Why does Smush it violate DMCA?

Hi Dave,

Smush.it uses the well-known software called jpegtran (http://www.ijg.org/) to remove all the meta information. While doing this we didn't realize the copyright problem, it was purely unintentional. The proper technical solution would be to strip needles meta data, while keeping copyright data, but the options in jpegtran allow for kind of all or nothing (or comments only) metadata preservation.

We looked at the best option to approach this and found a possible solution using imagemagick. The right solution looks a little difficult right now, because also our hosting provider is running an older version of image magick that doesn't have support IPTC manipulation.

And yes, btw, this tool is built by me and Nicole - two enthusiasts in their own time. It's not endorsed, nor even hosted by Yahoo. It was meant as a service to the community, not a tool to do harm.

Anyway: I'm happy to announce: smush.it now retains all the meta data, including copyright information.

This is a temporary solution to keep all the meta. The saving are no longer that impressive, but we're be working on the right solution that helps performance and doesn't harm artists.

Once again, thanks to you and everyone who brought the issue to our attention!

Stoyan:

Thanks for being so responsive. If you were not aware, there are many resources out there that allow one to manage the metadata in images. The problem has been that there are several places it can be stored, notably, in the Exif, IPTC-IIM, or XMP containers. Making sure that all of the values are synchronized can be a bigger challenge. However, one group has stepped forward to take on this challenge. Dubbed the Metadata Working Group, representatives from Microsoft, Adobe, Apple, Canon and Nokia, have released a White paper called Guidelines for Handling Metadata, that would be a good reference to check. Does Smushit have a metadata policy -- a set of acceptable practices and standards? If not, I would suggest looking this Metadata Working Groups white paper as a good place to start.

While I understand the need to make images smaller for the web (actually resizing the image rather than changing the image pixel attributes), I think that encouraging users to strip metadata wholesale creates a real issue. I know that for small web images, the amount of textual data within them can bloat, but this doesn't mean that you have to throw the baby out with the bath water. Even offering to remove all but the copyright notice, copyright URL, XMP rights (Rights Usage Terms) and Creator fields would still reduce the size of the image with causing it to become an Orphaned Work.

In using the tool, I see no attempt to warn users not to violate copyright, in fact, there doesn't even appear to even be an End User License Agreement anywhere on the site. It would be useful to at least have a warning to notify users that they must only run such a tool on their own images (these would be the only ones which they have the authority to modify). However, in looking at the Smushit tool, I don't even see a checkbox to acknowledge that these are the users property. In fact, the way it's set up, it seems to encourage the use of property other than one's own. This I find troubling, especially since when you launched, the tool removed metadata in the images, so if someone "borrowed" an image from another website, the version provided by Smushit was bereft of any information that might have helped the end user locate the original source at a later point in time. If this is to remain a fixture, I would recommend, that at minimum, the URL that is supplied for the original image be embedded into the image as a means to return to the original at some point in the future.

You mention using ImageMagick. This application does offer the ability to retain Image Metadata (at least IPTC Information Interchange Model types stored in a binary header). The other popular application GD does not have this feature, however, Laura Cotterman has provided a set of free functions that will work with GD and provide a means to preserve the metadata when resizing, watermarking, etc. You can find those on her "Image Metadata" website if you are interested.

Hope that helps, and thanks for making the appropriate changes.

David
--
Stock Artists Alliance Photo Metadata Project leader
http://www.stockartistsalliance.org/photometadata-project
Chairman, SAA Imaging Technology Standards Committee

Thanks Dave, this is really helplful

It's true that don't have any privacy or end-user agreement, in our defense, this is a personal project for me and Nicole (not a Yahoo project), so we only wanted to provide a useful service and were busy coding a cool app, and unfortunately not taking care of all the legal requirements.

the Metadata Working Group involvement is highly appreciated!

Good news is that temporarily we keep all the metadata while working on a clean solution. It was our hope to be of public service, not harm, I hope no one was injured :(Stoyan

A few suggestions, and a thanks for being so quick to respond to these concerns.

1. I see nowhere on SmushIt where it even asks/reminds a user to only use this tool on images to which they have rights - that is a crucial omission IMHO
2. I think, while technically interesting, the Firefox plugin and the URL fetcher interfaces simply encourage people to grab things off Websites not under their control - maybe not your intent, but that is going to be a very common use case; as a tool for optimizing images for legitimate users, I can't see many cases where I'd need to reference an image on a publicly available site if its mine and under my control - i'd focus more on making the uploading process easier. Note: Flickr does not, for instance, allow me to import images from public Websites into my Flickr account... not that it'd be technically difficult, but more likely because that would seem to simply promote snatch/grab of images.
3. jpegtran is fairly limited - if you need a toolkit (suitable for use in a Web app) for manipulating embedded metadata in image files, look into exiftool - it'll allow you to selectively strip metadata from just about any file you might deal with.

Best,

Roger Howard

Thanks Roger!

exiftool looks like a nice tool, I'll look into it switching to it.

we have no privacy policy (guilty!) and we have no experience writing one (we're not Yahoo, just two engineers) so any community help or templates will be highly appreciated

Please don't strip the metadata from my pictures. I work really hard to incorporate metadata into my photographs. Metadata has been very, very good to me, my friends, and my clients.

Walter Dufresne
Brooklyn, NY

To Walter and Stoyan:

I'd suggest maybe having two tools to choose from. One tool (the current version) will allow you to retain all of your meta data. The second tool would strip all metadata from the image (with the exception of the copyright info, of course) to compress the images as much as possible. Just a thought, that's all. I was very impressed with SmushIt the first (and unfortunately only) time that I used it.

Which brings me to a big question: What happened to SmushIt? I just used it no more than a week ago, and now all of a sudden, when I click on the little SmushIt icon at the bottom of my Firefox browser window, it simply loads up a blank white website that has no HTML in it at all. Could it possibly be a problem on my end?

Sincerely,

Jon Schuster

No offense to the artists, but the tool is much less useful now. Many of us were more than likely using it on our own graphics, or company graphics, and can't get the same SMUSH ability now, so the tool is kind of defeated. The concern is legitimate, but in this case I think it needed to be ignored, unless the artist could prove his work was being disseminated illegally, which I doubt he could. Politically correct bull****!

There is no liability under DMCA Section 1202 unless the copyright data was removed for the &quot;purpose of inducing, enabling, or facilitating copyright infringement.&quot;<br><br>And, anyways, if someone is using your image without your permission, their removing of your metadata is the least of your problems. And all you have to do to prove infringement is to show the original. If they claim fair use and include information elsewhere to indicate your copyright, the court will still decide only on the fair use case, and not whether there is&nbsp; metadata repeating the copyright information.