Jeff Lait

jlait

About Me

専門知識
Developer

Connect

LOCATION
ウェブサイト

Houdini Skills

Availability

Not Specified

My Badges

SideFX Staff
Since 7月 2005

My Tutorials

obj-image Masterclass
Building a Fluid Solver in Copernicus
obj-image Masterclass
H18 Pyro | Building a Combustion Model
obj-image Masterclass
Vellum Cloth 更新点 | H18 Masterclass
obj-image Masterclass
64 ビット処理 | H18 Masterclass
obj-image Masterclass
Attribute Paint | H18 Masterclass
obj-image Masterclass
Vellum Cloth | H17 Masterclass

Recent Forum Posts

Feather vs SDF 2026年6月21日20:30

Yep, if your "feather" amount is a slower drop off than the antialiasing is, it basically just replaces the antialiasing outright.

Consider 1 dimensions
1 0.5 0 0 0
If this is feathered at 0.1 per voxel
1 0.9 0.8 0.7 0.6

Essentially the peak at "1" will overwrite all the antialiased values around it, so they become meaningless.

By contrast, the SDF process picks one value to declare as the boundary and measures from that. But this means if you have a big patch with a values of 0.25, it will disappear if you have 0.5 as your boundary.

Feather vs SDF 2026年6月19日15:36

This is somewhat frustrating as "Feather: Circle" _is_ using the same eikonal equation as the SDF building. The catch is that in the Font SOP we can use the 0.5 cut-off of the antialiased font to get a sub-pixel starting distance; but in the feather case it effectively will trigger off the on-off status of the 1 pixels. So your smooth distance from a bumpy pixelated border will have wrinkles in the normal :<

Unfortunately as Feather is meant to work with grey scale input, we can't use a thresholding trick as there is no universal threshold to pick.

High memory consumption in COP. 2026年6月4日10:29

jsmack
How is memory managed on platforms with unified memory? With no main memory to evict to, is that setting used at all? Is there any unnecessary copying of layer data when moving from cpu to gpu compute? I'm finding even with 128GB of unified memory, it can be pretty constraining, I can't imagine what it's like with only 16GB of VRAM.

On these platforms you'll want to leave the VRAM percentage to 100 as you don't want to ever "swap" to main memory. As you note, that will just unnecessarily copy it to a new location leaving you no farther ahead. Worse yet, if it is a read-only layer; next time it is needed for the GPU it will be copied to the GPU (and hence to normal RAM) and "helpfully" keep the CPU copy around. The same argument applies if you run in HOUDINI_OCL_DEVICETYPE CPU mode.

With 16gb of VRAM you'd likley have a similar experience, just that you'll be hitting the swapping to main memory - so you'd basically just have an "extra" 16gb of RAM to work with.