[Cryptography] Finding Nemo's random seed

Kent Borg kentborg at borg.org
Tue Sep 5 14:38:33 EDT 2017


On 09/05/2017 01:06 PM, John Denker via cryptography wrote:
> -- Sometimes you want something that is entirely deterministic and 
> predictable, just complicated-looking, e.g. sea grass.

And presumably the folks writing the rendering software /did/ know this. 
The older rendering software knew how to keep track of its RNG state, so 
it could re-render frames without positions changing. It is common to 
re-render 3D scenes and fragments of scenes as components are added, 
tweaked, lighting changed, eventually better quality output generated, 
etc. The sea grass that Nemo is swimming through had better not jump 
from one rendering to another. Someone was very aware of this.

But it seems that wasn't explicit in the "source code" that they 
preserved for Finding Nemo. And somehow programmers are better at 
writing functions for importing data than they are for exporting data...

Similarly for any Monte Carlo simulation: frequently one wants to be 
able to rerun different version of the code on previous "random" data.

-kb
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20170905/e82a1b63/attachment.html>


More information about the cryptography mailing list