Performance Improvement for Voxels

[DRAWINGTANK]

Hi Andrew,

I am very Frustrated ....

The Problem that I am having is that voxels are only good (with my sys.) for basic stuff, very simple models and I get the out of memory & crashes

( freezes up ) very often if I try anything more complex

I know I requested 3DC 64bit for Macs before but what I am asking now is... there anything you could do performance wise or optimization in the code

to make life for guys like me with a " Normal/ Standard " computer working with voxels more accessible & fun ?

For what I want is to create in voxel and " STAY" In Voxel Mode for the " Final Detailing "

As an example in ZBrush I can sculpt a mesh with 15 million polys with 34 subtools and be abel to rotate, zoom in and out and sculpt with no real

some " lag " but not the complete freeze up as I experience in 3dc so often

3DC for me is not very stable and it's very frustrating to know the things I could create with voxels only if.... !

If you could make 3DC do more with less like ZBRUSH ( no CUDA needed) instead of requiring expensive video cards as a must to get good

results ... I am not saying that its bad to have a great video card and lots of ram but its not the norm, this would make voxels more appealing to a

broader audience, specially those of us who only have laptops

PS: It is interesting to note that ZBrush is only 32 bit for the moment but yet manages to be 1000x more efficient than 3DC , meaning with no special hardware needed you can achieve insane levels of detail... this in my opinion what gives ZBrush the edge over 3DC

I really hope some BIG improvements in Voxels are coming very soon

thank you for your time

Model Name: MacBook Pro

Model Identifier: MacBookPro3,1

Processor Name: Intel Core 2 Duo

Processor Speed: 2.4 GHz

Number Of Processors: 1

Total Number Of Cores: 2

L2 Cache: 4 MB

Memory: 4 GB

Bus Speed: 800 MHz

[haikalle]

These are very good requests and I hope that Andrew has time to take a look at this. But don't expect this feature to come any time soon.

This is a huge task even for a big company. Making voxel to work with voxel surface mode is the smart move right now. But things can change

very quickly, so let's keep our fingers crossed.

[Calabi]

Its not really a request thought its just an obvious observation.

I would hope for 3D Coat goes down the Sculptris route. Instead of the impossible Zbrush brute force approach(voxels cannot behave like pixols). Look for innovative ways around the performance deficiences.

[Taros]

Its not really a request thought its just an obvious observation.

I would hope for 3D Coat goes down the Sculptris route. Instead of the impossible Zbrush brute force approach(voxels cannot behave like pixols). Look for innovative ways around the performance deficiences.

I don't think sculptris is a good example. Sculptris is small, fast and nice because it has not much functions at the moment. It is in a early developement stage. If the programmer will do more, and the users will cry for more for shure, we will see what will happen with sculptris in two or three years.

[Calabi]

I dont mean for 3DCoat to emulate Sculptris literally. I'm talking about its design ethos, and the direction it went relative to everyone else. All others were screaming "Polygons we need more polygons!". Sculptris proved them wrong it does the same job with way less resources , and it has a more fluid and flexible workflow. It took a problem and solved it contrary to what everyone else was saying, the others didnt even know there was a problem till this guy came along.

[splodge]

ZBrush is a subdivision based modeler and this means it can temporarily drop to lower subdivision levels when necessary, this is what helps to keep performance good. But with ZBrush's subdivision surfaces you don't get the freedom of voxels.

Ultimately it's probably asking too much to want 3D Coat's performance to be the same as ZBrush's. Just as it would be asking too much to want the freedom of voxels when working with ZBrush's subdivision surfaces.

A sprinter can run faster than a long distance runner, yet the long distance runner can run further than the sprinter. So who is better at running?

Sculptris doesn't bother running, it gets the bus instead.

[polyxo]

ZBrush is a subdivision based modeler and this means it can temporarily drop to lower subdivision levels when necessary,

this is what helps to keep performance good. But with ZBrush's subdivision surfaces you don't get the freedom of voxels.

.

This is not entirely correct.

Zbrush is actually Subdivision Modeler with a Voxel Engine under the Hood.

One can simply append a Zsphere to any input-model and start changing the existing Shape very dramatically.

To make the existing and the new parts one combined mesh however a (very fast) quad-remeshing has to get performed first.

[Digital777]

I get what people are saying about sculptris, i can get quite good resolutions from that with less slowdown than i would in 3DC but i think that is mainly down to the way it works. Basically it applys detail when and where it is needed and you have lower resolution anywhere else that doesn't need it rather than just increasing everything together when you go up in detail. It also has a good reduction brush to lower things also. In terms of high res brush/surface details though sculptris is not that good in my opinion but for basic sculpting stuff it is great.

I think those types of ideas to just use what it actually needs and also tools to help optimize things would be a good idea for 3dc also. I think andrew is doing a great job so far and i know that voxels is a different system to other programs although they get similar results but hopefully this type of stuff can be done sometime also.

[AbnRanger]

Well, I certainly feel much of your frustration...much of the time. I think 3DC really needs to be more efficient with resources as well as fully utilize recent hardware advances. The lack of Multi-threading for the slowest of all functions (merge calculations) certainly remains an impediment. However, I don't think it's realistic to expect 3DC to go through such a dramatic paradigm shift anytime soon. Heck, Mudbox out-muscles ZBrush in raw sculpting speed, but you need a decent, newer system to take advantage of it...and I like a program that scales, in performance, along with your hardware. ZBrush only being 32bit is quite puzzling to me...even if it does fairly well with old technology.

Having said all of that, you do have the ability to Cache your layers, so that you can better manage your work within your system resources...and Caching lead to Multi-Res, which works pretty well. Even in ZBrush, you want to start low-res, and block out your model. Your laptop can handle that.

When you are ready to move up to higher resolutions, you can use the "Split" tool to separate sections of your model and cache the layers/parts you aren't currently working on...that allows you to do higher res work one section at a time. With a fair amount of planning, you can get considerably higher levels of detail, now.

[splodge]

.

This is not entirely correct.

Zbrush is actually Subdivision Modeler with a Voxel Engine under the Hood.

One can simply append a Zsphere to any input-model and start changing the existing Shape very dramatically.

To make the existing and the new parts one combined mesh however a (very fast) quad-remeshing has to get performed first.

ZBrush is actually a subdivision modeler with a marching cubes algorithm under the hood. But I suppose the "cubes" generated by the marching cubes algorithm could be seen as being voxels. But at no stage can you edit those voxels.

and yes, ZBrush enables the user to regenerate meshes (unified skin), but it's hardly as free form as voxels. But lets agree to disagree on that.

[DRAWINGTANK]

"ZBrush only being 32bit is quite puzzling to me...even if it does fairly well with old technology." -AbnRanger

I think ZBrush's ability to work extremely well on " regular sys. including laptops is I think one one of the biggest reasons why its the industry standard

it's great for a software to "scale performance" with better hardware but it's a mistake to have to " depend " on the latest, fastest sys. in order to have a

good experience with the app.

even-though Andrew is making " some" headway with proxys, in my opinion I wish he would just focus on voxels and get it to a level

that everyone, independent of how good your hardware is be abel to use voxels as I use polys in ZBrush

As a ZBrush user I can tell you that the new ZSketching is as good as voxel... in it's on flavor ( it's a matter of taste), you can create things out of thin air

and do a few things that voxels can't do at the moment such as " Sphere Depth control " have the sphere " follow a reference mesh ", scale, rotate &

move individual ZSpheres, preview as a poly mesh among many others.....

"I would hope for 3D Coat goes down the Sculptris route. Instead of the impossible Zbrush brute force approach(voxels cannot behave like pixols). Look for innovative ways around the performance deficiencies."- Calabi

I believe 3DC is going down the same path as Mudbox, the "brute force" approach

ZBrush " is 100% " innovation" remember they are doing things that no one-else is remotely doing all with a good old 32bit app.

I do agree that sculptris is a great example of how quick you can create something fresh.... Tomas Pettersson did it in six months as a hobby !

so yeah... we should expect a lot from the great Andrew

PS: I think some of you guys are underestimating Andrew by saying it's not " realistic" to expect this or that... remember how far 3DC has come in such

a short time ! it's just our job to remind Andrew whats important to us, the users

"If you think you can do a thing or think you can't do a thing, you're right." - Henry Ford

http://www.sculptris.com/story.html

Story

Hi.

I'm Tomas Pettersson, and I created this application for my own enjoyment over the past six months (starting at the end of November 2009). This is an unpaid hobby project, but it has essentially consumed all my waking time during this period.

I don't want to demand any payment for the application, but of course I will appreciate a donation if you feel like it is valuable to you. Take a moment to consider what it would be worth in your particular case.

A pizza? A movie ticket? A day's salary?

[Taros]

Great things needs time... as always. We all will get our features, one by other - sometime.

I have the time to wait and will support this project with my ideas and suggestions. And this is the most important: To say what you wish.

[DRAWINGTANK]

I found this discussion in the ZBrush forum regarding 3DCOAT, I think Andrew has a great opportunity to grab lots of costumers if he plays his cards right

the main problem with 3DC was performance issue............ now what a coincidence

http://www.zbrushcentral.com/showthread.php?p=699063&posted=1#post699063

below is a sampel

Quote:

Anyone who has been waiting endlessly should give 3DCoat a try due to the fact that Voxels (blows away poly sculpting), better topology tools, better UVs so on, not to mention (most important) that Andrew the developer of 3DC actually " communicates" with the users, how about that ! and is constantly fixing bugs & adding features on weekly builds that you can download for free

So if you are frustrated with how Pixologic is treating us ( lack of respect) I strongly encourage you guys to give 3DC a try

Perhaps if enough of us migrate (mac & win), Pixologic will realize that they are not the only game in town and change the way they do business

check it out:

http://www.3d-coat.com/

3DC is ok, but I have yet to see any convincing work done with it. The best 3DC work is nowhere near the best ZB work that I have seen.

Voxels do not (yet) "blow polys away", they are just different, at times better, at times inferior.

3DC does seem to be building itself as a better more rubust production tool, while ZB seems to add wacky, random features that artists later have to coerse into compliance with production.

ZB is still miraculously the better product. If 3D coat cleans up it's UI, adresses some performance issues, and gets some talent producing art to attract users, well... it would be a whole new ballgame.-Obz

[Taros]

I hope that the new brush engine, that is planned by Andrew, will change a lot of things. Especially the surface mode contains very high potential. The brush engine in zBrush is really something that is not compareable with 3DC currently, it's nearly perfect. The same with Mudbox, there the brushes feels good too.

Let's see what happen.

And because the detailed Models: Everybody know, that a tool grows with the community. So in the future very good examples will come by the users. I guess a lot of professionals are not able to show their work, until now. My projects are under an NDA, for example. As soon as I am able to do it, I will show my results. All of my current work was done in 3DC and will interest some people for shure. Maybe the results will change some peoples view outside there.

Best wishes

Chris

[AbnRanger]

I was frustrated too with the performance issues...mainly the long waiting time merging back from Surface Mode to Voxel Volume. That's why I still try to avoid using Surface Mode, even if it is relatively faster than Volume mode...the con outweighs the pros, in my experience. However, with the new Caching and Multi-Res features, it's a HUGE step forward. If you section your model up and cache what you aren't actively working on, you'd be amazed at what you can now do. It sounds like you haven't even tried that workflow yet....

Voxels introduce a lot more freedom in sculpting, but with that comes a bit of cost. Right now, it's that it requires a decent amount of RAM resources. If you could punch holes in models in ZBrush as easily and care free as in Voxels, it would hog RAM resources too...don't forget that. This is why I hope, once a lot of the bugs are squashed, that Andrew next replaces the current mesh editing engine and toolset in the Sculpt Room with the one in Surface Mode. Think of the implications that would bring...

[Greg]

If you section your model up and cache what you aren't actively working on, you'd be amazed at what you can now do. It sounds like you haven't even tried that workflow yet....

Don, I haven't tried that yet either. If you find a few spare minutes could you do a video showing that workflow, even if it's just on a simple model or something?

Greg

[DRAWINGTANK]

I was frustrated too with the performance issues...mainly the long waiting time merging back from Surface Mode to Voxel Volume. That's why I still try to avoid using Surface Mode, even if it is relatively faster than Volume mode...the con outweighs the pros, in my experience. However, with the new Caching and Multi-Res features, it's a HUGE step forward. If you section your model up and cache what you aren't actively working on, you'd be amazed at what you can now do. It sounds like you haven't even tried that workflow yet....

Voxels introduce a lot more freedom in sculpting, but with that comes a bit of cost. Right now, it's that it requires a decent amount of RAM resources. If you could punch holes in models in ZBrush as easily and care free as in Voxels, it would hog RAM resources too...don't forget that. This is why I hope, once a lot of the bugs are squashed, that Andrew next replaces the current mesh editing engine and toolset in the Sculpt Room with the one in Surface Mode. Think of the implications that would bring...

I have been using the proxy workflow and it has improved things for sure

However my frustration comes from been unable to first of all stay in voxel mode till the end and second the instability of 3DC with anything remotely more complex ( out of memory errors even though I got 4gb of ram )

Thats why I suggested on my other post " Local Resolution in Voxel mode " http://www.3d-coat.com/forum/index.php?showtopic=5921

to have the ability to "Inc. Res. only where its needed, maybe a combination of Proxy and Local Resolution will get us to the ZBrush Performance level

The " do more with less resources " approach will benefit even the users with the " monster sys. " btw so it's a great thing that we are discussing this in this forum.. making 3DC a better app I have no doubts that 3DC will surpass ZBrush at

some point, hopefully soon :

ps: It would be awesome if we had something like Increase Resolution & Decrease Resolution Brush... I think it has been mentioned before on the forum

[AbnRanger]

I have been using the proxy workflow and it has improved things for sure

However my frustration comes from been unable to first of all stay in voxel mode till the end and second the instability of 3DC with anything remotely more complex ( out of memory errors even though I got 4gb of ram )

Thats why I suggested on my other post " Local Resolution in Voxel mode " http://www.3d-coat.com/forum/index.php?showtopic=5921

to have the ability to "Inc. Res. only where its needed, maybe a combination of Proxy and Local Resolution will get us to the ZBrush Performance level

The " do more with less resources " approach will benefit even the users with the " monster sys. " btw so it's a great thing that we are discussing this in this forum.. making 3DC a better app I have no doubts that 3DC will surpass ZBrush at

some point, hopefully soon :

ps: It would be awesome if we had something like Increase Resolution & Decrease Resolution Brush... I think it has been mentioned before on the forum

Yeah...I sympathize with your plight. I have a new laptop (i7 w/ dedicated ATI 5730 1GB) but at the moment, it only has 4GB as well. Some files that I created on my desktop, I can barely open on the laptop. But as long as I manage to keep the RAM amount down under 4GB, it's pretty fast.

Now, Leigh Bamforth mentioned some tips (a while back) that essentially give you what you're asking. That is to use the "Copy" tool to paint areas where you want High Resolution, and leave the rest of the mesh lower res...you can merge the (High-Res) copied layer with the lower res, when you're done or ready to.

I'll try to make a quick video demonstrating some of these things in a little bit, so others can get a better idea too.

[Ratchet]

Just some points :

For Zbrush, sorry im' not against them at all, it's a good innovative company that have launched a new way of modeling

Each NEW VERISON version is FREE, you don't have to pay for it , there are not lot of companies doing that !

And sorry but the 3.5 and version 4 have brought a bunch of big new things like ZSpehere 2 (a little like Voxels) and Shadow Box !

And you don't pay for that !

Well, yes we all would like 3D Coat to be as fast as Zbrush, but the problem is pure Voxels use, CPU and memory use are intensive once use increase voxel resolution.

The best solution i see for now is purely based on sculptris :

Possibility to have detailled zones and non detailled zones on a same mesh : like Sculptris some parts would have lot of polygons and less on other parts of the same mesh. And a tool brush like in sculptris to reduce polygons on the mesh anywhere we want.

I think this could help A LOT in performance, caus why having lot of triangles on basic simple surfaces ?

[AbnRanger]

I think it's just a matter of time before Pixologic has to cave in and start charging upgrade fees. That is not a sustainable model, even though it helps generate plenty of buzz among the ZBrush community (as most of the crowd already has a seat). They have to rely on new sales only, and that is going to be to their eventual hurt, with all the competition coming on.

It's actually helpful to 3D Coat for them to sustain that policy, as Pixologic generates less income and thus fewer development resources...allowing 3DC and Mudbox to catch up.

Again, use Voxel Volume mode for what it is good for (cutting/Boolen type operations, blocking out a model, and getting to medium res before needing to switch to Surface mode), and use Surface Mode and Multi-Res for what they do well...even faster sculpting. Surface mode is very comparable to MB and ZB in raw sculpting speed. You just need a good chunk of RAM to take advantage of the benefits you find in Voxels. It's having to store more data, so that is the tradeoff...which each software in this category has at least one.