Uploading to Derpibooru
Uploading content is a piece of cake. You just hit the 'Upload' button and fill in the form. However, there's a few little details we'd like to explain better to you if you're interested.
We provide a few fields for metadata - tags are designed to let you group images together, and describe things in terms of their content. We also have a description field, which is intended mostly for people uploading original content to the Booru, or for more detailed description of images or context around the image. It can also be used for audio description of images for people with screenreaders.
For instance, the fact that an image contains Twilight Sparkle belongs as a tag, and if it's the Summer Sun Celebration, that's a tag, too. Fully describing an image like "Twilight sparkle standing on the steps in awe at Celestia at the Summer Sun Celebration, wide-eyed in awe" should be done in the description.
We also have some 'meta' tags - artist:artistname tags should be used to link an artist name to an image. There are also spoilered or hidden by default tags, which stop NSFW things from popping up when not wanted - these should be used where appropriate.
Finally there is the source URL field. This should link to the page on which the image was originally found. If you don't know, leave it blank, but try and find it first.
Scalable Vector Graphic uploads
We support SVG uploads - once we get them on the server we make PNG images out of them, but people can still download and view the SVG version on the links on the image.
We recommend you provide a sensible default resolution with your document - a couple of thousand pixels is plenty!
When you upload a GIF, JPEG or PNG, we do some checks on the image once it's been uploaded. Most images have un-needed data in them, which can be safely removed without affecting quality. We use a few tools to do this on your uploads, resulting in smaller file sizes for us to store, and faster page loads for everybody.
We use optipng to deinterlace and compress PNG images, fixing any encoding issues on the way. The specific command-line we use is:
optipng -i0 -fix -o5 your-file.png
We use jpegoptim to sort out JPEGs, using a maximum quality factor of 95, which yields no noticeable quality changes and significant potential savings (up to 50% on some images). The specific command-line we use is:
jpegoptim --max=95 your-file.jpg
GIFs are a bit more complex, as we treat all GIFs as probably animated, and so have to deal with all the frame processing (unlike some tools which just make a thumbnail from the first frame, which we think is a bit cheeky). We use gifsicle, with the following command-line:
gifsicle -O3 your-file.gif
We also use gifsicle for resizing GIFs, using --resize-fit. PNGs and JPEG images are resized with ImageMagick.
We also do basic deduplication using SHA512 hashes of images uploaded to Derpibooru. If there's a hash collision upon upload, which indicates two identical images, then the new upload will not be permitted - it'll throw an error up at you with a link to the other file found so you can go look at that instead.
In addition to our basic SHA512 exact-match deduplication we do perceptual, content based deduplication. If your image is similar to any already present in the database, but not exactly the same, we'll show you all the similar images and you can mark the image as a duplicate or confirm your image is indeed different. The algorithm for near-match deduplication uses a perceptually weighted RGB intensity average plus some other bits and pieces like aspect ratio.
We do all the processing in the background, and while we're doing it, we continue to serve the unoptimized file, so there's no noticeable difference for anyone. It is, however, noteworthy if you intend to download a file- you may wish to wait for the fully processed image to be available. Depending on the file type and file in question, it can take a while to process images - up to 5-6 minutes in some cases - and since we only have limited resources, if there's a queue, this could be a longer wait.
Basically- upload, and don't worry about it! We'll handle all the heavy lifting on our end, and once we finish processing the image, it'll be served instead of the old unoptimized one immediately.