
|
|
Kremzli
|
2024-02-28 09:41:00
|
isnt it like a flagship feature everyone loves? (if you mean that jpegli = that transcoding feature because i think thats what it is anyway)
|
|
|
lonjil
|
2024-02-28 09:41:19
|
no, jpegli is just a normal jpeg encoder that happens to be very good
|
|
|
Kremzli
|
2024-02-28 09:41:46
|
but when you transcode back into jpeg, isnt it the same output as if you were to just use jpegli
|
|
|
HCrikki
|
2024-02-28 09:47:39
|
jpegli is arguably more efficient at encoding jpg than mozjpg. lots of the old jpgs thatd benefit from lossless conversion to jxl were generated with old codecs
|
|
2024-02-28 09:49:21
|
a single library combining jpegli+jxl (encoding and decoding) is uniquely handy
|
|
|
lonjil
|
|
Kremzli
but when you transcode back into jpeg, isnt it the same output as if you were to just use jpegli
|
|
2024-02-28 09:57:27
|
when you transcode back to jpeg, you get whatever jpeg you already had. jpegli isn't really relevant.
|
|
|
Kremzli
|
2024-02-28 09:59:14
|
So it goes back up in size?
|
|
|
lonjil
|
2024-02-28 09:59:46
|
what do you mean?
|
|
2024-02-28 10:00:24
|
if you have a JPEG, and transcode it to JXL, the JXL will be smaller than the original JPEG. When you transcode back to JPEG, you get the exact same JPEG back that you had originally. So the same size as the original JPEG.
|
|
|
Kremzli
|
2024-02-28 10:00:55
|
Why doesn't it use jpegli?
|
|
|
lonjil
|
2024-02-28 10:17:13
|
what do you mean by that?
|
|
2024-02-28 10:17:29
|
jpegli is a tool that takes raw pixels and turns them into a JPEG image
|
|
2024-02-28 10:17:58
|
jpeg to jxl transcoding is a process that transforms a JPEG into a JXL with zero loss of information.
|
|
|
Kremzli
|
2024-02-28 10:40:39
|
So it's entirely separate?
|
|
|
lonjil
|
|
Kremzli
|
2024-02-28 10:55:14
|
Alrighty then
|
|
|
w
|
2024-02-29 12:28:29
|
if you're gonna support jxl then transcoding back into jpg isn't a feature
|
|
2024-02-29 12:28:50
|
tbh I don't see the purpose of the transcoding back feature
|
|
|
Kremzli
|
2024-02-29 12:31:11
|
Well it's very useful for compatibility, and jxl itself has enough features to push for itself too. Not just to be an archival format
|
|
|
w
|
2024-02-29 12:32:23
|
It only works for jpegs and takes extra processing from the server. If an app supports jxl, it would never use that feature
|
|
2024-02-29 12:32:58
|
I mean the jxl back into jpg feature
|
|
|
Kremzli
|
2024-02-29 12:33:21
|
That extra processing is like nothing and jpeg is more common for me, it's still going to provide savings
|
|
|
w
|
2024-02-29 12:33:53
|
I get the jpg to jxl conversion but never will it need to go back into a jpg
|
|
|
Kremzli
|
2024-02-29 12:34:22
|
That's when everyone updates their stuff, and unfortunately that takes a long time
|
|
|
w
|
2024-02-29 12:34:36
|
if the app doesn't support jxl, jxls that aren't from JPGs still won't work
|
|
2024-02-29 12:35:04
|
the extra processing and effort comes from implementing it. It's not simple
|
|
2024-02-29 12:36:45
|
It's a very specific path, like what are you gonna do? Encode everything to jpg before into jxl?
|
|
2024-02-29 12:37:23
|
May aswell either support it all or not at all
|
|
2024-02-29 12:38:12
|
(usually how it's done since it's easier to not have broken code)
|
|
|
HCrikki
|
|
w
tbh I don't see the purpose of the transcoding back feature
|
|
2024-02-29 12:41:18
|
its a fallback so that web services can upgrade their images but clients and apps lacking jxl support so far can decode something without servers needing to store 2 copies of every image
|
|
2024-02-29 12:42:54
|
directly encoding jxl from the og/lossless original hq source gives the best filesize decrease, but for some itd make more sense still generating the best jpgs then doing a lossless transcode to jxl (with guaranteed 20-25% filesize reduction), serving both
|
|
|
w
|
2024-02-29 12:43:09
|
Yeah it's a feature only to be used by the cdns
|
|
2024-02-29 12:44:59
|
like for an app like signal, not a real feature
|
|
|
HCrikki
|
2024-02-29 12:45:00
|
a shame cloudflare doesnt seem to do that out of the box. i recall seeing them activate that per account upon asking so maybe it could be made default sometime
|
|
2024-02-29 12:46:27
|
for embeds, thumbnails and existing images for sure itd be best doing straight jxl encodes from originals but most images a user would share would be jpgs
|
|
2024-02-29 12:47:45
|
conversions would need to run either at a server or a user' device and imo itd be neater for apps to handle that during upload for single images unless youre sharing a batch
|
|
|
w
|
2024-02-29 12:48:12
|
cdns never had an issue transcoding to other formats for compatibility in the first place
|
|
2024-02-29 12:48:32
|
the jpeg thing is extra and specific
|
|
|
VcSaJen
|
|
Nova Aurora
Deprecated?
|
|
2024-03-01 03:45:28
|
not quite, but it's still not considered "current gen" by Google & Co
|
|
|
username
|
2024-03-02 03:20:43
|
seems like there's a pull request for blender to add JPEG XL support however it seems to rely on a currently unmerged pull request on OpenImageIO
https://projects.blender.org/blender/blender/pulls/118989
|
|
2024-03-02 03:21:42
|
the currently unmerged pull request for OpenImageIO seems pretty close to being merged https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/4055#pullrequestreview-1910339267
|
|
|
Orum
|
2024-03-02 03:23:10
|
I know there has been a feature request to add it for a long time, but it's nice to see a pull request too <:Hypers:808826266060193874>
|
|
|
username
|
2024-03-02 03:24:15
|
there is a concern I have... does any of this handle depth/Z channels?
also pinging <@238552565619359744> since I remember him talking about the possibility/idea of Blender adding JXL support a couple of times in the past.
|
|
|
Orum
|
2024-03-02 03:26:02
|
you mean as a separate channel?
|
|
2024-03-02 03:26:27
|
honestly it's probably a lot more hassle than it's worth, but it could support it in theory
|
|
|
username
|
2024-03-02 03:27:18
|
don't other formats in Blender already handle depth channel saving and such already?
|
|
|
Orum
|
2024-03-02 03:27:39
|
🤷♂️ I always handle depth images separately
|
|
2024-03-02 03:28:24
|
so yes, it's possible, but trying to embed it as a separate channel (in most image formats) is a fool's errand
|
|
|
jonnyawsom3
|
2024-03-02 03:28:44
|
It should 'just' be a case of passing the depth buffer as a greyscale extra channel
|
|
2024-03-02 03:29:05
|
And thanks for the ping, on a train in London with nothing else to do
|
|
|
username
|
2024-03-02 03:29:59
|
I'm bringing this depth channel stuff up because doesn't JXL already have a channel dedicated to depth? like it feels like a waste to not use it in any form
|
|
|
Orum
|
2024-03-02 03:30:06
|
I think it would work in tif, and maybe exr?
|
|
2024-03-02 03:30:28
|
JXL just offers you a ton of channels and you can define them to be whatever you want IIRC
|
|
2024-03-02 03:31:15
|
> Up to 4099 channels. Main channels: either one channel for grayscale, three channels for RGB, or four channels for CMYK. The rest of the channels are optional and can be used to store alpha (either "straight" or "premultiplied"), depth, or thermal data.
|
|
|
jonnyawsom3
|
|
username
I'm bringing this depth channel stuff up because doesn't JXL already have a channel dedicated to depth? like it feels like a waste to not use it in any form
|
|
2024-03-02 03:31:21
|
Yeah, there's dedicated depth and thermal channels
|
|
2024-03-02 03:31:50
|
As said above
|
|
|
Orum
|
2024-03-02 03:32:37
|
of course getting software to handle so many channels is a feat in itself
|
|
|
username
|
2024-03-02 03:38:31
|
the only ones that really matter in the case of Blender / 3D rendering in general would be Alpha and Depth (maybe also Thermal in some cases) and if Blender came out of the door with it's JXL support having default saving of depth maps then it could set an example for any other professional software that handles images and might even bring JXL as being better known in such spaces
|
|
2024-03-02 03:40:17
|
when you export a PNG in Blender it already tries to save as much data to the file as it can including a bunch of metadata about the scene and how it was rendered (although strangely I can only find tools that can read it but not copy it)
|
|
|
Orum
|
2024-03-02 03:41:09
|
Sure, it would be nice to support it, but even if it does I don't think it'll do much to encourage the rest of the industry. I'd love to be wrong though. <:CatSmile:805382488293244929>
|
|
|
jonnyawsom3
|
2024-03-02 03:50:23
|
I mean, no downside to using the options specifically designed for it ;P
|
|
|
Orum
|
2024-03-02 03:51:38
|
well the main downside is when you export from blender to import into something else, and it has no idea what to do with that channel so it just ignores it... but blender will surely let you export the depth map as a separate image in such a scenario
|
|
|
username
|
2024-03-02 03:52:39
|
well the same thing would apply to the special metadata that Blender adds to PNGs
|
|
|
jonnyawsom3
|
2024-03-02 03:53:13
|
The solution there is just make it optional, which it already would be, and then you can export it separately
|
|
|
Orum
|
2024-03-02 03:57:08
|
Blender adds special metadata to PNGs?
|
|
|
username
|
2024-03-02 03:58:25
|
yes Blender adds a bunch of details about the scene and how it was rendered however when messing around exiftool can only read it but can't write it for some reason
|
|
|
Orum
|
2024-03-02 03:59:50
|
interesting... I converted most of my renders to webp in the past so I'm sure that got stripped out, but now I'll be using JXL so it'll be interesting to see if that survives transcoding
|
|
|
username
|
2024-03-02 04:04:04
|
huh I thought it saved way more info but anyway heres what it saves
|
|
|
Orum
|
2024-03-02 04:08:11
|
ah, so next to nothing
|
|
|
VcSaJen
|
|
username
the only ones that really matter in the case of Blender / 3D rendering in general would be Alpha and Depth (maybe also Thermal in some cases) and if Blender came out of the door with it's JXL support having default saving of depth maps then it could set an example for any other professional software that handles images and might even bring JXL as being better known in such spaces
|
|
2024-03-02 05:44:11
|
What about normal map, specular map, etc?
(Edit: ah, you said rendering, not textures)
|
|
|
jonnyawsom3
|
2024-03-03 02:07:57
|
Textures are just individual anyway
|
|
2024-03-03 02:08:43
|
And yeah, blender just embeds basic render statistics which you can overlay on the image itself instead
|
|
|
Quackdoc
|
2024-03-03 12:56:37
|
huh, neither gthumbs nor gimp supports jxl's converted ICC profiles. Tested with https://webkit.org/blog-files/color-gamut/Flowers-ProPhoto.jpg `cjxl Flowers-ProPhoto.jpg --lossless_jpeg=0 -d 1 test.jxl`
|
|
2024-03-03 01:00:26
|
prophoto and adobergb source images both fail
|
|
|
spider-mario
|
2024-03-03 01:07:46
|
cjxl _doesn’t_ convert it when I try
|
|
2024-03-03 01:07:49
|
it keeps it as is
|
|
2024-03-03 01:07:57
|
and it works in GIMP
|
|
2024-03-03 01:08:12
|
```console
$ tools/jxlinfo test.jxl
JPEG XL file format container (ISO/IEC 18181-2)
JPEG XL image, 1200x795, lossy, 8-bit RGB
Color space: 940-byte ICC profile, CMM type: "KCMS", color space: "RGB ", rendering intent: 0
Brotli-compressed Exif metadata: 25755 compressed bytes
```
|
|
2024-03-03 01:08:15
|
what do you get?
|
|
|
Quackdoc
|
2024-03-03 01:41:37
|
```ps
jxlinfo test.jxl
JPEG XL file format container (ISO/IEC 18181-2)
JPEG XL image, 1200x795, lossy, 8-bit RGB
Color space: RGB, Custom, white_point(x=0.345705,y=0.358540), Custom primaries: red(x=0.734698,y=0.265302), green(x=0.159600,y=0.840399), blue(x=0.036597,y=0.000106)gamma(0.555315) transfer function, rendering intent: Perceptual
Brotli-compressed Exif metadata: 25755 compressed bytes
```
|
|
2024-03-03 01:42:33
|
using `cjxl v0.10.0 19bcd827 [AVX2,SSE4,SSE2]` from arch's repo
|
|
2024-03-03 01:57:27
|
this is the encoded file
|
|
|
spider-mario
|
2024-03-03 02:03:42
|
thanks
|
|
2024-03-03 02:03:48
|
it does seem to work for me in GIMP
|
|
2024-03-03 02:04:07
|
but it’s weird that my cjxl doesn’t convert the color profile into an enum version
|
|
|
Quackdoc
|
2024-03-03 02:06:16
|
it works in gimp? maybe my versions are wack
|
|
2024-03-03 02:06:28
|
someone else reported this issue on phoronix though
|
|
|
spider-mario
|
|
spider-mario
but it’s weird that my cjxl doesn’t convert the color profile into an enum version
|
|
2024-03-03 02:07:27
|
oh… it does if I build with skcms instead of lcms2… 🥹
|
|
|
Quackdoc
|
|
Adrià
|
2024-03-04 01:32:15
|
JPEG XL support for dlib (C++ toolkit for ML, data analysis, etc) just got merged https://github.com/davisking/dlib/pull/2917
|
|
|
sklwmp
|
2024-03-04 01:41:04
|
> Moreover, I intend to use JXL for all my datasets from now on, since I will save about 20-40% of disk space with reversible pixel perfect conversion from JPEG and PNG. All that while decoding images faster than libjpeg-turbo.
<:BlobYay:806132268186861619>
|
|
2024-03-04 01:42:10
|
although, lossless jxl can still decode kinda slowly, especially compared to PNG
|
|
2024-03-04 01:42:14
|
i have a lot of large TIFF scans losslessly compressed, so this really bugs me sometimes
|
|
|
_wb_
|
2024-03-04 02:07:58
|
single-core, it's hard to beat PNG decode speed since that's basically just gunzip
|
|
|
jonnyawsom3
|
|
sklwmp
> Moreover, I intend to use JXL for all my datasets from now on, since I will save about 20-40% of disk space with reversible pixel perfect conversion from JPEG and PNG. All that while decoding images faster than libjpeg-turbo.
<:BlobYay:806132268186861619>
|
|
2024-03-04 04:00:42
|
> Done. If I (or someone) have the need for more JPEG XL features at some point, I will add them:
>
> grayscale_alpha_pixel: it would require a new pixel type
> storing the image in floating point: it would require rgb_float, rgb_alpha_float
> saving and loading animations: overload the load and save functions with std::vector
> transcoding from and to JPEG: this might be useful to convert imglab datasets.
|
|
2024-03-04 04:01:06
|
So "reversible pixel perfect conversion from JPEG" isn't quite right yet
|
|
2024-03-04 04:01:51
|
Unless they meant converting all their old images externally instead
|
|
|
Oleksii Matiash
|
2024-03-04 09:27:42
|
Iridient Developer (raw converter) added indirect support: Adobe DNG: Support for additional DNG formats including all new JPEG XL compression introduced in DNG v1.7.1 plus floating point and 32-bit integer formats.
|
|
|
CrushedAsian255
|
2024-03-04 09:33:28
|
Does anyone have an idea for a document icon for jxl files?
|
|
|
jonnyawsom3
|
|
Oleksii Matiash
Iridient Developer (raw converter) added indirect support: Adobe DNG: Support for additional DNG formats including all new JPEG XL compression introduced in DNG v1.7.1 plus floating point and 32-bit integer formats.
|
|
2024-03-04 09:33:48
|
https://www.iridientdigital.com/products/rawdeveloper_history.html#:~:text=JPEG%20XL
|
|
|
CrushedAsian255
Does anyone have an idea for a document icon for jxl files?
|
|
2024-03-04 09:34:24
|
Could use the SVG that spins in the corner https://jpegxl.info/
|
|
|
Adrià
|
|
sklwmp
> Moreover, I intend to use JXL for all my datasets from now on, since I will save about 20-40% of disk space with reversible pixel perfect conversion from JPEG and PNG. All that while decoding images faster than libjpeg-turbo.
<:BlobYay:806132268186861619>
|
|
2024-03-05 01:09:38
|
Thank you, most of my ML datasets are JPEG images, so I haven't been affected by the lossless decoding speed. But I'll keep that in mind.
|
|
|
Oleksii Matiash
|
|
https://www.iridientdigital.com/products/rawdeveloper_history.html#:~:text=JPEG%20XL
|
|
2024-03-05 07:15:26
|
I received announcement email, yes 🙂
|
|
|
jonnyawsom3
|
|
Oleksii Matiash
I received announcement email, yes 🙂
|
|
2024-03-05 03:23:30
|
Thought I'd add a link for people wanting to see themselves
|
|
2024-03-07 12:34:17
|
Very soon... Although I think we could help share info for the streaming they wanted too https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/4055
|
|
|
Quackdoc
|
|
Very soon... Although I think we could help share info for the streaming they wanted too https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/4055
|
|
2024-03-07 12:35:12
|
i've been waiting for this 0.0
|
|
|
lonjil
|
2024-03-09 12:44:40
|
I wonder what the timeline of JXL being added to the PDF spec is
|
|
|
HCrikki
|
2024-03-09 03:47:45
|
telegram desktop 4.15.1+ reportedly supports jxl on *macos* now (integrating 0.8.2 iinm). already did in the lin/win clients
|
|
2024-03-09 03:48:43
|
commit mention jxl and avif for macos, oddly the 4.15.1 release announcement just says 'telegram business features'
|
|
|
lonjil
|
2024-03-09 01:14:18
|
can someone confirm the following from reddit, about Windows support for JXL?
> it’s already in the insider, but partial
> thumbnails show, sometimes, you can set it as a wallpaper, from the slide shows, it’s beta, give it time to hit RC
|
|
|
|
veluca
|
2024-03-09 01:28:38
|
unfortunately I am not home for the weekend... but if nobody does so by tomorrow evening, for sure I will check
|
|
|
HCrikki
|
2024-03-09 01:36:32
|
mentioned before, no idea if any changes https://discord.com/channels/794206087879852103/803574970180829194/1201015191492640818
|
|
2024-03-09 01:37:06
|
https://images-ext-1.discordapp.net/external/qLgZ0LNdiXKAVd9Qc83swAMZ2racyswnoJ5tGXkvKgo/https/pbs.twimg.com/media/GEymaR2WUAAbZIO.png%3Alarge?format=webp
|
|
|
lonjil
|
2024-03-09 01:38:05
|
that was just a header and some stuff, no one reported actual support at the time. And we speculated that it would be something you had to install from the windows store (just like with some other codecs).
But if it is now generated thumbnails, that is a major indicator that it will be supported by default in base installs, which is big.
|
|
|
Tirr
|
2024-03-09 01:45:05
|
I'm updating my VM to build 26058
|
|
|
Tirr
it's `%SystemRoot%\system32\MSRAWImage.dll`
|
|
2024-03-09 02:44:25
|
this changed to `%SystemRoot%\system32\windowscodecs.dll`
|
|
2024-03-09 02:44:34
|
so it's builtin
|
|
2024-03-09 02:44:52
|
or, it will be builtin
|
|
2024-03-09 02:45:07
|
it doesn't load anything yet for now it seems
|
|
2024-03-09 02:47:26
|
latest build in canary channel is 26063 though
|
|
2024-03-09 02:52:15
|
ok I'm going to switch to canary channel to see if something changed
|
|
|
|
afed
|
2024-03-09 02:56:05
|
still not working as a wic decoder?
|
|
|
Tirr
|
2024-03-09 02:57:01
|
yeah I've tried photoviewer.dll, didn't work
|
|
|
lonjil
|
|
Tirr
this changed to `%SystemRoot%\system32\windowscodecs.dll`
|
|
2024-03-09 02:58:26
|
do you happen to know which codecs are in that dll?
|
|
|
Tirr
|
2024-03-09 03:00:34
|
nope, the only thing I know is that it's a mixture of various codecs linked together
|
|
|
lonjil
|
2024-03-09 03:01:16
|
I guess I'll install Win11 in a VM as well and snoop
|
|
2024-03-09 03:02:12
|
How is Windows these days? Ought I install a specific edition for it to not be annoying?
|
|
|
w
|
2024-03-09 03:02:58
|
education edition always
|
|
2024-03-09 03:03:33
|
but also because everyone gets a free education key
|
|
2024-03-09 03:12:55
|
im on 26063 and i dont see anything
|
|
|
|
afed
|
2024-03-09 03:14:46
|
for thumbnails and wallpaper, too?
|
|
|
w
|
2024-03-09 03:16:01
|
nothing
|
|
|
|
afed
|
2024-03-09 03:17:08
|
<:FeelsReadingMan:808827102278451241>
https://gist.github.com/sayurin/837e69f1f9ee43b7ff7a
|
|
2024-03-09 03:18:42
|
```WIC: Decoder
BMP Decoder
GIF Decoder
ICO Decoder
CUR Decoder
JPEG Decoder
PNG Decoder
TIFF Decoder
DNG Decoder
WMPhoto Decoder
DDS Decoder
Microsoft HEIF Decoder
Microsoft Webp Decoder
Microsoft Raw Image Decoder
Microsoft JPEG-XL Decoder
WIC: Encoder
BMP Encoder
GIF Encoder
JPEG Encoder
PNG Encoder
TIFF Encoder
WMPhoto Encoder
DDS Encoder
Microsoft HEIF Encoder
Microsoft JPEG-XL Encoder```
JPEG**-**XL <:PepeHands:808829977608323112> **-** not JPEG XL
though it's `revised this gist on Feb 7, 2023.`
|
|
|
MSLP
|
2024-03-09 03:19:24
|
so it's JPEG minus XL
|
|
|
Tirr
|
2024-03-09 03:19:55
|
it's written correctly as JPEG XL in the registry though
|
|
2024-03-09 03:24:18
|
it's JPEG-XL in build 22621
|
|
|
w
|
2024-03-09 03:25:22
|
ran it myself
|
|
2024-03-09 03:25:42
|
compile command `cl mftdumper.cpp /std:c++latest /LIBPATH:"C:\Program Files (x86)\Windows Kits\10\Lib\10.0.22621.0\um\x64" /link "ole32.lib"`
|
|
|
|
afed
|
2024-03-09 03:30:51
|
<:Poggers:805392625934663710>
```WIC: Decoder
Microsoft JPEG XL Decoder
WIC: Encoder
Microsoft JPEG XL Encoder
WIC: MetadataReader
JPEG XL Animation Properties Reader
JPEG XL Animation Frame Reader
WIC: MetadataWriter
JPEG XL Animation Properties Writer
JPEG XL Animation Frame Writer```
<:monkaMega:809252622900789269>
```WIC: PixelFormat
32bpp HDR10```
|
|
|
w
|
2024-03-09 03:32:18
|
i had jxl winthumb installed until a moment ago and I don't know if it's fully out
|
|
2024-03-09 03:32:25
|
and if that can affect that
|
|
|
|
afed
|
2024-03-09 03:42:29
|
even so, I don't think it would be **Microsoft** JPEG XL
|
|
|
w
|
2024-03-09 03:46:41
|
the new things are the metadatareader, don't know if that's real
|
|
|
Quackdoc
|
|
w
i had jxl winthumb installed until a moment ago and I don't know if it's fully out
|
|
2024-03-09 03:58:25
|
you can always reinstall it and see if you have a new entry
|
|
|
lonjil
|
2024-03-09 04:21:29
|
I remembered that I have a random win 11 pro key just lying around, so I'm currently installing the latest canary build
|
|
|
username
|
|
lonjil
can someone confirm the following from reddit, about Windows support for JXL?
> it’s already in the insider, but partial
> thumbnails show, sometimes, you can set it as a wallpaper, from the slide shows, it’s beta, give it time to hit RC
|
|
2024-03-09 05:04:33
|
I wonder if this would let you set the wallpaper directly through the context menu?
|
|
|
lonjil
|
2024-03-09 05:18:05
|
I tried a slideshow and it seems to not work
|
|
2024-03-09 05:26:04
|
I left a reply saying we can't get it to work, and asking exactly which build they are on
|
|
|
jonnyawsom3
|
2024-03-09 06:43:16
|
The same person made another comment saying they saw the wallpaper support in that old changelog we talked about thinking it was a typo for JXR
|
|
2024-03-09 06:43:40
|
So they could just be assuming, even if it would be strange to be so confident about it
|
|
|
Traneptora
|
2024-03-09 07:12:02
|
JXR is already in windows though
|
|
2024-03-09 07:12:47
|
and has been since winvista iirc
|
|
|
username
|
2024-03-09 07:13:41
|
wasn't available to use for HDR wallpapers until recently though
|
|
|
jonnyawsom3
|
2024-03-09 07:13:58
|
It was a bugfix
> Fixed an issue where it wasn’t possible to select .JXL files for your wallpaper slideshow.
|
|
|
sklwmp
|
2024-03-09 07:41:58
|
https://x.com/Kampidh/status/1764941071446909000?s=20
|
|
|
username
|
2024-03-09 07:42:46
|
last I checked Bluesky ruins images even worse then twitter....
|
|
|
sklwmp
|
2024-03-09 07:43:52
|
i haven't checked myself, but, that seems... bad
|
|
|
username
|
2024-03-09 07:45:08
|
I managed to get Bluesky to make a profile picture **315.8** **times** larger in file size (yes actually)
|
|
2024-03-09 07:45:50
|
for some reason their image handling enforces resizing of images to 1000x1000 or 2000x2000
|
|
2024-03-09 07:45:55
|
it's pretty bad..
|
|
|
sklwmp
|
2024-03-09 07:50:16
|
i wonder what would happen if you uploaded jxl art
|
|
|
username
|
2024-03-09 07:50:53
|
oh god, it would probably be decoded to pixels and then resized to 1000x1000 or 2000x2000 and then made a JPEG
|
|
|
jonnyawsom3
|
2024-03-09 08:02:14
|
Last I saw it allows upload but just serves a png
|
|
2024-03-09 08:02:24
|
It might have changed though
|
|
|
HCrikki
|
2024-03-10 12:24:28
|
updated adoption for mihon/tachiyomi. they updated libjxl from ancient 0.7.0 to 0.10.0 (nightly in mihon, not yet in stable or derivatives)
|
|
2024-03-10 12:27:11
|
the much faster enc/dec would be good for downloading chapters to read later as jpg->jxl, completely lossless yet consuming guaranteed 20% less storage but so far it seems theres only decoding
|
|
|
Quackdoc
|
2024-03-10 12:38:08
|
having mihon compress the files would be so nice, I know it was asked before tachiyomi did the dead
|
|
|
HCrikki
|
2024-03-10 12:46:10
|
conversion is normally out of scope for a reader, only jxl is in unique position to store local copies in better than lossless quality on the fly
|
|
|
lonjil
|
2024-03-10 12:54:07
|
I mean, it's not just a reader. It does stuff like downloading and has options like packing the images into cbz archives. Doesn't seem too far out of scope.
|
|
|
w
|
2024-03-10 04:24:09
|
the 0.10 isnt in-in yet and I am cooking encoding
|
|
2024-03-10 04:35:51
|
and so far jxl is extremely slow
|
|
|
Orum
|
2024-03-10 05:02:04
|
0.10 makes a big difference in speed
|
|
|
w
|
2024-03-10 05:18:51
|
on single core?
|
|
2024-03-10 05:19:19
|
small device hates it when all the cpu is being used
|
|
|
Orum
|
2024-03-10 05:25:15
|
oh, maybe not that
|
|
2024-03-10 05:25:37
|
I don't know anyone that even has a single core CPU any more...
|
|
|
w
|
2024-03-10 05:28:08
|
6 core phone cpu
|
|
2024-03-10 05:28:18
|
use even 4 of them and the phone cant do anything
|
|
|
Orum
|
2024-03-10 05:29:34
|
sounds like a scheduling problem
|
|
2024-03-10 05:29:43
|
but try a lower effort instead then?
|
|
|
w
|
2024-03-10 05:31:03
|
i only tested with effort 7 and 6
|
|
2024-03-10 05:31:11
|
but the sluggishness can be felt even with just decoding
|
|
|
Quackdoc
|
2024-03-10 05:31:53
|
I havent benched my s9+ yet, I should do so soon
|
|
|
Orum
I don't know anyone that even has a single core CPU any more...
|
|
2024-03-10 05:32:00
|
I do [av1_dogelol](https://cdn.discordapp.com/emojis/867794291652558888.webp?size=48&quality=lossless&name=av1_dogelol)
|
|
|
w
|
2024-03-10 05:32:05
|
i'm testing on my s10e
|
|
|
Tirr
|
2024-03-10 05:33:26
|
yeah s10e is... slow
|
|
2024-03-10 05:33:44
|
goes almost dying with multiple heif in gallery
|
|
|
HCrikki
|
2024-03-10 05:33:52
|
how are you decoding?
|
|
|
w
|
2024-03-10 05:33:55
|
it's still a flagship that's faster than 80% of all phones
|
|
|
HCrikki
how are you decoding?
|
|
2024-03-10 05:35:50
|
pretty much the same as decode_oneshot
|
|
|
Quackdoc
|
|
Tirr
goes almost dying with multiple heif in gallery
|
|
2024-03-10 05:38:27
|
heif can often just be ridiculously expensive to decode sometimes,
|
|
|
w
|
2024-03-10 05:47:51
|
decoding also noticably slower than other formats on a pixel 7
|
|
|
VcSaJen
|
|
lonjil
How is Windows these days? Ought I install a specific edition for it to not be annoying?
|
|
2024-03-10 08:50:18
|
With Enterprise Edition you have full control over Windows.
|
|
|
w
it's still a flagship that's faster than 80% of all phones
|
|
2024-03-10 09:03:42
|
Exynos 9820 is comparable to mid-range nowadays
|
|
|
lonjil
|
2024-03-10 09:48:56
|
is there a good app for testing jxl on android?
|
|
2024-03-10 09:49:10
|
I have a midrange with an older SoC
|
|
|
Lock
|
|
lonjil
is there a good app for testing jxl on android?
|
|
2024-03-10 09:51:16
|
i can only think of ffmpeg with termux lol
|
|
|
lonjil
|
|
HCrikki
|
2024-03-10 10:07:49
|
android cromite if its websites youre testing
|
|
|
Lock
|
2024-03-10 10:09:38
|
these fucking browser names being all chemicals
|
|
|
HCrikki
|
2024-03-10 10:25:33
|
integrate awxkee's xl decoder for coil or glide in your fave app or inhouse one otherwise
|
|
|
Demiurge
|
|
Tirr
goes almost dying with multiple heif in gallery
|
|
2024-03-10 10:26:24
|
Maybe it's using hardware decoding, which can only do one image at a time in series since the hardware is designed to decode video frames in series?
|
|
|
sklwmp
|
|
Lock
these fucking browser names being all chemicals
|
|
2024-03-10 11:38:41
|
well they are all based on "chromium"
|
|
|
jonnyawsom3
|
|
Very soon... Although I think we could help share info for the streaming they wanted too https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/4055
|
|
2024-03-10 08:22:25
|
> Now that we pass the imageinout_test, I believe that we are ready for a merge if you fix the clang format issue! Then we can come back to beef up this format with other improvements.
|
|
|
username
|
|
> Now that we pass the imageinout_test, I believe that we are ready for a merge if you fix the clang format issue! Then we can come back to beef up this format with other improvements.
|
|
2024-03-10 08:42:31
|
> Then we can come back to beef up this format with other improvements.
ooo I wonder if this is in reference to adding support for other channel types like depth?
|
|
|
jonnyawsom3
|
2024-03-10 08:58:16
|
<https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/4055#issuecomment-1986723143>
> Apologies if this isn't exactly the right time or place to ask this, But are there any plans to support depth/z channels and maybe even thermal channels? (could possibly be in a later pull request since this one is just "Initial" support?). This would be very useful for projects like Blender to have depth/z maps/data stored/saved in rendered images for example.
>
> libjxl does have support for handling these channel types and there's even a JXL file with such channels here as an example: https://saklistudio.com/jxltests/suz-multichannel-800px.jxl (RGB + Alpha (empty) + Depth + Thermal)
>
> I'm really not too familiar with OpenImageIO as a whole so maybe this would be out of scope? although it seems like the EXR support/code has depth/z channel handling in some form via z_channel? unsure, though I do see a few references to z_channel in the docs.
> OpenImageIO can deal with any number of channels having any names. So yes, totally within scope and nothing special at all.
>
> z_channel and alpha_channel are just a special designations in the ImageSpec to mark it for ease of clients of the library so you don't need to go searching for them every time. But it's in no way implies that those are the only channels you can have.
|
|
2024-03-10 09:00:17
|
Blender Stated this
> Logistics:
> The platform maintainers first need to decide if they're willing to accept a new dependency on libjxl because we'll need precompiled libraries and build system support for all the platforms if we decide to support the format at all.
>
> Then the next opportunity to update our library dependencies, for OIIO and libjxl, is around the transition from bcon1 to bcon2 scheduled for May 1st, so that would be the earliest that a commit on Blender's side could be made. This assumes OIIO has made an official release by then.
|
|
2024-03-10 09:00:45
|
OIIO releases monthly, so JXl should make it if nothing comes up
|
|
|
Quackdoc
|
2024-03-10 09:03:29
|
good OIIO is really the nifty tool
|
|
2024-03-10 09:03:53
|
it will be my goto for tiff to jxl probably
|
|
|
lonjil
|
|
lonjil
I left a reply saying we can't get it to work, and asking exactly which build they are on
|
|
2024-03-12 01:59:51
|
they did not respond in any useful way, so I assume they're mistaken.
|
|
|
HCrikki
|
2024-03-13 02:38:31
|
just updated to win11 pro insider dev **26058.1100** (releases 14 february - more exactly its **build 26058.ge_release.240209-1555**) and registry lists entries for microsoft jpegxl as system components (no prior support or addon used)
|
|
2024-03-13 02:39:10
|
trying to change wallpaper, it appears there **both** JXR and JXL
|
|
2024-03-13 02:39:35
|
anyone can check on their end with a pure out of box install ?
|
|
|
Traneptora
|
|
HCrikki
anyone can check on their end with a pure out of box install ?
|
|
2024-03-13 02:54:26
|
how would I install an insider pro dev build out of the box?
|
|
|
HCrikki
|
2024-03-13 02:54:57
|
install a fresh iso straight from ms or any other source
|
|
2024-03-13 02:55:12
|
i converted my vm install using a script
|
|
|
Traneptora
|
2024-03-13 02:55:15
|
but those typically aren't dev builds, right?
|
|
2024-03-13 02:55:29
|
a fresh ISO that is an insider dev build?
|
|
|
HCrikki
|
2024-03-13 02:55:43
|
canary is more experimental, dev already stabilized code
|
|
|
Traneptora
|
2024-03-13 02:56:06
|
no, I meant more like
|
|
2024-03-13 02:56:16
|
how can I install an insider build fresh from ISO
|
|
2024-03-13 02:56:26
|
I thought you had to enable the insider program in the settings and then run wupdate
|
|
2024-03-13 02:57:00
|
> Hello Windows Insiders, today we are releasing Windows 11 Insider Preview Build 26058 to the Canary and Dev Channels. We are releasing ISOs for this build – they can be downloaded here.
|
|
2024-03-13 02:57:02
|
Never mind
|
|
|
HCrikki
|
2024-03-13 02:58:13
|
just be warned its not recommended or possible to return to a more stable release branch without a clean reinstall, in case you used an existing vm for important tinkering
|
|
|
Traneptora
|
2024-03-13 03:02:12
|
I'm going to set up a new VM because the question was about a fresh install
|
|