As far as I know the common standard is the normal at the sample for Y, the projection of the first edge of the corner on the normal plane for X, and then the cross product between the two for Z, as a 3x3 matrix. In object space (like for local shapes) it's similar enough with the normal for the point, the projection of the first edge on that plane, and the cross between the two.
In texture space you like will need the former as pointperpoly is usually accounted for instead of artificially merged points. I can't promise this is how it works internally, it's just how I've done it in the past and what every (rare) piece of related doco in other softwares seems to work it out, but it's at least worth a try I reckon and see if you get a match. It should be relatively simple to at least try, generate a map, and then run a pixel diff to see variance. On Fri, Dec 6, 2013 at 10:18 AM, Ahmidou Lyazidi <ahmidou....@gmail.com>wrote: > have you tried to cross product the normal with the first neigbhor edge > direction? you can eventually do a quick propotype in ICE to see if that is > matching. > > Cheers. > -A > > ----------------------------------------------- > Ahmidou Lyazidi > Director | TD | CG artist > http://vimeo.com/ahmidou/videos > http://www.cappuccino-films.com > > > 2013/12/6 Daniel Brassard <dbrassar...@gmail.com> > >> Oops, should had read your last sentence, moving on. >> >> >> On Thu, Dec 5, 2013 at 8:21 AM, Daniel Brassard <dbrassar...@gmail.com>wrote: >> >>> Is a cross product between the normal and tangent vectors not giving you >>> the bi-normal vector? >>> >>> >>> >>> On Wed, Dec 4, 2013 at 6:50 PM, Matt Lind <ml...@carbinestudios.com>wrote: >>> >>>> Those calculations are for converting the normal in the normal map >>>> between color and euler space. The normal stored in the normal map needs a >>>> basis (ie: transform matrix) to act as a frame of reference so the renderer >>>> knows how to orient the normals in the map to the topology . That basis is >>>> derived from information in the Texture UV space, geometry normal, and >>>> Tangents and BiNormals vertex color properties. >>>> >>>> >>>> >>>> I need the algorithms for the computation of the basis, and I need to >>>> know how ultimapper compensates for the case when the binormals are not >>>> present. For example, does ultimapper assume the basis will always be >>>> orthogonal and do a cross product of the tangent and geometry normal to >>>> generate the binormal? >>>> >>>> >>>> >>>> >>>> >>>> Matt >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> *From:* softimage-boun...@listproc.autodesk.com [mailto: >>>> softimage-boun...@listproc.autodesk.com] *On Behalf Of *francisco >>>> criado >>>> *Sent:* Wednesday, December 04, 2013 3:36 PM >>>> *To:* softimage@listproc.autodesk.com >>>> *Subject:* Re: tangent/binormal space computation >>>> >>>> >>>> >>>> Hi Matt, >>>> >>>> will this help? >>>> >>>> >>>> http://softimage.wiki.softimage.com/xsidocs/tangents_binormals_SettingtheDataTypeforTangentsandBinormals.htm#Rga95703 >>>> >>>> >>>> >>>> Francisco. >>>> >>>> >>>> >>>> 2013/12/4 Matt Lind <ml...@carbinestudios.com> >>>> >>>> I need to write a variant of ultimapper to support our proprietary >>>> workflows, but our content (tangent space normal maps) currently uses the >>>> softimage standard. I need to know the algorithms Softimage uses so I can >>>> make an accurate conversion to our proprietary standards. >>>> >>>> >>>> >>>> Specifically, I need to know: >>>> >>>> >>>> >>>> - how the vectors are computed in the Tangents and BiNormal vertex >>>> color properties >>>> >>>> - how Ultimapper does a tangent space normal transfer when only a >>>> Tangents vertex color property exists on the hi res mesh. >>>> >>>> (eg; what assumptions does it make / what calculations >>>> does it do to compensate for lack of binormals?) >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> >>>> Matt >>>> >>>> >>>> >>> >>> >> > -- Our users will know fear and cower before our software! Ship it! Ship it and let them flee like the dogs they are!
