V4.1 BETA (experimental 4.1.17D)

[ajz3d]

Yeah...I thought that might be the same type of issue I was having previously. I tried for hours to get a decent result and nothing else worked...but that's just before Andrew added the Zones to the Bake Scan Settings Dialog. That's a good tip about the map size. Will have to keep that in my back pocket.

I'll report this problem on Mantis tomorrow.

Á propos zones and other bake scan settings, don't you think they should be unique for each retopo group?

---

 

Another bug. This time with Move tool (Through all volumes) and brush radius.

http://3d-coat.com/mantis/view.php?id=1388

 

Also, at least three times during one day I got a blank Save as... window, with only Save and Cancel buttons visible. I lost an hour of work once because of this. This bug seems to occur randomly and I wasn't able to reproduce it.

 

And what about the notorious problem with retopo vertices snapping to backfaces (in most cases backfaces have normals facing more or less 180 degrees away from polygons that face the camera, so WTF?) or snapping to underlying geometry when laying out retopo mesh vertices in "snap to outer faces" mode (when we retopo a floating geo for example)... Why on Earth does it happen in the first place? Smoothing with a Brush tool will almost always cause both issues to happen, and so will the Split edges tool.

---

 

I'm very sorry to say this, but I find this constant battle with various 3D Coat issues in all of its Rooms a little bit exhausting. I have this funny feeling that most of the time I spend with the program is wasted on finding workarounds to various problems that plague it.

 

I fully understand that this is a BETA version I'm using right now, but crucial functions for a 3D sculpting program, like correct normal and displacement map baking (anti-aliased), simply must work flawlessly in every version. The sample scene I provided is very similar to the example from the Happy Baking Guide, yet NM and displacement maps both bake with aliased edges wherever rays encounter a border of a floating geometry on their way... It's not a disastrous problem, but the question is: why a free program like xNormal can bake decent maps where a $379 worth software fails?

Edited January 20, 2014 by ajz3d

[AbnRanger]

I'll report this problem on Mantis tomorrow.

Á propos zones and other bake scan settings, don't you think they should be unique for each retopo group?

---

 

Another bug. This time with Move tool (Through all volumes) and brush radius.

http://3d-coat.com/mantis/view.php?id=1388

 

Also, at least three times during one day I got a blank Save as... window, with only Save and Cancel buttons visible. I lost an hour of work once because of this. This bug seems to occur randomly and I wasn't able to reproduce it.

 

And what about the notorious problem with retopo vertices snapping to backfaces (in most cases backfaces have normals facing more or less 180 degrees away from polygons that face the camera, so WTF?) or snapping to underlying geometry when laying out retopo mesh vertices in "snap to outer faces" mode (when we retopo a floating geo for example)... Why on Earth does it happen in the first place? Smoothing with a Brush tool will almost always cause both issues to happen, and so will the Split edges tool.

---

 

I'm very sorry to say this, but I find this constant battle with various 3D Coat issues in all of its Rooms a little bit exhausting. I have this funny feeling that most of the time I spend with the program is wasted on finding workarounds to various problems that plague it.

 

I fully understand that this is a BETA version I'm using right now, but crucial functions for a 3D sculpting program, like correct normal and displacement map baking (anti-aliased), simply must work flawlessly in every version. The sample scene I provided is very similar to the example from the Happy Baking Guide, yet NM and displacement maps both bake with aliased edges wherever rays encounter a border of a floating geometry on their way... It's not a disastrous problem, but the question is: why a free program like xNormal can bake decent maps where a $379 worth software fails?

That's a question only Andrew can answer. What I typically will do is try to record what is happening > upload the problem file (may want to zip it first, to keep the file size down), and if there is a real issue, Andrew will usually do his best to fix it. You may have heard this before, but it still holds true. I find the baking to be pretty good under most circumstances. You just have one instance where it may be problematic still. I'd be curious to see what Andrew has to say, when/should he take a look at the scene. I suspect it's because you have very thin overlapping parts.

 

The Voxel object also looks like it's rather low res. I would try to right-click on the layer > Extrude to thicken it a bit, and might add a bit more geometry to the Retopo mesh, to try and keep parts of the voxel object from poking through. You could always subdivide it once or twice when you merge to the paint room...to get a better bake, and then when you make your final export (out of 3D Coat), you can choose the LOW poly mesh export option, which will be the original state you see in the Retopo Room. In all of this discussion about Zones and different Retopo groups, we have no way of knowing if you have your baking cage too close or too far off the mesh. These are concerns in 3ds Max's Render to Texture engine, too.

 

You compare XNormal's result with 3D Coat's, but it's possible you set it up properly there, but not in 3D Coat. I typically find that if you adjust the cage properly, use the right mode, and set the Zones only in the problem areas, it comes out right. I might have a small spot here or there to clean up, but that's all.

Edited January 20, 2014 by AbnRanger

[wilson66]

You compare XNormal's result with 3D Coat's, but it's possible you set it up properly there, but not in 3D Coat. I typically find that if you adjust the cage properly, use the right mode, and set the Zones only in the problem areas, it comes out right. I might have a small spot here or there to clean up, but that's all.

 

Thing probably is, there is no such thing as 'zones' in xNormal. There are no extensive options to set. And when it still produces better results than 3DCoat, there has to be something awry with the code. 

 

Take a look at ZRemesher in ZBrush. It produces far better results without any guides than 3DCoat.

[AbnRanger]

Thing probably is, there is no such thing as 'zones' in xNormal. There are no extensive options to set. And when it still produces better results than 3DCoat, there has to be something awry with the code. 

 

Take a look at ZRemesher in ZBrush. It produces far better results without any guides than 3DCoat.

First of all, whether there are zones or not, if you don't have the overall baking cage set up correctly (goes for Render to Texture in 3ds Max, too), it's not going to give you great results. Zones are just there to help aid, when adjusting the cage is not quite enough. As for ZRemesher doing better than 3D Coat without any guides...that doesn't mean anything. Often times, no guides at all in 3D Coat gives better results that using guides.

 

Like I said, for most situations, it's working pretty good for me. You can see a clean baking result in the first video, and in the 2nd video, regarding Auto-Retopo...it still does some amazing work, but you can't just throw anything at it and expect perfect results. Just coming here and bashing the app cause it sometimes requires a little tweaking of parameters, isn't going to help matters. In fact, I sometimes get ready to report something "wrong" to Andrew, and upon further inspection, I realize I overlooked something or there was just user error, involved. It can happen...and I've been using the app almost daily for 6yrs now.

 

 

Edited January 20, 2014 by AbnRanger

[Andrew Shpagin]

Updated to 4.0.14

- 3D Lasso selection for freeze/vox/layer/vox extrude - see illustration - http://www.3d-coat.com/files/3dselection.jpg

Further it will be extended to other tools.

Actually this feature was important for our internal purposes, so it was inavoidable to do it, but actually it may be very helpful for everyone.

- Major scripting update, see manual (parametric dialogs, script - controlled strokes, retopo layers management, several examples)

- General cleanup and fixing Mantis issues.

4.1 expected soon, possibly next update will be 4.1 and then I will follow roadmap that I previously described.

[Andrew Shpagin]

Also, a few words about influence spheres.

Could you please add an option to toggle their visilibity on/off when previewing baking cage in "Bake scan settings" window? I currently have a scene with over 40 of them and its almost impossible to see the in/out depth preview through this crowd.

And by the way. Have you considered adding an option to load an editable cage that could be used as an alternative to influence spheres?

In next build there will be visibility icon.

In near future I will do depth to be brush based (actually volumetric)

[Carlosan]

ty !

[artman]

Updated to 4.0.14

- 3D Lasso selection for freeze/vox/layer/vox extrude - see illustration - http://www.3d-coat.com/files/3dselection.jpg

This is a very welcomed and important addition (I asked for this a long time ago I never expected it would happen one day.)

It improve usability of tons of tools in both paint and sculpting.

 

 

....be sure to really check into Mantis before releasing 4.1,I think there might be quite a lot of unresolved important issues forgotten there...

ex:Redo using Pose tool does not work.

 

edit:1)Also i agree anti aliasing 2x,4x.. in normalmap baking would be great.I also bake my normals in xnormal which is absurd.

2)Please make separation of voxel color layer by volumes optional,I got a file where I dont need it and 3DCoat creates 42 layers each time I bake...its a useful feature but when user dont need it can really be a nightmare.

 

Also please allow merging(combining) of objects in Surface mode for baking purposes,

I mean not booleans...just simple merging,keeping all self-intersecting parts as they are...with fast calculation...

Merging objects in SF mode is really superlong...sometimes hours...and does not work most of the time .

[spacepainter]

In 3D selection/voxlayer I can't get the selection cleared, only by returning to Epanel and reselecting 3d-select

O I see , the selection is only deleted, not the curve.

Edited January 20, 2014 by spacepainter

[JoseConseco]

+1 to Redo using Pose tool does not work.

Also cool trick that would be cool if it was feature:  when masking for pose tool, if you switch to sculpting brush - mask is gone. But if you switch to another sculpting layer, and then change brush, switching back to first layer - masking is preserved. It would be cool if masking from pose tool was working without need to switch to another layer. Tested in surface mode only.

 

Also could someone confirm bug with masking- it is giving me weird results (with simple pen pressure stroke type)

Edited January 20, 2014 by JoseConseco

[Javis]

Any word on the download speeds? It's been awful for over a month now for me (and others in the same region).

[AbnRanger]

+1 to Redo using Pose tool does not work.

Also cool trick that would be cool if it was feature:  when masking for pose tool, if you switch to sculpting brush - mask is gone. But if you switch to another sculpting layer, and then change brush, switching back to first layer - masking is preserved. It would be cool if masking from pose tool was working without need to switch to another layer. Tested in surface mode only.

 

Also could someone confirm bug with masking- it is giving me weird results (with simple pen pressure stroke type)

 

Yeah, the Pose tool is still messed up in more ways than one...been reported a few months ago, on Mantis, and nothing done about it. If you select the arm to pose > apply an FFD cage to it > select the controls points to manipulate/twist/bend the wrist > TRANSFORM, it affects the whole arm...not just the selected area. 3D Coat is applying a hidden gradient falloff from the selected portion all the way up to the end of the NON-SELECTED part. What's the point in that? If you select control points to manipulate, that is ALL you want to manipulate. This makes it impractical to use for Posing, in many cases.

 

Video report shown to Andrew:

http://www.screencast.com/t/KB8UXMFzJjL

 

Mantis Report:

http://3d-coat.com:8081/mantis/view.php?id=1322

 

I would not release 4.1 with the Pose tool not....well...Posing correctly.

Edited January 21, 2014 by AbnRanger

[AbnRanger]

Andrew if you do try to fix this before releasing 4.1, could you allow the user an option to reshape the FFD cage (so it better fits the object being manipulated)? Maybe an "Edit Cage Points" check box?

 

Also, there is inconsistency between the FFD cage in the Pose Tool and the Primitives tool. With the Primitives tool, you can arbitrairly set how many segments you want per axis. Not so with the Pose tool. You have to choose between a few presets. This is part of the inconsistency I was harping about a while back. Please make the options for the Pose tool FFD cage the same as it is in the Primitives tool. Thanks.

Edited January 21, 2014 by AbnRanger

[b33nine]

In the pose freeform tool, when you select "transform whole lattice" is it supposed to transform the mesh or just the bounding box? I would like to be able to transform the mesh with the transform gizmo but it won't do anything but transform the bounding box. Am I doing something wrong or does the tool not work that way?

[AbnRanger]

In the pose freeform tool, when you select "transform whole lattice" is it supposed to transform the mesh or just the bounding box? I would like to be able to transform the mesh with the transform gizmo but it won't do anything but transform the bounding box. Am I doing something wrong or does the tool not work that way?

With that selected, you can scale and rotate the whole lattice structure, but there is no ability to select the individual segments/point and transform those. That's why I'm asking Andrew to give us the option to do that. So, we can shape the lattice to fit the object more closely...than a simple rectangle. You can do this with the FFD deformers in 3ds Max.

[Andrew Shpagin]

Yeah, the Pose tool is still messed up in more ways than one...been reported a few months ago, on Mantis, and nothing done about it. If you select the arm to pose > apply an FFD cage to it > select the controls points to manipulate/twist/bend the wrist > TRANSFORM, it affects the whole arm...not just the selected area. 3D Coat is applying a hidden gradient falloff from the selected portion all the way up to the end of the NON-SELECTED part. What's the point in that? If you select control points to manipulate, that is ALL you want to manipulate. This makes it impractical to use for Posing, in many cases.

 

Video report shown to Andrew:

http://www.screencast.com/t/KB8UXMFzJjL

 

Mantis Report:

http://3d-coat.com:8081/mantis/view.php?id=1322

 

I would not release 4.1 with the Pose tool not....well...Posing correctly.

That is not right.There is my comment to mantis report:

 

Was you checking recent versions?

It works in completely different way. I spent a lot of time to fix this.

The issue:

http://www.screencast.com/t/KB8UXMFzJjL [^]

fixed completely. Now only selected area has influence and area that is far from selection is not affected. Have you noticed?

Of course selected cage subset has influence within range to nearest points but no more. It is logical. If you need just to rotate hand, why need to select whole arm?

 

[AbnRanger]

That is not right. I spent so much time to fix it. There is my comment to mantis report:

 

It still has a gradient falloff, between the selected points and the rest of the lattice. Please try it and you'll see. Me and Digman showed Raul in a Google Hangout session, and he said he would relay this back to you...but nothing ever came of it. Digman posted the Mantis report.

 

That gradient falloff makes it nearly unusable still. You can test this the same way in the video, on one of the Human primitives. If you select the points in the hand area, 3D Coat still affects the elbow and shoulder area. It was not fixed. Just now recorded yet another example showing it not working properly (4.0.13A)

 

https://docs.google.com/file/d/0B5zx7iMKxeu5SXJ0Rkl0ODVuMW8/edit

Edited January 21, 2014 by AbnRanger

[Andrew Shpagin]

It still has a gradient falloff, between the selected points and the rest of the lattice. Please try it and you'll see. Me and Digman showed Raul in a Google Hangout session, and he said he would relay this back to you...but nothing ever came of it. Digman posted the Mantis report.

 

That gradient falloff makes it nearly unusable still. You can test this the same way in the video, on one of the Human primitives. If you select the points in the hand area, 3D Coat still affects the elbow and shoulder area. It was not fixed. Just now recorded yet another example showing it not working properly (4.0.13A)

 

https://docs.google.com/file/d/0B5zx7iMKxeu5SXJ0Rkl0ODVuMW8/edit

I looked, and imo it works correctly. How do you suppose it should work? Now there is smooth falloff between selected points and non-selected - you see that only area between selected and first ring of unselected is deformed.  If there is no falloff (0 outside box and 1 in box) then surface will be just broken. Look explanation on the picture. If you have idea how falloff should look, please draw it.

[AbnRanger]

I looked, and imo it works correctly. How do you suppose it should work? Now there is smooth falloff between selected points and non-selected - you see that only area between selected and first ring of unselected is deformed.  If there is no falloff (0 outside box and 1 in box) then surface will be just broken. Look explanation on the picture. If you have idea how falloff should look, please draw it.

I would invert that curve, so that the falloff is very, very sharp. Can the user not use the same profiles as those for the brush pressure...especially the ones for the Spline Stroke Draw Mode? When adjusting the wrist (bend, twist, rotate, scale), there needs to be very little falloff. But there might be cases where you need more. If I were trying to do the same thing in 3ds Max, you could either adjust the cage (only) to pull down the nearest non-selected segement/loop....so that there is only a short area for falloff to occur.

 

Or you would use Soft/Gradient slection, which allows you to dial up or down the falloff range and even how much "Pinch" there is.

 

In this case, in the short term, maybe just supply a few falloff presets (Very Sharp, Sharp, S-Curve, Linear)?

Edited January 21, 2014 by AbnRanger

[artman]

Andrew,how do we make closed splines with new 3d lasso ?

 

btw it works pretty well but on my side when using extract tool selection is inverted.

Also when using Freeze tool the selection gets properly frozen but the entire object gets frozen too to a lesser degree.

Anyway,great addition really cool idea.

[AbnRanger]

Also, Andrew, would it be possible (after 4.1 is released) to apply CUDA acceleration to the Pose tool (using the CUDA 6 toolkit which has Unified Memory, now)...as you did with CUDA SMOOTH BOOST?

 

When you use it with "CONFORM RETOPO MESH"...it still can be very, very, very slow. I understand one can use the Multi-Res Workflow to help in some regard, but it's not always ideal/practical to do so (ie. Modeling and posing multiple objects simultaneously).

 

Remember...if one is going to use 3D Coat for modeling, the Pose Tool is the primary (model manipulation) tool. It's 3D Coat's Swiss Army Knife, and thus should be as robust as you can make it. The more it is developed and refined, the better 3D Coat is as a modeling solution.

 

Edited January 21, 2014 by AbnRanger

[Andrew Shpagin]

Andrew,how do we make closed splines with new 3d lasso ?

 

btw it works pretty well but on my side when using extract tool selection is inverted.

Also when using Freeze tool the selection gets properly frozen but the entire object gets frozen too to a lesser degree.

Anyway,great addition really cool idea.

They are always closed. Complete preview clutters viewport, so I am shouwing it non-closed.

but on my side when using extract tool selection is inverted.

 

What do you mean?

[Andrew Shpagin]

I would invert that curve, so that the falloff is very, very sharp. Can the user not use the same profiles as those for the brush pressure...especially the ones for the Spline Stroke Draw Mode? When adjusting the wrist (bend, twist, rotate, scale), there needs to be very little falloff. But there might be cases where you need more. If I were trying to do the same thing in 3ds Max, you could either adjust the cage (only) to pull down the nearest non-selected segement/loop....so that there is only a short area for falloff to occur.

 

Or you would use Soft/Gradient slection, which allows you to dial up or down the falloff range and even how much "Pinch" there is.

 

In this case, in the short term, maybe just supply a few falloff presets (Very Sharp, Sharp, S-Curve, Linear)?

From your words I still see no proper solution. Sharp falloff will just sharply distort surface.

What do you expect, how hand thansform should look?

[AbnRanger]

From your words I still see no proper solution. Sharp falloff will just sharply distort surface.

What do you expect, how hand thansform should look?

When rotating/scaling/twisting JOINTS, the falloff outside the selected points should be very short/small. All the deformation needs to occur strictly in the joint area....not half way up the arm. The current falloff curve is just too long...it's affecting parts no one intends to.

 

The Following would be some good Preset Falloff Curves (although it would be better to let the user create/adjust their own curves....but having a few presets is good either way). Linear, S-Curve, Sharp, Very Sharp

Edited January 21, 2014 by AbnRanger

[Taros]

+1 to Redo using Pose tool does not work.

Also cool trick that would be cool if it was feature:  when masking for pose tool, if you switch to sculpting brush - mask is gone. But if you switch to another sculpting layer, and then change brush, switching back to first layer - masking is preserved. It would be cool if masking from pose tool was working without need to switch to another layer. Tested in surface mode only.

 

Also could someone confirm bug with masking- it is giving me weird results (with simple pen pressure stroke type)

Check the masks, I guess there is one active. Maybe one from the paint room?

[JoseConseco]

Taros, I don't think so, since this happens in new file with default sphere and default mask pen brush. I checked this at work and  I have the same bug in here.  Just set mask brush curve to gaussian or sharp alpha and brush on sphere for 10 seconds with few strokes. I'm now 100% this is actuall bug. I will report it to mantis.

[AbnRanger]

Someone fussed about not being able to use/create/export Vector Displacement maps not too long ago. After some testing, I can't find a way to export them when you merge to the Paint Room, via Micro Vertex mode. But if you IMPORT a model into the Paint Room (MV), you can. What the heck?

[artman]

They are always closed. Complete preview clutters viewport, so I am shouwing it non-closed.

What do you mean?

You are right,they get closed anyway.. when making last points over first point yesterday it did not close,first point and last point would stay there on top of each other ...but this morning it it works fine. Very Strange.

(Please make it so that spline is not automatically applied when closing it,user might still want to Edit some points. Applying should only occur when doubleclicking/pressing Enter or Pressing Apply button.)

 

Also Today  Freeze works fine and Extract selection too.....again very strange as I used DefaultMale bust yesterday as well and did not change anything in the way I draw them.

Now only  issue I get is saving/loading spline give weird result when loading/applying.

there is kind of a weird double curve under it that was not there when saving.

 

I attached spline:

3dsavedspline.zip

[Malo]

I found something that confuses me.

 

How did Extrude (VoxTree -> rightclick on a layer) work or how did it calculate the size?

 

Take a look at this.

This is an object in Surface Mode with nearly 19.000 tris.

The ghost object is a copy of the main object and is extruded with 0.1

 

Now take a look at this.

Same object, subdivided in surface mode and switched into voxel mode with 1.3 million tris.

The ghost object is a copy of the voxel object and it is extruded with 2 to get the same result as i get with 0.1 in Surface Mode.

 

Why is there so a huge difference in the size of the extrude value?

I mean 2 is 2, why did i need in Voxel Mode a 20 time higher value to get the same result as i get in Surface Mode?

Wouldn´t it be better, if the Value would be the same in both modes?

Edited January 21, 2014 by Malo

[alvordr]

Malo,

 

It could be due to how the mesh is represented in either mode.  I never think of my mesh as being solid when in Surface Mode, as the name implies.  In Voxel Mode, however, I think of it as a solid filled with voxels, hence an increase in represented polys.