Jump to content
3DCoat Forums
Andrew Shpagin

3DCoat 4.7 (BETA testing thread)

Recommended Posts

On 5/31/2017 at 1:03 AM, Frimasson said:

I'm working on creatures sculpting for an advertising, in surface sculpting mode.
At some point, the director asked to add body parts (wings) 
I had to possibilities :
1 / To convert the mode to voxels, do booleans, then reconvert to polygons.
Even with a 160 000 000 voxels/poly models, a lot of fine details were lost
2 / try to  merge new parts with booleans
I have repetitive issues with merge booleans in 3dcoat. The bridge tool dosen't worked well either in this case.
I ended to do the boolean in blender with a 22 million poly model
Is it possible to have a more robust or permissive boolean in 3d Coat, that worked even ii a have to fix some holes, or mesh issues ? It is the main bottleneck in my workflow

Thanks

I tested AbnRanger's method and it does work, just like mine below maybe not in all cases but our answers are only to help with your problem the best we know how. Future improvements like a re-projection feature, we would all love to have.

The below is once you have all your revisions done and approved by the art director. You got the sign off on it.

 The method is not worry about doing a real boolean. I do not know how many layers you have so the example will be only two and assume you only want one retopo mesh and one retopo object layer.

In this example I am not using Names Correspondence as I have only one retopo object layer. Once you know if you can use this method then, I think you can figure out to use it with Names Correspondence.  

Clone a copy of the body and wing layer. Now hide the other layer two original layers, then merge visible with no true booleans. Convert this to voxels, you do not need all the super high detail here as this becomes object you to do your retopo mesh over, a more simplified version.

Once you have competed your retopo mesh,hide the simplified version and unhide the two original layers.

Do all the stuff for baking to the paint room, etc etc.

Once in the paint room paint out on the normal map any occlusion errors where the wing meets the body. This can be done in 3DC or Photoshop.

Now the above is a simple object being only two layers but it can be done on more complex objects. The retopo mesh where the wing and body meets needs to be real clean to be paint out any occlusion errors in the paint room or photoshop in a easy fashion.

I would test this method on some simple objects at first to get a feel of it.

I would love to see a re-projection of detail feature in 3DC. Where we can have a lower mesh and then re-project the high details to it from one or all layers. Maybe the transition area would need up some clean but that would be ok.

Will the above work out in all cases, I can not say but I have used it with success.

merge.jpg

Edited by digman
  • Like 1

Share this post


Link to post
Share on other sites
19 hours ago, ray said:

same issue I meet often, maybe a solution could be if 3D Coat could detect, when switching back from voxels to surface, which areas of voxels were not changed and use the original surface for these areas

Slow quality loss from switching is always an issue, also in the other direction, when working in surface mode for a while, and then going back to voxels.

Only when using Surface tools in voxel mode everything is fine, these functions keep the original voxels.


I nailed the problem, and try to find a quick way to fix it in 3d coat
Actually, i open the mesh in Zbrush, create auto groups by topology continuity, split the loose groups and delete them, closes holes, and re-export the model in 3dc

Share this post


Link to post
Share on other sites

when the boolean fail, most of the time, hole appear around the intersection
thantk for the advice, digman :)

 

In summary : closing holes in 3dc dosen't fix the problem and takes hours

A projection tool could be a nioe addition
A mirror tool to mirror only part of a model, could be a nioe addition too

hole.jpg

Edited by Frimasson

Share this post


Link to post
Share on other sites
On 6/2/2017 at 5:33 AM, Carlosan said:

There is a request ticket open about this issue, please feel free to add a +1. Thx

0002188: mesh booleans fail way too often.

Yes, a reworking of the surface mode boolean operation is much needed. 

I can merge surface layers into one layer without booleans. Export the model and use two other programs to do the boolean operations with success.

ZCore--- import into ZCore, switch to dynamesh with enough resolution to capture the details. The boolean operation is successful. Switch back to regular polygons and I can export this model back to 3DC.

MeshMixer--- Turn to solid and the boolean operation is successful. Export back to 3DC again.,

Having a more robust boolean operation would save us the extra steps.

Edited by digman
  • Like 2

Share this post


Link to post
Share on other sites

What of a switch to voxels that accurately calculates the needed resolution? Best workaround: plan ahead.

Share this post


Link to post
Share on other sites

The original idea of freeform sculpting was to allow the freedom of not having to worry about topology, experiment, and if the result was good enough, push it the extra mile to the finish line.

With this roadblock it's still a dream.
So plan ahead is not acceptable.

  • Like 3

Share this post


Link to post
Share on other sites
3 hours ago, Tony Nemo said:

What of a switch to voxels that accurately calculates the needed resolution? Best workaround: plan ahead.

The idea is good, but I think there's a good chance that the voxel density required for this kind of "lossless" conversion in some cases would be so high that it would lead to frequent system instability caused by workstations running out of RAM. Or viewport freezing on low end cards due to insane amount of triangles queued for rendering.

A total overhaul of 3D Coat's surface booleans system might be the only way to go. SideFX managed to pull off a similar task. They ditched their old cookie operator and completely rewrote boolean ops. The result is a boolean operator that I have yet to see failing a given task.

On my feature wish list for the sculpt room, this overhaul is a #1.

  • Like 1

Share this post


Link to post
Share on other sites
2 hours ago, ajz3d said:

The idea is good, but I think there's a good chance that the voxel density required for this kind of "lossless" conversion in some cases would be so high that it would lead to frequent system instability caused by workstations running out of RAM. Or viewport freezing on low end cards due to insane amount of triangles queued for rendering.

A total overhaul of 3D Coat's surface booleans system might be the only way to go. SideFX managed to pull off a similar task. They ditched their old cookie operator and completely rewrote boolean ops. The result is a boolean operator that I have yet to see failing a given task.

On my feature wish list for the sculpt room, this overhaul is a #1.

I believe that this is the only right way 3DC should continue. I've been testing SideFX since it's release and it is pretty interesting how they solved the ( Cache ) Part and I love it! With C4D it`s a beast  ( Zero Lags, bugs Yay!):) I would really like to see the Developers of 3DC to take action to what the community is asking for instead of leaving us in the cold. Sorry for being harsh, but you guys really need to focus on repairing the bugs and stop adding new ones. Still good job with your Renderman Plugin ( not needed, but ok. ) The 3DC community from Beginner to Professional can use your Product with a speed up workflow, but the Program needs some cleaning. I myself have slowed down fighting the frustrations with 3DC bugs and errors. It feels very slapped together and I am hoping for the day a clean and (True Stable Version ) to come out..

 

It would be real nice to see Pilgway grow as a Strong tree and not  melt with the wrong melting cheese.. ( if you understand the metaphor)

Edited by Speike-Styles
  • Like 2

Share this post


Link to post
Share on other sites
9 hours ago, ajz3d said:

The idea is good, but I think there's a good chance that the voxel density required for this kind of "lossless" conversion in some cases would be so high that it would lead to frequent system instability caused by workstations running out of RAM. Or viewport freezing on low end cards due to insane amount of triangles queued for rendering.

A total overhaul of 3D Coat's surface booleans system might be the only way to go. SideFX managed to pull off a similar task. They ditched their old cookie operator and completely rewrote boolean ops. The result is a boolean operator that I have yet to see failing a given task.

On my feature wish list for the sculpt room, this overhaul is a #1.

+ 1000

Share this post


Link to post
Share on other sites

hello,

It would be nice to have a release note for the Mac & Linux version too. It's hard to know if any of the bugs are solved (no loading bar, incremental rendering not working on Retina display, non-US keyboard not recognised).

Also, Is it possible to have an shader for emissive light ? (I know you can do it with smart materials, it would just be a little faster in some cases).

Share this post


Link to post
Share on other sites
23 hours ago, ajz3d said:

The idea is good, but I think there's a good chance that the voxel density required for this kind of "lossless" conversion in some cases would be so high that it would lead to frequent system instability caused by workstations running out of RAM. Or viewport freezing on low end cards due to insane amount of triangles queued for rendering.

A total overhaul of 3D Coat's surface booleans system might be the only way to go. SideFX managed to pull off a similar task. They ditched their old cookie operator and completely rewrote boolean ops. The result is a boolean operator that I have yet to see failing a given task.

On my feature wish list for the sculpt room, this overhaul is a #1.

I've been waiting for 4+ yrs for Sculpt Layers, and it has one of the highest levels of community support on Trello, and even this forum. Andrew is supposedly focused on the Paint workspace, so who knows when he'll ever get back around to the Sculpt and Retopo workspaces.

Share this post


Link to post
Share on other sites
On 5/26/2017 at 0:22 PM, tartinebrule said:

Still no luck producing a build for Linux that would finally allow us to be on par with windows in terms of feature and bug fix?

1
come on... I made it easy for you, just copy paste in a terminal :
 
gcc -o 3dcaot 3dcoat -I

Thanks!

T.

Share this post


Link to post
Share on other sites

Think I found a new bug in 4.7.29 with the brush/aitrbrush tool in the paint room. If I am using the brush tool in stamp mode and I apply color/normals to my model in the texture editor view (2D view) it seems to level artifacts all over before I let go of the mouse button. Trying to hit the undo key does not remove these artifacts either. I have to manually paint them away. 

Share this post


Link to post
Share on other sites

Hi I happend to find some metalness, and color (Alberto) seems not restrict correctly, with icon or layer option.

1 load obj in Paint room,  I set new layer for metalness , color, and roughness.  

2 I set metalness (black and white) on metalness layer, (color wihte some part), set all other opacity as zero. but metalness opacity as 100% about this layer.

3 In texture editor, I set view as color. to see diffuse color texture. I believe it should be perfect newtral grey, but actually the color seems effected by metalness too.

4 Then I try to erase "color only"  from Top menu>Layers>erase unfrozen part., but set Normal and glossy channel disable. (by icons) to restrict erase only about color.

5 now it seems remove "color" from metalness layer, but at same time it remove all metalness white color. then fill black again.

through this step, I set layer visible only about metralness and Layer zero. then other layer cause no effect.  and I only modify texture on metalness lyaer.   If  metalness will not be separated from color channel?  

Edited by tokikake
add detail more

Share this post


Link to post
Share on other sites

Ah sorry I could confirm, when I export color map it seem not effected by metal layer. :blush:

I believed 2d texture editor with color mode, may show color which not effected by Metalness,(alberto)  but it seems count metalness as same as smoth shading, then it not matter.

(though I hope if there will be option (alberto color only option), to quick see current alberto map which not effected metalness and roughness, without change metalness opacity,.

Share this post


Link to post
Share on other sites

4.7.30 [beta]

- Possibility to change resolution of texture, attached materials will be automatically resmpled.

- a lot of problems related to attached materials fixed

- correctly updated material ball in layer attachment preview.

- square alphas support.

- Problems of paste UV fixed.

  • Like 7

Share this post


Link to post
Share on other sites
10 hours ago, Andrew Shpagin said:

4.7.30 [beta]

- Possibility to change resolution of texture, attached materials will be automatically resmpled.

- a lot of problems related to attached materials fixed

- correctly updated material ball in layer attachment preview.

- square alphas support.

- Problems of paste UV fixed.

Thanks a lot for the square alpha support. :good2:

Folders for the Preset panel... hint hint... LOL...

Edited by digman

Share this post


Link to post
Share on other sites

4.7.30 [beta] on Mac

Shaders (except in "default" and "skins" folders) are not working anymore. When I try to use them, it display the default shader.

3Dc can't find Renderman.

Share this post


Link to post
Share on other sites

4.7.31 [beta]

- better support of square alphas. Less clipped at edges.

- All tools are correctly compatible with layers attacments.

- merge down, apply blending are working correctly with "Replace depth" and materials attaching.

  • Like 3

Share this post


Link to post
Share on other sites

"better support of Square Alphas, Less clipped at edges"

Good news as I was just going to post my test results showing the clipped edges problem. I will test today later and report back here with the results.

Share this post


Link to post
Share on other sites

Select Layer0

Attach any smart material - the layer is visually filled

Now detach the smart material - the layer keep filled by the smart material.

---------------------------------

In the download file only the version GL is v31. DX still is version 30

  • Like 1

Share this post


Link to post
Share on other sites

I'M downloading .31 because I got this with .30:

2017-06-10_1315.thumb.png.5d6c32fe26a2dcc2ad421a36f8e474e0.png

Painting caused a color change from 'skin' to purple and white (the colors are 'X' out but hard to see in this pic).

Share this post


Link to post
Share on other sites
Guest
This topic is now closed to further replies.

×