Hi,

I have a need to flatten the uv's of many islands. They are hair cards and there could be several thousand of them.

The UV Flatten node works perfectly for this however it's very slow.

As far as I can tell, most of the slowness seems to come from the packing part of the procedure.

I don't actually need the uv islands to be packed for my workflow.

However, there doesn't seem to be a way to disable the packing part of the UV Flatten node that I can see.

Does anyone know of a method to flatten the uv islands without packing them afterwards?

I'm hoping that if I can find a method for doing this, it may be much faster for my workflow.

Thanks

Hi,
the uv flatten does not pack uv islands as long as you keep the layout constraints on manual layout and dont provide any seam group.
I'm not sure what your setup is, but as you can see in the figure below, the box islands are not packed and do overlap.

And for such a workflow, the better approach is to create first the uvs for the hair card sample and then scatter them with the uv included or transfer the uvs from the sample to the scattered final hair cards.

vik_lc
Hi,
the uv flatten does not pack uv islands as long as you keep the layout constraints on manual layout and dont provide any seam group.
I'm not sure what your setup is, but as you can see in the figure below, the box islands are not packed and do overlap.

I do have Enable Manual Layout enabled and aren't providing any seam groups, however it still spends ages packing.

The only difference is that I need to have Rectify Group set to "*" in order to have the uv islands straightened out.

vik_lc
And for such a workflow, the better approach is to create first the uvs for the hair card sample and then scatter them with the uv included or transfer the uvs from the sample to the scattered final hair cards.

I agree completely however I didn't make these hair cards. They are assets from the Daz store. Some have good uv layouts and some don't.

I found that if I put the UV Flatten node inside a compiled For Each block, running over a single connected primitive per loop, it finished MUCH faster. Presumably because it doesn't have to do any recursive packing?

Have you tried checking Preserve Seams and Preserve Layout?

Thanks Tomas.

Checking those options does prevent the packing phase from occurring