Thank you very much but honestly, if it's going to be too much of a pain please don't bother on my account. I have a suitable workaround for it in Houdini but it requires some user intervention and I am just trying to automate as much as possible. But, it's easy enough for users I think.
However, if you would still like to solve it, the issue occurs when any geograft is applied to the character as, due to the way they work, a geograph must hide at least one polygon. Male/female anatomical elements for example, but any geograft you happen to have in your content library will probably do it.
From memory, the issue with earlier versions of Sagan was that the vertex orders didn't match but the counts did. Whereas in later version the vertex orders do match but the counts don't, at least until you delete any isolated/unattached vertices resulting from geografts. Speaking purely for my own workflow, I can now automatically correct for any differences in vertex orders, but correcting for the isolated/unattached vertices requires some user intervention. So I guess, if i had to go with one over the other I'd pick the older method where the counts matched but the orders didn't nessecarily match.
But like I say, it's certainly not a showstopper for my project so please don't worry about it. I was mainly just curious if you were still working on Sagan but I know you have other, very cool sounding projects on the go as well.
Thanks
Peter
Hi Peter,
As one of the guys who atually slogged through the act of getting all this to work, I admire both your efforts and results, so I don't mind trying to figure this out.
Besides, I truly believe that I have some kind of low grade OCD because this gnaws at me even though you say you've got a workaround. In the DSON, there's another field that has something to do with associating pairs of vertices, which sounds kind of promising, but once again, the documentation is not what one would hope for. I'll ask David Vodhanel and Thomas Larsson directly.
Thank you very much but honestly, if it's going to be too much of a pain please don't bother on my account. I have a suitable workaround for it in Houdini but it requires some user intervention and I am just trying to automate as much as possible. But, it's easy enough for users I think.
However, if you would still like to solve it, the issue occurs when any geograft is applied to the character as, due to the way they work, a geograph must hide at least one polygon. Male/female anatomical elements for example, but any geograft you happen to have in your content library will probably do it.
From memory, the issue with earlier versions of Sagan was that the vertex orders didn't match but the counts did. Whereas in later version the vertex orders do match but the counts don't, at least until you delete any isolated/unattached vertices resulting from geografts. Speaking purely for my own workflow, I can now automatically correct for any differences in vertex orders, but correcting for the isolated/unattached vertices requires some user intervention. So I guess, if i had to go with one over the other I'd pick the older method where the counts matched but the orders didn't nessecarily match.
But like I say, it's certainly not a showstopper for my project so please don't worry about it. I was mainly just curious if you were still working on Sagan but I know you have other, very cool sounding projects on the go as well.
Thanks
Peter
Hi Peter,
As one of the guys who atually slogged through the act of getting all this to work, I admire both your efforts and results, so I don't mind trying to figure this out.
Besides, I truly believe that I have some kind of low grade OCD because this gnaws at me even though you say you've got a workaround. In the DSON, there's another field that has something to do with associating pairs of vertices, which sounds kind of promising, but once again, the documentation is not what one would hope for. I'll ask David Vodhanel and Thomas Larsson directly.
Donald
Thanks Donald, I really appreciate you taking the time to look into this.
I too suffer from OCD when it comes to these sorts of projects, like when I added days of work to my project simple because I didn't like Epics naming conventions for metahuman control curves, and so couldn't bring myself to use them :)
I naively assumed that the answers would lie in David's DazToUnreal fork as part of the ML Deformer code, but it doesn't look like it's in there, although I don't really know what I'm looking for.
Also, I meant to upload the screenshot of the geograft I tested in my last post, for reference.
Nice to know I'm not the only one bothered by little things more than I should be :)
Can you explain what I'm looking at? Because it looks like the hidden polygons have been removed, but not the resultant unused vertices. And which case was this, when the count matches, but not the order? And I notice that the vertices at the geograft's seam seem to match up with every other vertex of the g8m, so the g8m had been subdivided one time more that the geograft? In DS or Houdini?
Happily, I've got time to work on this more tonight.
Nice to know I'm not the only one bothered by little things more than I should be :)
Can you explain what I'm looking at? Because it looks like the hidden polygons have been removed, but not the resultant unused vertices. And which case was this, when the count matches, but not the order? And I notice that the vertices at the geograft's seam seem to match up with every other vertex of the g8m, so the g8m had been subdivided one time more that the geograft? In DS or Houdini?
Happily, I've got time to work on this more tonight.
Sorry for the late reply. For some reason I almost never get notifications from the Daz forum and so have to remember to check back on threads from time to time.
In the image below, the starting point was a sub-d 1 figure in Daz with the geograft applied. Nothing has been done in Houdini to alter the geometry in any way.
I've exploded the alembic geometry into the two shapes that get exported. I'm calling them shapes (not sure what Daz calls them internally, probably figures). The delineation is based on the "name" attribute.
The FBX and the Alambic shape 1 match perfectely both in terms of vertex counts and order.
So, ideally the Alembic shape 2 should not be exported from Daz at all.
Alembic shape 2 seems to be the original geograft geometry at it's base sub-d level as well as the unattached vertices of the figures geometry that get replaced by the geograft, also at the base sub-d level.
Understood about the notifications; I don't know why a get a few but miss the vast majority, like your response above.
OK, I think I understand. I think David's logic will filter out the faces of shape 2, but I'm not sure what to do with the loose vertices. Part of the problem is that I have never had a clear definition for what a Shape is, nor what an Assembly is. They're in thge object model but I'm not sure what they are for. I'll poke around.
I Installed the latest version of plugin (and previous as well) but I can't see it in Edit Menu. How to fix it?
See the note in the very first post:
Note: Currently, after installing, you must manually configure Sagan to appear in DAZ Studio's menu. Go to Windows | Workspace | Customize... and see the attached image. Basically, find Sagan on the left-side pane under Miscellaneous, and drag it to the desired location on the Menus tab of the right-side pane. Under Main Menu Bar | File, for example.
You may also want to right-click on Sagan on the left-side pane and select "Change Keyboard Shortcut" and choose something like Alt-S.
Does anyone know why I would be getting a failed export saying "vertex count changes"?
Thanks
Which figure, using what export settings, to which application, making what alterations, returning how, and with what settings? As i stands all we can do is paraphrase the message.
Does anyone know why I would be getting a failed export saying "vertex count changes"?
Thanks
Which figure, using what export settings, to which application, making what alterations, returning how, and with what settings? As i stands all we can do is paraphrase the message.
Thank you, Richard, I couldn't have said it better.
Stated like that, I have no idea. I'm going to need A LOT of details because that message is really just a sanity check that should never actually happen. Sagan figures out the geometry at the first exported frame, i.e. what the vertices and faces are, and assumes those values for all subsequent frames. That was part of the speedup from v2 to v3... it doesn't recalculate things that should be immutable, like the vertex count and the faces. Sagan assumes that the topology of subsequent frames remains immutable, and the animation is just the vertices moving around. The error you are encountering says that Sagan detected that that assumption was violated. The point is that the topology should not have changed, but somehow did.
Comments
Hi Peter,
As one of the guys who atually slogged through the act of getting all this to work, I admire both your efforts and results, so I don't mind trying to figure this out.
Besides, I truly believe that I have some kind of low grade OCD because this gnaws at me even though you say you've got a workaround. In the DSON, there's another field that has something to do with associating pairs of vertices, which sounds kind of promising, but once again, the documentation is not what one would hope for. I'll ask David Vodhanel and Thomas Larsson directly.
Donald
Thanks Donald, I really appreciate you taking the time to look into this.
I too suffer from OCD when it comes to these sorts of projects, like when I added days of work to my project simple because I didn't like Epics naming conventions for metahuman control curves, and so couldn't bring myself to use them :)
I naively assumed that the answers would lie in David's DazToUnreal fork as part of the ML Deformer code, but it doesn't look like it's in there, although I don't really know what I'm looking for.
Also, I meant to upload the screenshot of the geograft I tested in my last post, for reference.
Thanks again!
Hi,
Nice to know I'm not the only one bothered by little things more than I should be :)
Can you explain what I'm looking at? Because it looks like the hidden polygons have been removed, but not the resultant unused vertices. And which case was this, when the count matches, but not the order? And I notice that the vertices at the geograft's seam seem to match up with every other vertex of the g8m, so the g8m had been subdivided one time more that the geograft? In DS or Houdini?
Happily, I've got time to work on this more tonight.
Sorry for the late reply. For some reason I almost never get notifications from the Daz forum and so have to remember to check back on threads from time to time.
In the image below, the starting point was a sub-d 1 figure in Daz with the geograft applied. Nothing has been done in Houdini to alter the geometry in any way.
I've exploded the alembic geometry into the two shapes that get exported. I'm calling them shapes (not sure what Daz calls them internally, probably figures). The delineation is based on the "name" attribute.
The FBX and the Alambic shape 1 match perfectely both in terms of vertex counts and order.
So, ideally the Alembic shape 2 should not be exported from Daz at all.
Alembic shape 2 seems to be the original geograft geometry at it's base sub-d level as well as the unattached vertices of the figures geometry that get replaced by the geograft, also at the base sub-d level.
Hi Peter,
Understood about the notifications; I don't know why a get a few but miss the vast majority, like your response above.
OK, I think I understand. I think David's logic will filter out the faces of shape 2, but I'm not sure what to do with the loose vertices. Part of the problem is that I have never had a clear definition for what a Shape is, nor what an Assembly is. They're in thge object model but I'm not sure what they are for. I'll poke around.
Thanks for the example, it really helps clarify.
Donald
Hello everyone,
Quick question, is the plugin still works on new Daz3D update? Thank you so much :)
It should. I haven't heard any issues specific to the new versions of DS.
Can confirm it's been working fine for me on DS 4.22
I Installed the latest version of plugin (and previous as well) but I can't see it in Edit Menu. How to fix it?
See the note in the very first post:
Thanks!
Hello.
Does anyone know why I would be getting a failed export saying "vertex count changes"?
Thanks
Which figure, using what export settings, to which application, making what alterations, returning how, and with what settings? As i stands all we can do is paraphrase the message.
Thank you, Richard, I couldn't have said it better.
Stated like that, I have no idea. I'm going to need A LOT of details because that message is really just a sanity check that should never actually happen. Sagan figures out the geometry at the first exported frame, i.e. what the vertices and faces are, and assumes those values for all subsequent frames. That was part of the speedup from v2 to v3... it doesn't recalculate things that should be immutable, like the vertex count and the faces. Sagan assumes that the topology of subsequent frames remains immutable, and the animation is just the vertices moving around. The error you are encountering says that Sagan detected that that assumption was violated. The point is that the topology should not have changed, but somehow did.
Going to need a lot more details.
just a wild guess but is there an animated VDB in the scene or a Fluidos animation
something that loads geometry per frame?
Oh wow, Wendy, that's an excellent guess. I wonder if that's it.