New work for Napster and p2p

Posted 29 Mar 2001 at 23:10 UTC by Malx Share This

Why only mp3?
You could distribute images and photoes same way, but latest are made by self.

Most of people make photos. Some of them really great.
Most of people just put them on Web to get rid of them. They are not against people downloading them. (Arn't you? ;).
Sometimes one want to find some specific image for page (say apple), but it is hard to find one.
With simple descriptions and some P2P programm (server in which you could setup public photo/images dir) it whould be easy to find one you need. Or just to make changable desktop image.

The main difference with mp3 - almost all are creators of photos/images, but almost none of mp3 :)

Info to specify is: size XxY px, title, list of objects on it, place(country/city/street),
camera type, film type, scanner. Also quality could be specified.
You could thinks of adding md5 check to (image+info) data, so it whould be easy to find author of original image.

Differences between music and images and habits, posted 30 Mar 2001 at 00:39 UTC by khazad » (Journeyer)

I don't think the P2P model fits well for images now. (Napster is not really P2P, to start with).

Why P2P works for music ?

  • the current music file formats carry along within the files a description of its contents. Many image formats don't allow this, and even when they do, you can count on your fingers how many people fill that field with useful information.
  • people (the peers) are willing to keep a reasonably large music collection (many people keep gigabytes of MP3s in their hard disks. Best music playback is achieved when the files are in a fast, mounted 24/7, media. Most people already have the music files available 100% of the uptime there, and most are online for at least some hours each day, sharing is just a step further. And they don't even need to index the files, the ID3 tags are already there!

    How this differs from images ?
    The usual "image experience" is not "playing back" a collection of image files in a playlist (you don't really watch a slideshow of your JPEGs every morning, do you?). So the image are better placed on removable media -- zip disks, CD-Rs. At least for me, my image collections are kept in 96.1 MB zip disks (if you want to believe 100,000,000 B == 100 MB, go ahead...). The "image experience" is 1) I need an image to tinker with 2) I grep the 'ls -lR' of my image disks (kept in my /home) for what I want, 3) mount the disk 4) fire up gimp 5) copy whatever needed inside gimp 6) umount disk. The images aren't "there" to be shared, many times they aren't mounted. Also, classifying images for searchability can be boring -- altavista must know that, a media search for "tolkien dwarf" returned a page with pictures of crocodiles for me :)

    So provinding a service where images are indexed and available online is probably better achieved by concentrating the images in one or a few servers, with 24/7 storage for them.

    When we started sharing music we needed special programs to do that -- Netscape, Opera, Mozilla, et caterva don't know how to "display" music files, and even if they did, you'd probably get angry since these browsers freeze up so much -- not reliable enough for playback. So here comes Napster with search primitives for bitrate, name, line speed, etc.

    Also, image files usually aren't as large as music files. HTTP transfers are enough.

    But browsers can show images pretty well. And getting them to browse some other formats (Gimp's XCF, TIFF, PCX...) doesn't look so tough (as long as the browser in question is under an open source license that encourages contributions from users). I'd say the solution here is an image collection based on HTTP with an associated database (MySQL, PostgreSQL) and server-side code to search for images and contribute to the archive (e.g.: PHP, CGI, Zope/Wikis). Better reliability could be achieved by mirroring the site.

    Now, a search for photo+album on freshmeat will give you plenty of this kind of tool.

    Even though P2P networks are nice, and many things will get shared on them when the P2P-thing really happen, from music to images to flashes (All your base II - CATS's wrath ?) to videos to text docs. For now, P2P just isn't the "right here, right now" solution for image sharing.

    One thing we could start thinking is a wrapper file format for images, with a header containing image dimensions, size, keywords, author info, and a data chunk containing the image file in its original format (PNG, JPG, etc), so that images could be shared without the aid of a database backend and still easily opened (modifying imaging programs to "extract" the file is not tough, and a command line tool to extract the image file from the "wrapper archive" can be written in 30- minutes (given the wrapper format spec).

  • JPEG, GIF89a, PNG and TIFF files can contain comments, posted 30 Mar 2001 at 01:40 UTC by jmg » (Master)

    Take a look at rdjpgcom and wrjpgcom that are part of the jpeg library that almost everyone uses on Unix. I tag most of the files that I index on my home page with this. This lets me throw in a copyright and a description. Then my indexing script will pull the description out of the comment block for the index.

    If you take a look at the index of Blue Mountains (notice the indx extention? It invokes my Python script that dynamicly builds the index. It also resizes the images to your current preferences.) you will see the descriptions. If you click on an image (not a thumbnail, I haven't gotten it to push the descriptions to the resized images yet), and then view an image like myself in the Blue Moutains and look at the Page Info (with Netscape, not sure you can do this with IE), you will see the copyright and description of the file.

    The technology already exists out there for tagging files, and has been for a long while. Just the applications haven't been making use of it.

    Porn, anyone?, posted 30 Mar 2001 at 18:30 UTC by kelly » (Master)

    Of course, the real reason why there's no interest in p2p distribution of images is that there's no real demand for graymarket copies of copyrighted photographs, with one exception: porn. A p2p distribution system for pornography would probably be in extremely high demand -- until people started sending kiddie porn over it, and the feds arrest everyone even remotely involved. Facilitating copyright infringement (Napster) is merely a civil offense. Facilitating the distribution of child pornography, on the other hand, is a federal felony.

    real "p2p" doesn't care about the data type, posted 30 Mar 2001 at 18:30 UTC by splork » (Master)

    I believe napster is the only one that restricts what the bits are supposed to contain. Mojonation even has an often overlooked feature of being able to publish and view websites as a single piece of content (perfect for photo galleries). Just point it at the top level html file or the directory that contains it when publishing and it will spider it for relative links, publishing it as a whole.

    New Advogato Features

    New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

    Keep up with the latest Advogato features by reading the Advogato status blog.

    If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!

    Share this page