Posted under Bugs & Features
When I try to upload anything from skeb, it keeps failing with an error message like "Upload failed: https://skeb.jp/@wobumi/works/16 doesn't contain any images." even if I try one of the posts from skeb already uploaded here which definitely have images. Tried Chrome and Firefox and both had the same problem. Any way to fix it?
For this platform I just download and upload the image, then use the link as a source.
You can use the direct image URL to upload. It'll even automatically link to the Skeb page on upload without the need to change the link manually. I personally use the Upload to Danbooru extension to right-click the image and upload. What it won't do is grab any commentary...
Skeb uploading is currently broken, see issue #5773 on github
@Alixiron If you want to get your images unwatermarked you can do this with one simple trick
In order to do it automatically you will need a help of your system to call Skeb API.
Here is a curl query template which gets the response with the unwatermarked image URL
curl https://skeb.jp/api/users/<artist_name>/works/<commission_number> -H 'Authorization: Bearer <your_extracted_token>' -H 'Cookie: request_key=<your_extracted_request_key>' -H 'User-Agent: <user_agent>'
Notice that the URL after curl keyword is not an URL to the web page, it points to API. For example https://skeb.jp/@usumeshirouchan/works/65 (webpage) becomes https://skeb.jp/api/users/usumeshirouchan/works/65 (API call).
After all is said and done a real query can look like this:
curl https://skeb.jp/api/users/usumeshirouchan/works/65 -H 'Authorization: Bearer eyJhbGciOiJSUzI1NiJ9.eyJkYXRhIjp...' -H 'Cookie: request_key=fb58e6fc:aed...' -H 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36'
Here are useful one-line scripts in Powershell (built natively in Windows)
curl https://skeb.jp/api/users/<skeb_artist_name>/works/<commission_number> -H 'Authorization: Bearer <your_extracted_token>' -H 'Cookie: request_key=<your_extracted_request_key>' -H 'User-Agent: <user_agent>' | ConvertFrom-Json -AsHashtable | Select-Object -ExpandProperty 'article_image_url'
Copy paste the link in output and download it from your browser.
For Linux I guess you can Bash and replace the pipeline | ConvertFrom-Json -AsHashtable | Select-Object -ExpandProperty 'article_image_url' with | jq -r '.article_image_url' but I didn't test it myself.
curl (curl https://skeb.jp/api/users/<skeb_artist_name>/works/<commission_number> -H 'Authorization: Bearer <your_extracted_token>' -H 'Cookie: request_key=<your_extracted_request_key>' -H 'User-Agent: <user_agent>' | ConvertFrom-Json -AsHashtable | Select-Object -ExpandProperty 'article_image_url') -o "<output_path>"
output_path is a target path to the file in your system, for example "C:/Test/output.jpg", the image will be saved as such.
My local results
Alternatively to system console you can download Postman program and use it to call the API.
Open a new tab, paste the API url, fill the headers with creds, click Send button and that's it. Example
CTRL + click the link and it opens in your browser.
- Copy paste the webpage link as a source
- Copy paste the commentary
- Tag meticulously
- Don't forget about skeb commission!
- Post it
- Make the world a better place
It's enough to configure it once, after that you only need to replace an artist name & a work number and press Enter.
As always only the first image in a Skeb collection is free of the sample watermark.
Updated
This method you describe sounds kind of like an exploit, yeah it's there for the taking but isn't this basically like taking a private / paid image?
Well if its fine, why not.
Freshblink said:
This method you describe sounds kind of like an exploit, yeah it's there for the taking but isn't this basically like taking a private / paid image?
Well if its fine, why not.
Not even remotely close. This is literally just using the API, which is like, the normal thing to do.
SonMati69 said:
Show
@Alixiron If you want to get your images unwatermarked you can do this with one simple trick
Extract your Skeb creds
- Open any Skeb.jp page
- Ensure you're logged in
- Open dev tools in your browser, for example by clicking F12 in Chrome or Firefox
- From local storage copy the value of token property, example
- From cookies copy the value of request_key cookie, example
Get the unwatermarked image
In order to do it automatically you will need a help of your system to call Skeb API.
Here is a curl query template which gets the response with the unwatermarked image URL
curl https://skeb.jp/api/users/<skeb_artist_name>/works/<commission_number> -H 'Authorization: Bearer <your_extracted_token>' -H 'Cookie: request_key=<your_extracted_request_key>' -H 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36'Notice that the URL after curl keyword is not an URL to the web page, it points to API. For example https://skeb.jp/@usumeshirouchan/works/65 (webpage) becomes https://skeb.jp/api/users/usumeshirouchan/works/65 (API call).
User-Agent part is a random, but valid source of the request, it's necessary to represent yourself as a web browser.
After all is said and done a real query can look like this:Show
curl https://skeb.jp/api/users/usumeshirouchan/works/65 -H 'Authorization: Bearer eyJhbGciOiJSUzI1NiJ9.eyJkYXRhIjp...' -H 'Cookie: request_key=fb58e6fc:aed...' -H 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36'Use system console
Here are useful one-line scripts in Powershell (built natively in Windows)
Get unwatermarked image URL
curl https://skeb.jp/api/users/<skeb_artist_name>/works/<commission_number> -H 'Authorization: Bearer <your_extracted_token>' -H 'Cookie: request_key=<your_extracted_request_key>' -H 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36' | ConvertFrom-Json -AsHashtable | Select-Object -ExpandProperty 'article_image_url'Copy paste the link in output and download it from your browser.
For Linux I guess you can Bash and replace the pipeline
| ConvertFrom-Json -AsHashtable | Select-Object -ExpandProperty 'article_image_url'with| jq -r '.article_image_url'but I didn't test it myself.Download unwatermarked image
curl (curl https://skeb.jp/api/users/<skeb_artist_name>/works/<commission_number> -H 'Authorization: Bearer <your_extracted_token>' -H 'Cookie: request_key=<your_extracted_request_key>' -H 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36' | ConvertFrom-Json -AsHashtable | Select-Object -ExpandProperty 'article_image_url') -o "<output_path>"output_path is a target path to the file in your system, for example "C:/Test/output.jpg", the image will be saved as such.
My local resultsUse Postman
Alternatively to system console you can download Postman program and use it to call the API.
Open a new tab, paste the API url, fill the headers with creds, click Send button and that's it. Example
CTRL + click the link and it opens in your browser.Upload the file to Danbooru
- Copy paste the webpage link as a source
- Copy paste the commentary
- Tag meticulously
- Don't forget about skeb commission!
- Post it
- Make the world a better placeRemarks
It's enough to configure it once, after that you only need to replace an artist name & a work number and press Enter.
As always only the first image in a Skeb collection is free of the sample watermark.
Had some errors with the curl one for some reason but postman worked like a charm while also being easy to setup following your example. Thanks a lot.
Updated
Unwatermarked article_image_url versions have lower quality settings than the watermarked versions (fm=jpg&q=45&w=696) and are not always available. There is some more information about them available at https://github.com/mikf/gallery-dl/issues/1031.
Freshblink said:
This method you describe sounds kind of like an exploit, yeah it's there for the taking but isn't this basically like taking a private / paid image?
Well if its fine, why not.
This is the same method which danbooru.rb is using, but w/o getting the 429 HTTP error
https://github.com/danbooru/danbooru/blob/master/app/logical/source/extractor/skeb.rb
It seems that Skeb backend uses the value of User-Agent header to validate the request, so it cannot be random.
My 3 months old creds became invalid when combined with user-agent from another browser.
A trio of bearer token, request key cookie and user-agent header is used to tell Skeb API: "Hello, I'm a real user from the browser, totally not some scrapper~"
I updated the tutorial above, here is the short of it:
This time I did more tests, I signed out and in to Skeb to check whether it works with refreshed creds, fake values, etc
Thanks to @Dead_Princess for reporting the issue :thumbsup:
