SHA1 and MD5 are cryptographically secure hashes, which means they are intended to be non-reversible. Murmur3 isn't cryptographically secure so it is much faster. -ben
On Wed, Aug 15, 2012 at 3:13 PM, Alan Fregtman <[email protected]> wrote: > What's Murmur's benefit over say, SHA1 or MD5? > > > On Wed, Aug 15, 2012 at 2:27 PM, Ben Houston <[email protected]> wrote: >> >> The basis of hashing in Alembic is the murmur3 hash. I'd recommend it. >> >> http://en.wikipedia.org/wiki/MurmurHash >> >> -ben >> >> On Wed, Aug 15, 2012 at 2:19 PM, Nicolas Burtnyk <[email protected]> >> wrote: >> > I've used hashes of the geometry data itself (vertex and polygon data) >> > to >> > identify meshes, but this obviously doesn't work if you want the hash to >> > remain stable between edits of the mesh. If you want to know if 2 >> > meshes >> > are identical with just different names though, it's extremely reliable >> > assuming you have a good hash function. >> > >> > -Nicolas >> > >> > >> > >> > On Wed, Aug 15, 2012 at 10:49 AM, Alan Fregtman >> > <[email protected]> >> > wrote: >> >> >> >> I know, but normal artists will extract geo and do other things that >> >> would >> >> destroy any metadata I could put on the object itself. >> >> >> >> >> >> On Wed, Aug 15, 2012 at 1:42 PM, Matt Lind <[email protected]> >> >> wrote: >> >>> >> >>> You’re describing Asset Management. >> >>> >> >>> >> >>> >> >>> A solution is to put metadata on each mesh (ie: custom property or >> >>> userdata blob) with a unique ID and register the asset somewhere such >> >>> as in >> >>> a database. If the mesh is renamed, it’s name will be irrelevant as >> >>> you’ll >> >>> be tracking the ID in the custom property, not the mesh name. You can >> >>> quickly find the custom properties via FindObjects(). You can resolve >> >>> collisions of duplicates using the mesh name. >> >>> >> >>> >> >>> >> >>> Matt >> >>> >> >>> >> >>> >> >>> >> >>> >> >>> >> >>> >> >>> >> >>> >> >>> >> >>> >> >>> From: [email protected] >> >>> [mailto:[email protected]] On Behalf Of Alan >> >>> Fregtman >> >>> Sent: Wednesday, August 15, 2012 8:28 AM >> >>> To: XSI Mailing List >> >>> Subject: Anyone ever tried generating a unique hash based on topology? >> >>> >> >>> >> >>> >> >>> Hey guys, >> >>> >> >>> >> >>> >> >>> I'm trying to keep track of a few thousand meshes which are being >> >>> renamed >> >>> between stages/departments and I was thinking it might be worth a shot >> >>> to >> >>> generate a hash based on topology so that I could track them better. >> >>> >> >>> >> >>> >> >>> Has anyone tried this already? I got pretty close hashing a string of >> >>> component counts and bounding box values... In a section of 415 meshes >> >>> there >> >>> were only 8 hash collisions. >> >>> >> >>> >> >>> >> >>> Any ideas what else I can consider to make the hashes more unique? Or >> >>> an >> >>> alternate solution to tracking existing sets of meshes with same >> >>> topology >> >>> but potentially inconsistent naming? >> >>> >> >>> >> >>> >> >>> Cheers, >> >>> >> >>> >> >>> >> >>> -- Alan >> >>> >> >>> >> >> >> >> >> > >> >> >> >> -- >> Best regards, >> Ben Houston >> Voice: 613-762-4113 Skype: ben.exocortex Twitter: @exocortexcom >> http://Exocortex.com - Passionate CG Software Professionals. >> > -- Best regards, Ben Houston Voice: 613-762-4113 Skype: ben.exocortex Twitter: @exocortexcom http://Exocortex.com - Passionate CG Software Professionals.
