Moved from page
I moved the following from the page. It's a good idea, but the implementation is flawed — one of the shares shows the image. — Matt Crypto 08:36, 30 November 2005 (UTC)
Yes, I did a mistake. The second image was also intended to be black&white whereas it contained some additional colours that gave it away. I changed it to monochrome and now I hope it's ok, but it's still better if you check it :). Thanks for the notice. Random user name 20:12, 12 December 2005 (UTC)
The images look OK to me now. So I moved them back to the article. It seems terribly complicated, though, in order to communicate a well-known logo. Would a much lower-resolution image that encoded a short bit of text be worthwhile? Perhaps something about the squeamish ossifrage or something else from the list of famous ciphertexts? --DavidCary 16:31, 23 December 2005 (UTC)
I don't quite understand the use of 'complicated' here. Is the problem with the size of the image or the wording of the description? Random user name 02:44, 31 December 2005 (UTC)
Fascinating idea, but I don't have Matlab. Any way of showing what the result is supposed to be without Matlab? —Preceding unsigned comment added by Mcswell (talk • contribs) 15:51, 5 April 2006 (UTC)
- It is also possible to take one image and overlay it on top of the other in Photoshop or The GIMP with ~50% transparency. Also, the matlab code given doesn't work in the latest matlab (2006b, I don't know if the behavior in question changed since previous versions) because the image 1 is in 8 bit RGB and image 2 is in 16-color indexed grayscale. Matlab thus gives a 3 dimensional array with white = 0 black = 255 for image 1 and a 2 dimensional array with white = 0 black = 15 for image 2. I imagine it'd be easier to fix the images rather than add the necessary code to manipulate the images in matlab to the desired form. Drooling Sheep 10:33, 14 January 2007 (UTC)
Is this worth adding to the article?
- Doug Stinson's Visual Cryptography Page —Preceding unsigned comment added by DavidCary (talk • contribs) 22:42, 23 December 2005 (UTC)
Is it just me, or does it seem strange that the article intro says that visual cryptography can be done without computers, but the example uses code to generate a new image? Timbatron 16:51, 22 March 2007 (UTC)
- If you print both of the images on transparency paper and overlay them, you'll see the image. The example code does the same. By the way, if you have an lcd screen and open the two images in two tabs in Firefox, and switch between them, you'll see the combined image for a moment. 220.127.116.11 19:21, 3 May 2007 (UTC)
I decided to be bold and change the example that was in here, for several reasons:
- I strongly agree with Timbatron's comments above -- the article is about cryptography without computers, yet gives code for decrypting the shares.
- Two static grey blobs of pixels do not really illustrate the idea. The magic is when you overlay the transparency and the image suddenly appears.
- Any reader should be able to see the idea of overlaying things, without having to save these images / know how to use photoshop / print things out onto transparencies / know how to run matlab code / know that they should toggle really fast between two firefox tabs on their LCD monitor / etc...
So I made up a smaller animated example that I hope gets the point across better. I removed all the code snippets (some vandal had already removed half already). Comments welcome.
I tried to explain a little bit about how the sharing is performed and why it is a valid secret sharing. If needed, we could also include a chart as on Doug Stinton's page, describing how each pixel is shared (I used different shares than in that diagram though). // Blokhead 16:42, 26 August 2007 (UTC)