I set up a bunch of Copy Instancing using “stamp(”../copy1“, ”instance“, 1)” on a switch node and “floor(fit01(rand($PT),1,10))” on the copy node variable “instance” value. All my Geos connected to the switch are connected in an orderly fashion from 1 to 9.

Beside this I then try to recreate the same with a Point-old instancing using the same “floor(fit01(rand($PT),1,10))” pointing at my instances, so “/obj/Geo`floor(fit01(rand($PT),1,10))`/”. As it's way faster rendering that way, i'm using Redshift.

For reasons i still ignore i don't get the same Geo distribution on my points with both methods.

When using the Point-old instancing method i can see the instance number in the Geometry Spreadsheet, but the copy stamp node on the other side does not create an attribute for that, so i don't know how to compare tthe result of my rand(@PT) which i would assume should be identical anyway.

Any idea what i'm missing to get the same results with both instancing and copy stamping please?

Cheers,

A.

If you wanted to compare the exact attribute values, you could do it by simply creating a point attribute and setting its value to the “instance” variable you're stamping, via the “stamp” expression.

That said, you're using a deprecated workflow. Rather than stamping variables, you should consider using a compiled for-each block to handle the switch between instances. I did a write-up [www.toadstorm.com] on this a while back that's a helpful overview for copying and varying instances, and it links to Jeff Wagners's much more in-depth tutorial [vimeo.com] on compile blocks if you need more details.

Also, there's a Redshift attribute you can place on a Geometry node that enables Redshift to read packed primitives as instances… try enabling “Instance SOP Level Packed Primitives” under the “Redshift OBJ > Instancing” tab. Might be easier than trying to go the old school instancing route.

toadstorm
If you wanted to compare the exact attribute values, you could do it by simply creating a point attribute and setting its value to the “instance” variable you're stamping, via the “stamp” expression.

That said, you're using a deprecated workflow. Rather than stamping variables, you should consider using a compiled for-each block to handle the switch between instances. I did a write-up [www.toadstorm.com] on this a while back that's a helpful overview for copying and varying instances, and it links to Jeff Wagners's much more in-depth tutorial [vimeo.com] on compile blocks if you need more details.

Also, there's a Redshift attribute you can place on a Geometry node that enables Redshift to read packed primitives as instances… try enabling “Instance SOP Level Packed Primitives” under the “Redshift OBJ > Instancing” tab. Might be easier than trying to go the old school instancing route.

Thanks for the insights. I'll look into it. Although i fixed my problem by adding a “Null node / Empty obj merge” into the switch (on the first input) above my copy stamp and now my copy node gives the same result as point-old instancing. Well the result was the same already, but the instance applied to each value was offset-ted because of the switch first input being “0” and not “1”, silly me :$

Thanks a lot for your time, i'll follow your links now and do some reading. Appreciated.