3D-Coat 3.2 updates thread.

[jwiede]

OSX build (ideally with fix to the color pick issue)? Please? Pretty please?

[AbnRanger]

OSX build (ideally with fix to the color pick issue)? Please? Pretty please?

No Soup for you! (Soup Nazi on "Seinfield")

[Andrew Shpagin]

OSX build (ideally with fix to the color pick issue)? Please? Pretty please?

Oh, please excuse. I had several crazy days. At least in Monday I will do new build.

Possibly earlier. Possibly even today if I will be able.

Do you mean crash as a picker issue?

[jwiede]

Right, I was referring to the crash that occurs when you pick a color off the object.

Glad to hear of the new build coming, thanks!

[Andrew Shpagin]

Right, I was referring to the crash that occurs when you pick a color off the object.

Glad to hear of the new build coming, thanks!

OSX Build uploaded too.

Please check V issue, I was not able to reproduce it on OSX.

[ifxs]

thanks for the OSX build Andrew, been looking forward to testing it :-)

[michalis]

Ptex is crashing on macosx, 16 threads xeon macpro. Here's the record. Fix it please, it works with few poly models only.

[michalis]

Now, this update crashes all the time, so voxels are working, I can export a triangulated mess (decimated), go to blender or zbrush and do the rest. I can't paint any textures or bake or UV or retopo . Its enough for me but very disappointing.

[Andrew Shpagin]

Updated to 3.2.04 (Win+Mac, Linux comes today or tomorrow)

Changes:

- Ptex export/import done in beta stage. You will be able to export/import color/specular/displacement separately. Ptex files are much closer to textures so I placed export/import ptx files in Texture menu

- Merging retopo room with reference mesh/painted mesh -> per pixel painting will respect color of the reference/painted mesh. In previous builds normalmap was baked correctly but color was not baked t per pixel painting if reference mesh was used.

- In Retopo->select tool Relax will act only on vertices selected in Retopo->Select->Verices. Snap button will work in the same way.

- Fixed render problem when rendering custom size frame was not properly centered.

- I fixed multiple problems with snapping in retopo tool. Problems was especially visible on the "creature" model.

- I fixed issue with crash in Linux while using 'V' key.

- I improved cylinder-like manipulator to avoid controls squashing.

- Fixed PSD export issue - incorrect (dotted) padding.

[ifxs]

!!! Andrew to the rescue again with another release! Thanks Andrew, cant wait to get into 3.2.04(OSX) !!! Did I mention 3DC rocks?!

EDIT#1: Thanks for the update Andrew... testing time ;-)!!!!! cant wait to use improved PTEX for microvert displacements!!!

EDIT#2: I just rant 3.2.04(OSX) PTEX through some quick tests with microverts, WOW!!! the ability to selectively up res areas of the displacement is soooooo very wonderful! CANT WAIT FOR 64-Bit OSX to push this to the limits!!!

[Andrew Shpagin]

!!! Andrew to the rescue again with another release! Thanks Andrew, cant wait to get into 3.2.04(OSX) !!! Did I mention 3DC rocks?!

OSX build is already there.

[Andrew Shpagin]

Linux build uploaded too.

[Maximus3D]

Great update Andrew, thank you very much! it's appreciated

/ Magnus

[thecoolmuseum]

in Ptex mode, Topological symmetry flip polygon UVs.

[AbnRanger]

Thanks for the diligence, Andrew...very big update.

[philnolan3d]

Very happy with the retopo snapping fix, my Armitage character model is now back in the works.

[digman]

The new local resnapping through selected vertices is a great addition. I preselect the vertices that will be resnapped,sculpt in that area, return to retopo and resnap. It goes give more control over what polygons get moved in the snapping process. I'm still am having problems though as stated in the bug report. This a great update, Thank you!

[Alan]

Updated to 3.2.04 (Win+Mac, Linux comes today or tomorrow)

Changes:

- Ptex export/import done in beta stage. You will be able to export/import color/specular/displacement separately. Ptex files are much closer to textures so I placed export/import ptx files in Texture menu

- Merging retopo room with reference mesh/painted mesh -> per pixel painting will respect color of the reference/painted mesh. In previous builds normalmap was baked correctly but color was not baked t per pixel painting if reference mesh was used.

- In Retopo->select tool Relax will act only on vertices selected in Retopo->Select->Verices. Snap button will work in the same way.

- Fixed render problem when rendering custom size frame was not properly centered.

- I fixed multiple problems with snapping in retopo tool. Problems was especially visible on the "creature" model.

- I fixed issue with crash in Linux while using 'V' key.

- I improved cylinder-like manipulator to avoid controls squashing.

- Fixed PSD export issue - incorrect (dotted) padding.

Thanks a lot for implementing ptex export/import so quickly!

I look forward to testing it out with prman this weekend.

[pbowmar]

Alan, FYI it's not clear how the polygons are mapped to the ptex tiles, the simple "poly 0 == tile 0" does not appear to work.

I've asked Andrew in another thread how the mapping is handled. So, don't get your hopes up

Cheers,

Peter B

[Active Rendering]

Hi Andrew,

really unbelievable your speed of updating! Great work.

Cheers,

Sascha

[Lawless]

Updated to 3.2.04 (Win+Mac, Linux comes today or tomorrow)

Changes:

- Merging retopo room with reference mesh/painted mesh -> per pixel painting will respect color of the reference/painted mesh. In previous builds normalmap was baked correctly but color was not baked t per pixel painting if reference mesh was used.

Hello Andrew,

You say you merged the retopo room with which room? I still see retopo room and paint room in my wife's copy of 3DCoat. I'm not sure I understand your meaning here.

-Brandon

[artman]

Hello Andrew,

You say you merged the retopo room with which room? I still see retopo room and paint room in my wife's copy of 3DCoat. I'm not sure I understand your meaning here.

-Brandon

The way its written makes it confusing but what he means is:

Merging from retopo room to per-pixel-painting over a Reference mesh will correctly bake the loaded texture of the Reference Mesh.

[Javis]

The way its written makes it confusing but what he means is:

Merging from retopo room to per-pixel-painting over a Reference mesh will correctly bake the loaded texture of the Reference Mesh.

That is correct sir. There was a major bug whereby merging/baking wasn't being completed for some channels (color/normals from vox/meshes).

[Alan]

Alan, FYI it's not clear how the polygons are mapped to the ptex tiles, the simple "poly 0 == tile 0" does not appear to work.

I've asked Andrew in another thread how the mapping is handled. So, don't get your hopes up

Cheers,

Peter B

Thanks for the heads up Peter. I didn't end up getting to test it out anyhow, but it looks like I wouldn't have got very far.

I'll track down your other thread and follow it.

[Lawless]

The way its written makes it confusing but what he means is:

Merging from retopo room to per-pixel-painting over a Reference mesh will correctly bake the loaded texture of the Reference Mesh.

Thanks!

[philippe leprince]

I made a quick ptex test using 3dc 3.2.04 Linux 64 and Pixar's PrMan 15.0.

Even though it works at first glance, it is currently un-usable with an external renderer.

Here is what I found :

ptex is only usable on subdivision surfaces (catmull-clark or loop), but PrMan and 3dCoat treat them differently :

• PrMan subdivides the faces at render-time using either a catmull-clark or loop scheme.
• 3dCoat pre-subdivides them with a different-looking scheme to allow you to paint on flat polys.
  • There should be an option in the import dialog to choose your subdivision scheme.

3dCoat exports ptex tiles per subdivided faces instead of per un-subdivided face.

• If a face has been subdivided once, the final ptex tile for that face should be made of 4 3dCoat faces stictched together into one texture.
• If you look at the ptex paint demo on youtube, you can see around 1:16 that faces are subdivided but represent a single texture tile.

There is currently no way to garantee that connectivity data will be preserved accross the pipeline.

• I'll explain :
  • Say you model something in Maya. Maya has its own data representation (vertex/face ids and connectivity).
  • You export that model to an obj file to import it in 3dCoat. The order in wich the vertices are written to this file is likely to change, compared to Maya's original structure.
  • You import the obj in 3dCoat : the vertex/face ids and connectivity data is likely to change again based on the way it is stored internaly.
  • You paint until you are happy, export the ptex files.
  • You plug that ptex file into your shader and the faces are all over the place, because the faceids / connectivity data are not the same as the original mesh.

  [*]The solution :

  • Import a fed files with the mesh so that 3dCoat may be able to either use the same ids/connectivity or remap it's own internal data structure when exporting the ptex files.
  • fed files are simple text files containing the connectivity info for each face in the mesh. They are described in the ptx SDK.

My test ptex files from 3dCoat looked incorrect color-wise, although they loaded fine in Pixar's ptxview utility.

I am still very interested in seeing this work in 3dCoat and remain available for further discussion / testing.

Thanks

Edited March 3, 2010 by philippe leprince

[ifxs]

Philippe: thank you for the thorough review of PTEX / PRMAN thus far.

[Andrew Shpagin]

Thank you Philippe for deep investigation!

Yes, now it works per subdivided face but I am changing it to work per initial faces.

And I am using Catmull-Clark subd scheme. Where do you see difference?

My test ptex files from 3dCoat looked incorrect color-wise, although they loaded fine in Pixar's ptxview utility.

How does it look? Red and blue channels swapped?

[philippe leprince]

Thank you Philippe for deep investigation!

You are welcome.

Yes, now it works per subdivided face but I am changing it to work per initial faces.

Great.

And I am using Catmull-Clark subd scheme. Where do you see difference?

I might be wrong. I will test a bit more tomorrow.

How does it look? Red and blue channels swapped?

It is a bit hard to tell.

I am attaching a couple of snapshots for you to see.

Thanks.

P.S. : The last snapshot is a display artefact I get when I turn on the low-poly view mode.

[ifxs]

I too get the display artifact on many PTEX merges(into scene) when viewing the file in low poly view. However for me, it usually only happens on one side or the other of a symmetrical mesh, hmm.

EDIT: I've also noticed some issues when backing to PTEX in regions with Tri's in the retopo room. I am well aware you should retopo with quads for PTEX, but I just wanted to see what would happen. I get polys that have vertices that "jump" quite far away from the surface, even when I turn off the displacement layer in the paint room, the actualy mesh has the "wild" vertices, thus causing flipped polys and faces. If I take that same retopo mesh with tri's and merge it to a PTEX scene without backing displacement, none of the vertices move in the wile manner. I'll see if I can get some images up soon.