magneto Posted September 4, 2012 Share Posted September 4, 2012 (edited) Hi, I don't seem to see any change when I use the Recompute Point Normals option of the Transform SOP. I see that the viewport shows some difference in length of the normals when I toggle this option, but I assume this is an artifact in the new viewport? Couldn't try the old viewport because of a crash. But when I compare the normal values after a transform, the Details View show no change when this option is turned on or off, which means the normals do not change right? Anyway here is the picture where the first picture is the original Grid, the second picture shows the result of Transform SOP with rotatez set to 90, 3rd picture is the same, except the Recompute Point Normals option is turned off. Different results in the viewport but the numerical values shown in the Details View are the same. So what does Recompute Normals do exactly? Edited March 4, 2013 by magneto Quote Link to comment Share on other sites More sharing options...
edward Posted September 4, 2012 Share Posted September 4, 2012 Your example is very hard to tell the difference because the Transform SOP will also rotate the N attribute at the same time. Try just pulling a singly polygon on the grid upwards. 1 Quote Link to comment Share on other sites More sharing options...
magneto Posted September 4, 2012 Author Share Posted September 4, 2012 (edited) Thanks Edward. Do you mean before transforming the grid? I did it like that and here is the result: Basically it shows before the transform and after, but the 2nd image is the same for both states of the Recompute Point Normals option. The details View shows numerical change for 1-2 points for the last digit, so like -0.652345 vs -0.652346. Edited September 4, 2012 by magneto Quote Link to comment Share on other sites More sharing options...
petz Posted September 4, 2012 Share Posted September 4, 2012 try to select and ROTATE just one point of the grid and you´ll see the difference. hth. petz 1 Quote Link to comment Share on other sites More sharing options...
magneto Posted September 4, 2012 Author Share Posted September 4, 2012 Thanks petz, I saw the difference now. So Recompute Point Normals also changes normals of the neighbours of the input Group? Also it looks like other normals are changed slightly, not by the same transform, but relative to the actual points in the input Group. Quote Link to comment Share on other sites More sharing options...
magneto Posted September 5, 2012 Author Share Posted September 5, 2012 Just to follow up on this, my observations are these. Please correct me if I am wrong or just say yes if I am correct anyway It seems like the point normals (N) are always transformed, whenever you transform points. This is due to point normals being Vector attributes, thus are always transformed. The "Recompute Neighbour Point Normals" option in the Transform SOP basically allowing you to separate this behaviour in case you don't want the point normals transformed. Quote Link to comment Share on other sites More sharing options...
edward Posted September 5, 2012 Share Posted September 5, 2012 "Recompute Point Normals" says to recalculate the N values such that they are the "average" [1] of the point's adjacent face normals. (as opposed to being "transformed" which means to take the N values and multiply each one by the inverse transpose of the transform matrix) 1. Well, "average" if the faces were coplanar. 1 Quote Link to comment Share on other sites More sharing options...
magneto Posted September 5, 2012 Author Share Posted September 5, 2012 Thanks Edward, it's crystal clear for me now Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.