[ / / / / / / / / / / / / / ] [ dir / random / bane / clang / hentai / htg / loomis / mde / pone / rule34 ]

/hydrus/ - Hydrus Network

Bug reports, feature requests, and other discussion for the hydrus network.
Name
Email
Subject
Comment *
File
Password (Randomized for file and post deletion; you may also set your own.)
Archive
* = required field[▶ Show post options & limits]
Confused? See the FAQ.
Embed
(replaces files and can be used instead)
Voice recorder Show voice recorder

(the Stop button will be clickable 5 seconds after you press Record)
Options

Allowed file types:jpg, jpeg, gif, png, webm, mp4, swf, pdf
Max filesize is 16 MB.
Max image dimensions are 15000 x 15000.
You may upload 5 per post.


New user? Start here ---> http://hydrusnetwork.github.io/hydrus/

Experienced user with a bit of cash who wants to help out? ---> Patreon

Current to-do list has: 2,017 items

Current big job: Catching up on Qt, MPV, tag work, and small jobs. New poll once things have calmed down.


HookTube embed. Click on thumbnail to play.

ac1d61  No.8583

windows

zip: https://github.com/hydrusnetwork/hydrus/releases/download/v302/Hydrus.Network.302.-.Windows.-.Extract.only.zip

exe: https://github.com/hydrusnetwork/hydrus/releases/download/v302/Hydrus.Network.302.-.Windows.-.Installer.exe

os x

app: https://github.com/hydrusnetwork/hydrus/releases/download/v302/Hydrus.Network.302.-.OS.X.-.App.dmg

tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v302/Hydrus.Network.302.-.OS.X.-.Extract.only.tar.gz

linux

tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v302/Hydrus.Network.302.-.Linux.-.Executable.tar.gz

source

tar.gz: https://github.com/hydrusnetwork/hydrus/archive/50dc4f82cb55ae21b62d1e41a2bcc169b82b1d41.tar.gz

I had a great week. I've finished up the new 'simple downloader' and wrote a new way for subscriptions to run simultaneously.

simple downloader improvements

I have rounded out the new simple downloader, which replaced the old 'page of images' downloader last week. Its parsers have more descriptive names and it can import and export parsers for sharing. And it remembers which parser it had selected through a session restart.

I also reintroduced the old 'all files linked by images in a page' behaviour from the old page of images downloader. This parse method, which fetches anything linked to by an image in the entered URL, is a useful catch-all for threads from as-yet unsupported imageboards and other miscellaneous tasks.

I also added new parsers for mewch threads, webmshare file pages, and imgur image albums. The imgur album parser only works for the first page of results (about ten images) right now–we'll have to wait for the cleverer gallery parser to get full support.

simultaneous subscriptions

Subscriptions can now run in parallel! The new setting for how many can run at once is under options->downloading. I have kept the default as 1, so you will not see any change if you do not touch anything, but if you would like to bump it up to 2 or 3, please give it a go and let me know how it works for you.

This is new code, so I recommend not turning it up to 11 yet. I've improved how some of the bandwidth calculations work here, but I can't promise a very heavy situation will not result in a traffic jam.

misc

After a long time of hating the name, I have changed 'explicit tags' to 'additional tags'.

If you like, you can now set media to scale to 100% by default, even when the media is larger than the media or preview canvas size.

Some unusual searches that were formerly not returning any results (like 'akaiha_(akaihasugk)' on danbooru) should now be fixed. This was due to bad site-side markup not parsing correctly clientside. Users who run from source may like to check to see if they have 'html5lib' installed, which is the new parser I now prefer above the old 'lxml'.

new html parsing options

This is for advanced users getting into the new parsing system only!

To get the old 'all files linked by images in a page' behaviour working again, I had to update the new html parsing formulae to walk up the tree (ascending) as well as search down it (descending). So in any normal parsing step, as well as saying 'find all the img tags', you can now say 'walk back up the tree until you hit an 'a' tag'!

Also, you can now set a descending rule without an explicit tag name, so you can say 'find all tags with class="thumb"' without having to specify div or span or whatever. If you are a mad man, you can also just search for 'all tags' without any qualifiers, wew.

full list

- improved how simple downloader parsing formulae are stored and passed around

- the edit simple downloader parsing formulae panel now has an 'add defaults' menu button

- the simple downloader formula edit panel handles its buttons better and can now do import from/export to clipboard/png and reimport from the defaults, and fixes duplicate names automatically

- simple downloaders now remember their current parser through a session save/load

- simple downloaders will set a new default parser selection for new simple downloader pages on any choice event

- renamed all the default simple downloader parsers to more clearly explain what they do

- added 'all files linked by images in page' to the simple downloader, which does the old page of images' behaviour, and is hence pretty decent for most imageboards

- added a simple mewch thread html parse formula to the simple downloader

- added a very simple webmshare parse formula to the simple downloader

- added 'imgur image album' to the simple downloader, which will fetch the first 'page' of results from an image album. full parsing will have to wait for the gallery downloader update

- subscriptions can now run simultaneously! you can set the max number at options->downloading page–the default remains 1, so default behaviour will not change

- if subscriptions guess they can eat some bandwidth in the next 30s but it turns out they can't, they will bandwidth-override that last job after the 30s expires. this will stop some edge cases from causing subs to potentially hang for hours just on one last request

- 'explicit tags' are now renamed to 'additional tags'

- you can now set media to initially scale to 100% even when it is larger than the media or preview canvas

- the html parsing formula can now 'ascend' as well as search down the DOM tree–so, for instance, you can find all 'img' below an 'a' and then jump back up to the 'a' to fetch the 'href' of that image link!

- html parsing formulae can now search without a tag name–so, for instance, 'find all tags with class = "whatever"' is doable, without having to specify div or span etc..

- the html parsing formula rule panel is hence a bit richer. it also presents a preview of the rule's string as you edit

- the client no longer needs lxml to boot, and if html5lib is present, it will prefer that for html parsing. if both libraries are missing and html parsing is requested, it will raise an exception explaining the error

- the builds now include html5lib, which is better at recovering from some bad markup than lxml, and as a result some parsing jobs that formerly did not work (such as 'akaiha_(akaihasugk)' on danbooru) now do

- rewrote how services test their functional status to better account for different states. the logic here should be a bit better now–repositories will still be able to process if their network side is down, but everything will pause if any pause is set, and it should all propagate up to higher levels of error catching and status reporting better

- hydrus can now deal with really big decompression bombs (>~180 megapixel)

- filtered out en-dashes (as an alternative to standard hyphens in a "? + -" link-prefix) from booru tag parsing

- fixed a string generation issue that was stopping date decode string transformations from summarising themselves

- fixed some catastrophic boot failure gui reporting

- cleaned up a double-exit error on dialogs that could occur when spamming exit events with a script like AutoHotkey

- improved some dead cpu thread clearout code

- updated misc old code

- misc cleanup

next week

With the simple downloader finished, I want to get into the meat of the gallery downloader. The first step of this will be allowing drag and drop of arbitrary booru/whatever file page URLs onto the client.

____________________________
Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

ac1d61  No.8584

I figured out how to make Ebola-chan's symbol in unicode this week:

☜🌢☞

Please spread it as appropriate.

Also here is a leaner version of the rifle that has been floating around:

︻╤╦━╾╴

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

9de0be  No.8585

just going to make a recommendation for the simple downloader.

Personally, the vast majority of what I download will run through all linked files through images in page, and if it doesn't work though that, it's going to have to go through a gallery downloader.

Im sure in the future there will be a great use case for it but I can't see it right now.

would there be an easy way to put a checkbox on it to lock out any parser change other than what you have as default/last used?

It would largely be piece of mind more than anything else. I just have the feeling i'm going to accidently a large 20K image download at some point and not have the links to re feed the downloader.

that said, thanks for the update and getting the old functionality back, albe to go though some of a backlogged threadlist now.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

c56f36  No.8586

>>8584

Based dev.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

f5eb96  No.8589

I never thought I'd see the day that our glorious dev would grace us with his baneposting abilities. Thanks for all you hard work my man!

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

ecd947  No.8590

File: c77ebb1ee067fa9⋯.jpg (120.86 KB, 680x671, 680:671, 1464423673018-1.jpg)

>the amount of things in the to-do list increased again instead of decreasing

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

c14e2b  No.8591

>>8590

>>Devbro sticking to his weekly output well into the 2030's

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

cc35e5  No.8592

i love you

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

fd56cf  No.8593

File: 9573a2c756712ef⋯.jpg (193.46 KB, 600x800, 3:4, 9573a2c756712ef35bd5c12697….jpg)

Fucking nice. Thanks dev.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

e1c832  No.8597

appreciated as always

one thing i have noticed is that exif rotation metadata is correctly adhered when rendering thumbnails, but not in the image viewer itself. any chance of getting this little one fixed up?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

e14ba6  No.8598

File: f94569bf82e551c⋯.jpg (45.07 KB, 800x559, 800:559, ebils.jpg)

>>8591

>2030 comes in

>at this point hydrus has become a partially sentient AI that can automatically tag all forms of media by artist/creator, resolution, source, character names, series names, drawing type, format, and automatically downloads and archives from every source on the world at the users disposition and configuration

>Current To-do list has 37,654 items, because we still need a parser for IRL

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

f5eb96  No.8599

>>8598

> we all still have problems with data hoarding

Also dev PLS unpin the old update threads it's making my autistic checking of /hydrus/ a lot harder.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

5d45c2  No.8600

Please add the ability to set a shortcut key to the "set all selected as alternates" option. I've been setting alternates manually as I import images to avoid unnecessary work with the dupe finder, but I just imported 2000+ images I have to go though and each image has several alts. The right-click menu is way too slow, a shortcut key would be very useful.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

e14ba6  No.8601

>>8600

>to avoid unnecessary work with the dupe finder

dude what. just use the dupe finder, its amazing since he added that shortcut menu to the side.

i ran over 2000 duplicate pairs in a couple of days

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

e14ba6  No.8602

is there any way to download art from twitter?

i cannot understand why people use twitter, of all fucking social media available, to upload their fucking art, when its not designed for such a purpose, but they do. and there's a couple artists twitters that i've been eyeing

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

5eb770  No.8603

>>8602

Not built in. Hydrus_dev is working on various overhauls that may allow him [or third parties like "us"] to eventually add more downloaders in an efficient fashion.

I'm going to guess that if third parties can actually pick this up easily, we're probably going to see reddit/imgur and twitter and instagram etc. downloader script integrations for it. Most of these already have some downloaders out there on github and so on anyhow.

For now, you simply want to use an external application and just import the resulting folder.

If you also need twitter's tags, I guess you're going to have to find out how to regex them from the filename or wrangle them into a companion .txt file. Hydrus can be made to use either of these to record tags on import.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

e14ba6  No.8604

>>8603

eeeh i can do tags on my own, at least some general tags like artist name, character name, and whats going on in the picture. but i have no idea how to bulk-download from twatter because of their shitty weird interface.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

5eb770  No.8607

>>8604

> eeeh i can do tags on my own

Ok.

> i have no idea how to bulk-download from twatter because of their shitty weird interface

Well, obviously twitter themselves do not provide this?

They want your attention on ads and behavior data they can sell; not host easy media downloads in bulk.

Anyhow, you can still use software to get what you want. Use one of the multiple twitter downloader tools published on aforementioned github. They're typically fairly trivially easy to use from a command line.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

e14ba6  No.8608

deleted >>8605 >>8606 because im retarded i accidentally un-ticket "auto-add tags parents" and didnt realize until now

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

5109b0  No.8609

>>8583

>imgur image album

i was working on imgur album parsing but i stopped due to new job

each album/gallery has a chunk of json in the html with all the images in it


every script tag
regex
(?s)widgetFactory\.mergeConfig\('gallery', {(.+?)}\);

string manip regex
(?s).+image\s+:\s(.+?),\n.+

should output json that you can further parse to get all imgur pics from album

i ran out of time to actually going through and parsing the json though

note: is if the image is a gif then you can check if prefer_video is true and return .mp4 instead.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

0a76cf  No.8610

Sorry if this is already reported, but after updating (from like 29x) new parent tags do no longer work until i reboot hydrus - they then work just like old ones.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

cfecbc  No.8611

>>8583

This is a pretty massive feature request but I think it would fit naturally in with hydrus. How about adding some tensorflow script that scans all your tags and learns to identify the patterns that correlate with each tag in each image? Considering we all have huge libraries of manually curated, unique images, it seems like a machine leaning auto tagger that we could opt to train on our personal libraries is a natural fit. Like it could do all its training then the user could approve each image's added tags individually or all at once.

Hydrus is phenomenal btw, thanks for making it.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

ac1d61  No.8612

File: fba6abd6d4a8093⋯.jpg (195.46 KB, 640x960, 2:3, fba6abd6d4a80938edfec55a2b….jpg)

>>8585

Adding a persistent lock state for the dropdown choice is more complicated than I want to go right now, but a simple way to guarantee just one parser would be to edit them (hit the cog) and delete the rest. If you only need to access another parser type once in a blue moon, you can always add one or more of the defaults back in the same edit panel.

>>8590

>>8591

>>8598

>>8599

We're all heading into the darkness together, lads. I don't know if it ends, but I will keep pushing as long as I can. My general assumption is that the 2020s are going to be fucking wild, so I don't want to miss it.

And if we can birth some cheeky machine learning algorithms along the way, I'll consider that a true success. If we could use our existing content stockpiles to train machines to create better quality memes and other fun on demand, we might just make creator-tracking and hoarding (and hence hydrus's original concept) completely irrelevent. Perhaps then we could rest.

Our meme corpus is far in advance of theirs, as is our electronic sentience, and their… ethical inflexibility has allowed us to make progress in areas they refuse to consider.

>>8599

Thanks–I must have missed them last week.

>>8597

Can you post an example file that does this? And if you go thumbnail right-click->reparse files and regen thumbnails (which needs help->advanced mode switched on to see), does that change anything?

I hope to have a maintenance routine that will retroactively correct all the bad entries here, but I want to make sure I am doing it correct in the first place.

>>8600

I will see if I can slip this in. Also, if you haven't seen them yet, please try the duplicates processing page's new buttons as >>8601 suggests, as these are a good stopgap until I can do some larger workflow overhauls of this system.

>>8602

>>8603

>>8604

>>8607

The new 'simple downloader' has a simple 'twitter image' parser that'll do single posts for now. The new gallery overhaul will allow you to bolt a 'put in username, get image tweets back' search component to the same basic system and add tags and all that.

The 'drag-and-drop URLs on the client' stuff should be a half-step towards this, so you'll be able to drop twitter URLs on the client soon enough.

Twitter gif/video seemed a little trickier from my recent looks at it. It gets embedded dynamically by javascript somehow. I'll give it another look when it comes time to write a more complicated parser, but we might have to look for some aggregator feed or something that does the twitter-end API legwork for us.

>>8609

Thanks m8, this is great. Do you have an example imgur album that includes gifs?

I haven't read up on it yet, but apparently they have a decent explicit API as well. Someone linked me to this:

https://imgur.com/a/uAFvn

https://imgur.com/ajaxalbums/getimages/uAFvn/hit.json

Which will be a nice option when we get around to the more complicated gallery downloader, which will allow url class redirection to that api link.

>>8610

Thank you, I will check this.

>>8611

Thanks. This is the grand hope and long term plan. We build up a giant metadata corpus and then train the new generation of machine learning shit to do interesting things with it. I would love in the next three to five years to have some kind of auto-tagger going. Even just filling in the easy stuff like basic nouns and character names, or providing decent suggestions for a human to approve. Once we have some building blocks in place, we can then presumably go nuts.

I do not know a lot about this field. Have you used Tensorflow before–would you recommend it? Is there an easy job you think ML could do in hydrus that we could start on?

If you know, what are typical CPU/HDD loads for this stuff? In some examples I have already seen, the trained database can be multi-GB just for a few hundred different tags. Is the industry using GPU-based OpenCL/cuDNN to accelerate any of this stuff, or does it require a different set of hardware acceleration (like this: https://www.nvidia.com/en-us/data-center/tesla/) to do real work?

>>8586

>>8589

>>8592

>>8593

👍

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

e14ba6  No.8613

>>8612

>The new 'simple downloader' has a simple 'twitter image' parser that'll do single posts for now. The new gallery overhaul will allow you to bolt a 'put in username, get image tweets back' search component to the same basic system and add tags and all that.

>The 'drag-and-drop URLs on the client' stuff should be a half-step towards this, so you'll be able to drop twitter URLs on the client soon enough.

>Twitter gif/video seemed a little trickier from my recent looks at it. It gets embedded dynamically by javascript somehow. I'll give it another look when it comes time to write a more complicated parser, but we might have to look for some aggregator feed or something that does the twitter-end API legwork for us.

amazing. hydev you are a true hero.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

cfecbc  No.8615

>>8612

I wish I could answer your questions about ml. I only do programming recreationally and have hardly scratched the surface of tensorflow, but from what I've seen image recognition isn't hard to implement at all. The question of scaling across a variety of commercial hardware is an interesting one. Does hydrus have an api?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

b6f8f8  No.8616

In regards to the new sankaku limit, would it be possible for the downloader to grab the list of files, but instead of grabbing the files from sankaku directly, instead try to download them using ipfs?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

9de0be  No.8617

>>8612

with the persistent lock, instead of deleting them would it be possible to add a disable? there are ones like twitter or imgur that I will likely end up using in time and dont really want to delete because what if.

though I will go through with deleting them if I do end up fucking an import up.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

e14ba6  No.8618

File: d0f5ce7c951b7b3⋯.jpg (498.52 KB, 1066x1600, 533:800, 1334865717847.jpg)

will we ever get a parser for yiff.party pages?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

f5eb96  No.8620

>>8618

I'm trying to get a hold on those downloader formulae things, can you explain a bit more about how that website works and what pages you want scraped? I'll see if I can take a look at getting a downloader formulae working for it.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

e14ba6  No.8623

>>8620

i have no idea how it works, yiff.party is not specifically aimed at furry, disregard the name, but at any and all patreon content.

users use the sites functions to "rip" posts from a patreon page and upload them to the site. The developer of it refuses to add a "download all" button to the site because of bandwidth. you also need a cookie to be allowed access to the site. You'd have better luck going to the yiff.party thread and asking there directly, the dev should be browsing casually. I'm just a user reaping the benefits tbh.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

314faf  No.8624

>>8612

For what it's worth, once I get the cash to fix my good desktop I'm planning to see if it can sustain a bit of HyperGAN so I can try to auto-generate more niche porn using part of my Hydrus collection.

I'll get back to you if it yields any good results.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

c14e2b  No.8626

>>8616

How the hell would that even work? Why would you assume that someone has those files lying around and has them uploaded on the network?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

2cda3e  No.8631

im having a problem with pixiv, or maybe its a well known current-issue?

when i try to download pictures that are in multiple pages the download fails. i've tried through Subscriptions and through the raw URL downloader, no luck. any help??

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

d158ef  No.8636

>>8612

>Our meme corpus is far in advance of theirs, as is our electronic sentience, and their… ethical inflexibility has allowed us to make progress in areas they refuse to consider.

Is that The Algorithm from Alpha Centauri?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

ac1d61  No.8642

HookTube embed. Click on thumbnail to play.

>>8636

Why contain it?

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

ac1d61  No.8643

File: c90bc58ee404867⋯.png (2.94 KB, 512x84, 128:21, yiff.party_attachments_htm….png)

File: 86a2a6f9785a6d7⋯.png (2.94 KB, 512x84, 128:21, yiff.party_post_file_html.png)

>>8615

There is no API yet, but I would like to build a way for the client to accept some GET/POST stuff for common commands like importing files and applying tags for users who would like to automate this stuff with scripts.

I did a little more research on this, and it looks like DNN is getting popular for consumer and low-level enterprise stuff here and getting rolled out in new drivers and libraries, so I think that's the ticket. If vidya devs start employing that for basic pattern recognition for simple AI strategies or something and there is demand for specialised hardware in every GPU, I figure this'll be a big fucking deal in ten years.

>>8616

>>8626

This is too complicated for now. I've been thinking about maybe doing a 'public URL repository' at some point, so users can share file-url mappings, which would be helpful here. If you were informed one way or another that the files you wanted were on sankaku but you had no bandwidth for it, it wouldn't be super difficult to write a search routine to look up any other known URLs of the files as shared by the repo, and you could then fetch it from a booru or IPFS or anything another user had seen.

>>8617

I mention delete/restore as a quick solution in your case because you can restore from default really quickly from the edit dialog. Just click the button and select from a list.

That said, maybe a button that launches a 'select-from-list' is better here as the main choice control. I want to do the same thing with the broader subscription gallery selector, which is its own dropdown mess atm. I will think about this a bit.

>>8618

>>8620

>>8623

Another user sent me these pngs this week. I have not tested them but if they work out, I will integrate them into the client defaults. Please feel free to play with them.

>>8624

Thanks, I would be very interested.

>>8631

Yes, unfortunately manga pages are not supported in the old system (it can't handle >1 image per post). This new system I am moving the gallery downloader to should be able to handle them, so I hope to roll out an update for it as I move everything over. This is a highly requested feature, so you won't be able to miss the fanfare when it happens. Let me know how it works for you when it does.

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

e1c832  No.8658

File: 6e3a99d6c650edf⋯.jpg (1.9 MB, 4032x3024, 4:3, IMG_3266.JPG)

>>8612

I've attached a sample image that incorrectly handles rotation metadata. In windows the hot dog is vertical, but in hydrus it is horizontal in the thumbnail and squashed with incorrect proportions in the media viewer.

hope this helps

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.

ac1d61  No.8672

File: 940527fad6c2610⋯.png (260.47 KB, 827x622, 827:622, dog.png)

>>8658

Thank you–I actually get pic related on my dev machine. If you go help->advanced mode and then right-click the thumb and hit 'reparse files and regen thumbnails', does that change the image?

If you open help->about, what version of OpenCV and Pillow does your client think it has? Do you happen to have the 'load images with PIL' setting under options->media set?

I notice that your thumb shows the dog vertical, but if I click on this in Firefox to embed it, I get it horizontal (-90 degrees).

Disclaimer: this post and the subject matter and contents thereof - text, media, or otherwise - do not necessarily reflect the views of the 8kun administration.



[Return][Go to top][Catalog][Nerve Center][Random][Post a Reply]
Delete Post [ ]
[]
[ / / / / / / / / / / / / / ] [ dir / random / bane / clang / hentai / htg / loomis / mde / pone / rule34 ]