Jump to content
3DCoat Forums

3D-Coat 3.7 updates thread


Recommended Posts

  • Member

Personal comment (not speaking for pilgway !)

If you think as opencl as a way to get hardware acceleration from gpu on ati cards, think again. I won't go into details but think opengl and ati cards and you'll get an idea of what I mean.

I know it's not a perfect solution but at least it exists for ATI cards, Cuda is closed to only Nvidia hardware, I would prefer an open platform that might still need better implementation over one that is tied to one companys product. OpenCl and support for it on different hardware will not grow without demand for development. But like I said, I'm not a programmer, I'm just expressing some wishful thinking.

Link to comment
Share on other sites

  • Advanced Member

Lack of UVs is a mixed blessing. There are benefits but also downsides like not being able to work on it in 2D (including the link to Photoshop). You can increase res even better with the LiveClay tools, then it's exactly where you need it. To export it you still need to retopo, uv, and merge to the Paint room from the Retopo menu. It's really just the same as before when merging with one of the shaders.

I actually never did that before either. I have only really tried voxels without painting and did not have much luck with retopo so just exported the decimated meshes automatically. I'll see if I can figure it out.

You don't need to voxelize your mesh, you can import it as a surface mesh (merge into scene and click "without voxelizing" , you'll keep your mesh as-is.)

About 1) that won't we relevant soon, Andrew just made resample able to kee^p color, meaning you can increase/decrease polycount without affecting (at least when going up) too much your paint.

I didn't know about that - thanks. I suppose that it would still entail the same export/retopo process Phil mentioned above - or does this method retain the geo and UV's of the original mesh? If so is there a guide somewhere on how to bake the polypainted texture back?

In the grand scheme of things, I think it's a lot like Ptex was...seemed to have a lot of potential, but in every day workflows, not a lot of value. I just never find myself using Ptex as it's too limiting, and the UV layout process in 3D Coat is so fast...it seems like a good object lesson for Andrew to learn from. Just because something seems to have a benefit here and there, doesn't mean it's worth the development time invested. When viewed against the ability to paint in both 3D and 2D mode (regular painting), Voxel Painting will lose it's shininess before long.

What will impact every user, every minute or their workday is instead of pursuing shiny new novelties like this and Ptex, focus on raw application performance and refinement, and do not stop until you have become the gold standard in the industry. It might not be easy, but nothing ever truly lasting and beneficial is. I know Andrew is very capable of it, but it will take a major shift from the mad rush to crank out as many features as possible, to making the application as refined and polished as possible.

You have a tendency to make your point a bit forcefully and it's rubbing people the wrong way, but I don't disagree with the basic idea that fundamental tools should be fully functional before adding new stuff. I also actually agree with the premise behind your point that CUDA acceleration could be improved. Andrew said that he could get 20% improvement but didn't seem too keen, but I actually think that is a very significant improvement and would be well worth the investment. For regular users that can add up to a massive difference in production time. Now, CUDA is less interesting to me as I am a paint user, not a voxel user, but the point that such large improvements could be made, but aren't seen as a priority, is a problem in my view as well.

Regarding PTEX, I agree with the others who posted on the subject. The problem isn't that PTEX isn't useful, it's the lack of render support that holds it back in my workflow. Better support is coming and with that I would happily switch over to PTEX and forget about PP and MV painting (well, assuming the performance in the paint room was improved to make it more useable in hi-res).

/b

Link to comment
Share on other sites

It's cool, but it doesn't give me any real benefit over what I could do before.

I will have to disagree with that. This new option is so much faster for painting. Even in my first few tests I was able to do things that I couldn't do before because the Paint room performance couldn't handle it. So that alone is a huge benefit, not to mention you can now continue sculpting after you've started to paint, this is also a big benefit. Yes, if you are using Microvertex you can sculpt to a certain extent but when I tried even making some small changes on my current project it caused me big headaches. Even then, what if you want to use per-pixel?

I know this will probably get shot down by most users who favor Cuda or who have already invested in nVidia hardware but I think it might be a bit more benificial for Andrew to develop gpu acceleration for 3d-coat using opencl. I know this would mean more work for Andrew but if he could get an opencl build optomized for Nvidia cards first then work on adding optomization for AMD cards he would draw attention from a larger group of users.

I think Open CL would be a waste of time IMO. Cuda has been around longer so it's more mature, plus nVidia has opened it up now so even ATI can implement it.

Link to comment
Share on other sites

  • Reputable Contributor

As always there are alot of users and many different workflows. I do agree that before Version 4 is released that Andrew spend a few months on fixing and improving areas that need it. Getting rid of bugs etc. No need to state them again here in my post as they have been addressed already a number of times.

Andrew could put out an official guide for testing a version 4 beta and RC release. This way the community could have a guide to go by and Version 4 would be a solid release. This creates a very, very strong sales point for new users and upgrade for current and past users...

Link to comment
Share on other sites

  • Member

I think Open CL would be a waste of time IMO. Cuda has been around longer so it's more mature, plus nVidia has opened it up now so even ATI can implement it.

I don't think Nvidia has opened it up quite as much as I would like, It appears the "open" only comes from there adoption of an open compiler. Nvidia still has control over who can get the source so I don't think they will open things up to a competing hardware vendor. For power users who take advantage of cuda for scientific and medical computing this is great news but for software developers OpenCl is still the most portable platform. We may see users develop methods for running cuda applications on ati hardware though?

Link to comment
Share on other sites

  • Advanced Member

So getting a model into the surface mode and painting on it is pretty easy, but how do I get that painting out? I tried baking but it only baked out the base shader colour, not the painting. Any tips?

Also: it was mentioned that you can increase the res of the model in surface mode (using LC tools I believe) but I cannot find a way to just bump up the res. Where is that tool located? I tried the resampling thing, but that seems to do something quite different.

tks

/b

Link to comment
Share on other sites

  • Member

Would be nice to either have an option for live clay brush to behave like *MudBrush* or have mudbrush have a liveclay option. At the moment Liveclay brush simply wrecks any ridge detail. See my pic for clarification.

Particluarly noticable after 2+ strokes with LC. It bumps out top of the ridge disproportionately, wrecks the side of the ridge and the area where the ridge joins the surface, whereas mud feels great but unfortunately is not LC enabled.

This is pretty much the only(but at the same time the biggest) reason i still prefer Sculpris over 3DC for sculpting with dynamic tesselation. LC brushes are very unpredictable/destructive.

Also zbrush navigation more does not work with some brushes and there's still no screenspace rotation(the one that's actifacted by holding shift rotating then releasing shift). It's an extremely useful navigation feature i keep waiting to see. In mudbox it's standard cakked trackball navigation(dont need to hold down any button there it's on by default)

post-4167-0-02103800-1327205501_thumb.jp

Link to comment
Share on other sites

  • Advanced Member
Read back my sentence. Its pretty obvious that Im implying Ptex can do that too.

You're right, I misinterpeted your comment. Too many late nights getting Windows and all my apps set up again lol. ;)

My point was that Ptex require displacing a lower res topology and its very hard to achieve good displacemnt in 3Dcoat unless topology is has absolutly no tricky areas.This is why vertex painting is much straightfoward...you can paint right away,without a projected carcass mesh.

You're still going to need a low res cage for many applications though. Rendering just isn't all that great in 3D Coat and the ones that are don't handle a lot of polygons very well (though they are slowly improving). Vertex painting is cool, but only so long as you can bake to multiple map types for export, in which case UV's of one sort or another will still be required. This is why I asked about baking vertex painting to ptex earlier. This path will make life a lot easier once ptex becomes better supported, provided I've understood it all correctly (obviously not a guarantee). I see Andrew has stated going in the other direction will be possible at some point in the future too (convert map to vertex paint), so it seems he does have his eye on this area at the moment. Maybe he'll speak up on this, provided he has some time to spare from being a one man army lol. :)

Updated to 3.7.03C

- New tool intodiced - layer. It is intended for medical application, but will be vy useful for cloth, hardsurface modeling and many other. See example - http://3d-coat.com/files/VoxLayer.jpg

I'm having a lot of fun with this tool. It allows me to create interesting, complex geometry very quickly and use primitives as the scaffolding. I have a question (possibly a request) though. Is it possible to sharpen/blur the selections you make, like with masks in ZBrush? I'm having a lot of trouble getting clean edges with VoxLayer and Smooth All isn't all that helpful, especially if one doesn't want soft rounded edges. If not, perhaps simple plus and minus buttons in the dialog box would do. A check box for getting rid of all selections after clicking apply would be helpful too. It's easy to forget they're still there sometimes, especially when their hidden behind the new geometry, and it can result in extra unnecessary extractions happening (increasing calculation time) which overlap and may be hard to notice until it's too late.

Speaking of the Smooth All button, a Sharpen All button would also be helpful. Something like ZBrush's ability to polish, like when remeshing a dynamesh. I think I may have found a glitch with Smooth All as well. Try loading the third sphere from the startup screen, hit the Res+ button twice, then hit the Smooth All button a couple of times. You should see a whole lot of stair stepping emerge. I discovered this when using the sphere as scaffolding for creating some VoxLayer stuff. I was getting stair stepping in my extractions and figured smoothing the sphere first before drawing selections on it would solve the problem, but it only makes things worse. They can be smoothed out using the smooth brush, but this is rather laborious and time consuming. I'm still new to 3DC so if anyone can point out mistakes I'm making or features I'm overlooking, go right ahead and do so please.

PS: I actually have a growing list of problems and things I'd like to see, but I'll leave it at that for now. Hopefully this is the right thread to be posting in regarding such issues.

Link to comment
Share on other sites

  • Member

I'm having a lot of fun with this tool. It allows me to create interesting, complex geometry very quickly and use primitives as the scaffolding. I have a question (possibly a request) though. Is it possible to sharpen/blur the selections you make, like with masks in ZBrush? I'm having a lot of trouble getting clean edges with VoxLayer and Smooth All isn't all that helpful, especially if one doesn't want soft rounded edges. If not, perhaps simple plus and minus buttons in the dialog box would do. A check box for getting rid of all selections after clicking apply would be helpful too. It's easy to forget they're still there sometimes, especially when their hidden behind the new geometry, and it can result in extra unnecessary extractions happening (increasing calculation time) which overlap and may be hard to notice until it's too late.

Are you after clean edges with VoxLayer like this?

voxellayer.jpg

  • Like 1
Link to comment
Share on other sites

Andrew needs to correct me if I am stating something inaccurately, but, I believe the development push, currently, with 3DC is to rely less and less on custom, powerful hardware and more and more on efficient software routines to produce the speed and fluidity of painting and sculpting that most users are asking for.

This is the reason for spending so much development time on technologies like LiveClay, and now Direct Painting.

If even an average hardware setup can accomplish this kind of fluidity, then 3DC becomes more attractive to all types of users - not just the small numbers of "power users" that now use competing products requiring equally powerful hardware configurations.

Greg Smith

  • Like 1
Link to comment
Share on other sites

  • Reputable Contributor

Andrew needs to correct me if I am stating something inaccurately, but, I believe the development push, currently, with 3DC is to rely less and less on custom, powerful hardware and more and more on efficient software routines to produce the speed and fluidity of painting and sculpting that most users are asking for.

This is the reason for spending so much development time on technologies like LiveClay, and now Direct Painting.

If even an average hardware setup can accomplish this kind of fluidity, then 3DC becomes more attractive to all types of users - not just the small numbers of "power users" that now use competing products requiring equally powerful hardware configurations.

Greg Smith

One of my computers qualities for the above statement... :o My other one runs linux and is faster and more powerfull.

LiveClay and Polypainting helps the single core old lady of a computer to keep zipping along. I might keep her just to test how 3DCoat works even on a computer that by today's standards has both feet in the grave... ;)

Link to comment
Share on other sites

Andrew needs to correct me if I am stating something inaccurately, but, I believe the development push, currently, with 3DC is to rely less and less on custom, powerful hardware and more and more on efficient software routines to produce the speed and fluidity of painting and sculpting that most users are asking for.

This is the reason for spending so much development time on technologies like LiveClay, and now Direct Painting.

If even an average hardware setup can accomplish this kind of fluidity, then 3DC becomes more attractive to all types of users - not just the small numbers of "power users" that now use competing products requiring equally powerful hardware configurations.

Greg Smith

That is the general impression I was under as well.

Link to comment
Share on other sites

  • Advanced Member

So getting a model into the surface mode and painting on it is pretty easy, but how do I get that painting out? I tried baking but it only baked out the base shader colour, not the painting. Any tips?

Also: it was mentioned that you can increase the res of the model in surface mode (using LC tools I believe) but I cannot find a way to just bump up the res. Where is that tool located? I tried the resampling thing, but that seems to do something quite different.

tks

/b

Just bumping this. Anyone able to let me know? I didn't have much luck and ended up going back to just using PTEX.

tks/b

Link to comment
Share on other sites

  • Advanced Member

I dont have time to get into it now as its my bed time, but I believe I responded to it several posts ago.

Yes, I was asking for some clarification because "using LC tools" to increase the res didn't get me too far. Perhaps I should have been more clear on the other question: even if you merge geometry in using surface mode, thus maintaining the geometry, it's *still* necessary to go through voxels and retopologize and UV etc to get the direct painting out?

Link to comment
Share on other sites

In most normal cases yes, you still need to do the retopo, and set up a uv map, then bake (merge) to the paint room and export from there. You can just export the voxel model, but most of the time it's too dense and doesn't have good clean topology for animating with.

Link to comment
Share on other sites

  • Advanced Member

In most normal cases yes, you still need to do the retopo, and set up a uv map, then bake (merge) to the paint room and export from there. You can just export the voxel model, but most of the time it's too dense and doesn't have good clean topology for animating with.

Thanks Phil.

So another question then: rather than bake to a retopo model, is it possible to bring the original geometry back in and bake from voxel to that? Then I won't have to rebake it yet again to get the maps back onto the original mesh, which would be the plan in most cases.

/b

Link to comment
Share on other sites

  • Advanced Member

Absolutely. You just import it (original low poly model, created outside of 3D Coat?) in the Retopo Room. It should come in exactly at the same position and scale as the merged voxel object. If you had UV's on the original model it will show up in the Retopo Room, so you don't have to do them again.

Cool - thanks!

/b

Link to comment
Share on other sites

  • Advanced Member

So, I tried the OSX build. I merged the "creature" I jumped in paint mode, I tried to paint and OSX was dead. Just like this. No force quit or anything. Just a hard reboot. (in safe mode)

Can you please check this? Before we start arguing with Andrew or anybody else about my cynical way of speaking.

Under 10.6.8, nvidia graphics.

Link to comment
Share on other sites

  • Advanced Member

Is there a basic "subdivide" function in surface mode? I was directed to LC tools to increase res, but I cannot find any way to do it aside from 'resample' which seems to do something totally different. Is there a way to do it without actually changing the topology with one of the mesh?

Thanks /b

Link to comment
Share on other sites

  • Advanced Member

Is there a basic "subdivide" function in surface mode? I was directed to LC tools to increase res, but I cannot find any way to do it aside from 'resample' which seems to do something totally different. Is there a way to do it without actually changing the topology with one of the mesh?

Thanks /b

One way is to use a LC brush with zero depth and set the resolution in the Tools palette, but it would be nice to have a global increase/decrease res in surface mode.

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
 Share

×
×
  • Create New...