Jump to content
3D Coat Forums
Taros

Twitter discussion: What Andrew is currently doing.

Recommended Posts

From my conversation with Andrew, I was under the assumption it was up/down res like ZB/MB, for voxels. I believe it was using surface mode in some way as well.

Share this post


Link to post
Share on other sites

"Primitives cage can be tweaked using not only points but edges and faces too."

This is nice, I was just wishing I could move two adjacent points at once, like I would select both and move in LightWave. Selecting an edge takes care of that.

Share this post


Link to post
Share on other sites

I'm wondering if Surfacing tools/mode aren't being worked to provide a similar workflow and alternative to ZB's and MB's, as I thought I saw a comment on one of the updates about being able to bake out a normal/displacement map...not sure if I read that wrong. If we were able to merge in a low poly model and use all the surfacing/voxel tools, and bake the results to a normal map on the original model (no need for retopologizing), that would be fantastic.

That would be awesome thus saving tons of work...

Share this post


Link to post
Share on other sites

I'm wondering if Surfacing tools/mode aren't being worked to provide a similar workflow and alternative to ZB's and MB's, as I thought I saw a comment on one of the updates about being able to bake out a normal/displacement map...not sure if I read that wrong. If we were able to merge in a low poly model and use all the surfacing/voxel tools, and bake the results to a normal map on the original model (no need for retopologizing), that would be fantastic.

I'm a bit confused by this since part of the reason I like and use 3dc as an alternative to zb or mud (both of which I own and have used in production) is because it allows me to not only work faster, but more fluidly by sculpting first in voxels, retopologizing over that, then uvs, then baking and painting - all within the same app. Retopologizing over a sculpt is much faster for me than buidling a poly mesh first and then sculpting detail over it. This becomes more true the more detailed and complex a project is. For simple things like head busts or the occasional low poly body it's not as noticeable, but for a detailed cinematic piece, it's much easier to retopo over a sculpt as a start and then finish off with polygon modeling. 3dc is the first app I've used where I could literally start a project from scratch and finish it within the same app. I find the need for editing polygon models by hand vert by vert is becoming more and more unnecessary in general. I love that 3dc allows for this. For me, the surface tools are only really necessary because they allow you to squeeze a bit more performance out of the app when detailing, and because they are easier to implement tools for that match zb in terms of brush control - as with the new chisel brush or pinch, for example, not to mention the freeze/masking capability. The biggest weakness in voxels for me has been the difficulty in achieving more extreme detail in sculpts. Hopefully the evolution of the surface tools will fix that.

Anyway, I actually wanted to discuss this recent tweet from Andrew:

"Every volume will be merged to the separate layer while merging from voxels to per-pixel. It gives much better control over baked color."

I don't understand what it means. Is it only something that will happen if you want to quadrangulate from voxels or does it apply to retopologized stuff as well? WIll it create separate mesh chunks and bake their unique color/shader properties to separate UV and image maps? Or does it mean that volumes with their own shader and color settings will simply get their own image layer when baked, but stay within a single mesh/uv set?

Share this post


Link to post
Share on other sites

Or does it mean that volumes with their own shader and color settings will simply get their own image layer when baked, but stay within a single mesh/uv set?

I think its gonna be that.

Just a wild guess . :)

Share this post


Link to post
Share on other sites

Hope so! =]

Yes it is!!! :)

But it does not work with all shaders,I get marble2 but not marble1 .

It works very well and the layer can be used as a mask to.

The only remaining problem is shader bump being prisoner of the main normal map.

Share this post


Link to post
Share on other sites

Looks like Ptex may not be a plugin after all?

I made save/load scenes with Ptex support and possibility to merge from voxels/retopo to paint room using Ptex.

Share this post


Link to post
Share on other sites

I made save/load scenes with Ptex support and possibility to merge from voxels/retopo to paint room using Ptex.

It opens possibilities for this workflow:

-Voxel design with med/hires details done with surface tools-

-Quadrangulation to Ptex with shader baking-

-texture work on Ptex model(color,spec and depth)-

-retopo and UVs using 3DCoat tools(1X lowpoly for game+ 1x higres mesh for cinematics(subdivided game lowpoly)-

-texture baking tool(normal map for lowpoly,Displacement for mesh for cinematics)-

You get 2 versions of same character ready for both movie and game. :brush:B)

(Exactly like Microverts but with no artifacts due to islolated uv islands generated by automapping,

and much better resolution and speed in texture work with depth, color and spec using Ptex.)

Of course,I have no idea what king of displacement is using Ptex.

But the screenshot Andrew posted looks very crisp.

And many approaches beside this one can be thought of. :)

Very exciting!

Share this post


Link to post
Share on other sites

looks like andrew added some PTEX speed improvements and expand and contract selection for PTEX texturing, VERY necessary for PTEX work IMO. Thanks andrew...

Looks like you beat nearly the entire industry in integrating PTEX into an app other than a render engine, congrats and thanks again!!!

hoping for multithreading for merging in voxels (so my screen doesn't freeze forever when using 4X & 8X res voxels) ;-) P.S. can anyone even work with an 8X res voxel object, wish I could :-( 20gigs of RAM 8x-2.8ghz

Share this post


Link to post
Share on other sites

P.S. can anyone even work with an 8X res voxel object, wish I could :-( 20gigs of RAM 8x-2.8ghz

4x, 8x, all depends on what you started with. If I start with a 40 triangle mesh my 1st gen iPhone would be able to handle 8x. :p

Share this post


Link to post
Share on other sites

4x, 8x, all depends on what you started with. If I start with a 40 triangle mesh my 1st gen iPhone would be able to handle 8x. :p

I get about 0.5 FPS on a primitive sphere increased to 8x

and and endless merge bar when attempting to take ANY of my voxel sculpts from 4x to 8x

the same results appear for me in OSX 3D coat and Win764 3Dcoat as well(different boot on same CPU)

quadcore xeon 2.8, 20gigRAM, Radeon 3870 512mb

Share this post


Link to post
Share on other sites

Well I made this test object and you can see it's currently at 34x, but it only takes up 5 million triangles. My avatar orc was around 18-20 million before I retopo'd and started painting. You can see my specs in my signature.

Share this post


Link to post
Share on other sites

Well I made this test object and you can see it's currently at 34x, but it only takes up 5 million triangles. My avatar orc was around 18-20 million before I retopo'd and started painting. You can see my specs in my signature.

wow!, something seems like it could be wrong with my 3D Coat install(win and mac) or my CPU, I cant push a sphere primitive above 8x, and you can do sculpting on a 34X voxel object, something seems very wrong here...

I also want to thank you big time for taking the time to reply and test this, your help is much appreciated :-)

Share this post


Link to post
Share on other sites

I was just trying to point out that the 8x number doesn't really mean anything. I scaled up the model and the 34x dropped to 4x. What really matters is the triangle count.

Share this post


Link to post
Share on other sites

I was just trying to point out that the 8x number doesn't really mean anything. I scaled up the model and the 34x dropped to 4x. What really matters is the triangle count.

thank you again for the help :-). Given your reply, may I ask what triangle count you can "max out at", meaning how high can you push it(on your system) before a merge bar will just sit there and process forever, 10mil, 15mil, 30million? I'm speaking specifically of the poly count in voxels mode(i know voxels arent polys). Any help would be greatly appreciated.

any other users that can chime in on this would also help :-) Thanks a ton

Share this post


Link to post
Share on other sites

I've only gone as high as about 25 million before it was just too sluggish to be usable and I moved on to retopo. I've never had it just hang like you say, I don't think.

Share this post


Link to post
Share on other sites

thank you again for the help :-). Given your reply, may I ask what triangle count you can "max out at", meaning how high can you push it(on your system) before a merge bar will just sit there and process forever, 10mil, 15mil, 30million? I'm speaking specifically of the poly count in voxels mode(i know voxels arent polys). Any help would be greatly appreciated.

I can comfortably work at a 20-25 mil safezone "per layers" range and up to approx 3-5 layers + a resonable assortments of 2mil sub-layers.

Navigation is slow.But brush speed is above 60fps.

50mil trees approx.

I have an old quad with Gtx280 and 6gig ram.Nothing out of space here.

(of course at 20 mil you use small brush,its not even a limitation it is just well...the way it works.

We get perpetual even topology and crazy 3d editing tools but we need to downsize brush half each time we increase res.It is a small price to pay.Anyway,we will see multires soon in surface mode Im sure.

Share this post


Link to post
Share on other sites

I've only gone as high as about 25 million before it was just too sluggish to be usable and I moved on to retopo. I've never had it just hang like you say, I don't think.

wow, even on my win7 64bit 3Dcoat I haven't been able to get it up to 25mil, or even past 20mil(i'm pretty sure. 3D coat usually starts to get VERY slow when its RAM allocation gets around 2.7-2.9gigs, on OSX and Win7-64bit. Again I have 20gig of 800mhzDDR2 RAM, and an ssd that I have much of my win7 files read and written to. Very odd that you can push it that high, when I cant even merge in an OBJ above 20mil without the merge progress bar stalling(for over 12 hours) after about 2.7gigs of RAM is allocated to 3D Coat, and I've been trying this since V 3.1 on both OSX and Win7-64. Most 64-bit 3d/MoGraph apps easily blast past the 2.7gig RAM line well into the 8+gig territory, for me.

Odd, others(Artman as well) can push it above the 20mil line. i'm holding out for a fermi GTX480 card soon, get rid of this ATI card, although it still does a great job in most 3D apps, its a RAdeon 3870 with 512 of ram and.

Thanks for the reply Artman & philnolan3d. I'm enjoying 3D coat far more than the other sculpting apps I use, love the retopo too, however my need to merge in OBJ's at over 20 million seems to be holding things back for me :-( My OBJ's arent 20mil of course, but, to bring them in with enough res in the vox tree, I have to push it well over the 10 mil mark before I start getting the med/high level details in the exported OBJ.

Has anyone tried to merge in an OBJ with the polycount in voxel mode being over 20million, or just created objects within Voxel mode that end up being above 20mil? MAybe it just an import/merge issue with pushing the polycount that high.

I too heard the rumor of multires in vox surface mode, that would be SWEEEETT.

wish merge was multithreaded. Wonder what the memory paging in 3D Coat is like.

Share this post


Link to post
Share on other sites

wow, even on my win7 64bit 3Dcoat I haven't been able to get it up to 25mil, or even past 20mil(i'm pretty sure. 3D coat usually starts to get VERY slow when its RAM allocation gets around 2.7-2.9gigs, on OSX and Win7-64bit. Again I have 20gig of 800mhzDDR2 RAM, and an ssd that I have much of my win7 files read and written to. Very odd that you can push it that high, when I cant even merge in an OBJ above 20mil without the merger bar stalling after about 2.7gigs of RAM is allocated to 3D Coat, and I've been trying this since V 3.1 on both OSX and Win7-64. Most 64-bit 3d/MoGraph apps easily blast past the 2.7gig RAM line well into the 8+gig territory, for me.

Odd, others can push it above the 20mil line, i'm holding out for a fermi GTX480 card soon, get rid of this ATI card, although it still does a great job in most 3D apps with its a RAdeon 3870 with 512 of ram and.

20 mil is a chunk...I mean its resonable to wait 10 min for it.

Try increasing virtual memory instead maybe :huh: Its just a stupid suggestion.

I have it at 20gig.

I tried it and if I remember it wasnt that long...4-5 minutes maybe

But anyway,what I mean is maybe your max is like 19mil or something :)

Its really a benchmark value,i mean the last STage..hte one in which you scult only in surfacmode.

Any realistic object can be recreated with a 10mil tree(cutting operations ect..).

any organic madness can be created withn a 10mil with brushes.

10mil head.8mil torso.2mil strap.4mil gun.8mil booths.1 mil boothsornements...ect...

Share this post


Link to post
Share on other sites

I'm curious why you need to import at such a high resolution. If I'm going to import an OBJ (or LWO in my case) I usually import at less than 1 mil and gradually increase as I need to from there.

BTW this thread is getting a bit off topic, really should be continued elsewhere if it's not about Andrew on Twitter.

Share this post


Link to post
Share on other sites

I'm curious why you need to import at such a high resolution. If I'm going to import an OBJ (or LWO in my case) I usually import at less than 1 mil and gradually increase as I need to from there.

On my one of my current projects I'm actually using 3D Coat to import a nearly finished 3D mesh of a figure so the voxels can automatically watertight her for 3d printing, and of course do some extra sculpting cleanup in 3D Coat before exporting ;-)

The figure is complete with eyelids, fingernails, all the usual female figure details... and I cant seem to get 3D coat to import/merge her in above 20 mil, thus, details like the eyelids come in jaggy, and lacking the res I need when I go in to smooth out the jaggies. If I could just get her in there with a bit more voxel res, she might import and look closer to her OBJ exported appearance, and I would then have enough vox res to properly smooth out her eyelids.

Again, I GREATLY appreciate your help, as the whole increase/decrease density/size VS increase resolution in 3D Coat isnt as intuitive as I would like, and the manual IMO lack thorough explanation in this area(IMO).

Sorry for getting off topic, i've created threads, but not many respond(akaik) to many of the threads on the forum with this particular question, figured I'd take advantage of your wisdom while I could. :-)

Share this post


Link to post
Share on other sites
Again, I GREATLY appreciate your help, as the whole increase/decrease density/size VS increase resolution in 3D Coat isnt as intuitive as I would like, and the manual IMO lack thorough explanation in this area(IMO).

I agree, i have had problems with slowdown in the past when increasing i think this type of stuff might be a good idea for the video manual series maybe so you have a simple mesh and it shows how to go up in quality without slowdown on most computers. It seems mostly for the voxel tools right now but commonly asked subjects like this would be great also for that if they like this idea.

Share this post


Link to post
Share on other sites

×