blender's multires really needs someone to look at it... seriously

General Forums Blender and CG Discussions
1

well, multires, we all know it is some awesome tool for sculpting, but it has also a lot of annoying “features”… the latest i just suffered is this:
i have a model in multires, which is on level 7 and weighs in 5 million quads … i have rendered 25 million polygon models (without multires) well on my mac in blender, in just 6 minutes with use of 9 gig of ram, all worked well, now, i just wanted render the model with the multires and “only 5 million” quads , i hit render, and what happens? … nothing, hammering but nothing else, blender builds a swap disk (mind you my mac has 32 gig ram) of 84 gig for rendering this model??? (this is what activity monitor told me…) it took almost 30 minutes for my mac to become responsive and recover from the mess blenders multires created on my harddrive, and of course not a single pixel was rendered after all this time waiting and listening the hammering on the harddrive … (since i broke off the “rendering” as soon my mac was again responsive)

now i ask, how does blender uses 84+32 gig of ram for rendering a multires models of 5 mill quads???

so, please, someone look into this, seriously!! … and no, i cannot send the blend file, its almost a gig big… i am sure it is a quite normal file, just the (portrait) model and a basic material nothing fancy, i am (in sculptings area) an experienced blender user …

i really am annoyed, multires breaks the workflow, we cannot access the vertices from higher levels which is vital in so many things, for example for vertex paint, or weigth paint, we get spikes when we are not ultra careful , we cannot recalulate subdivisions properly,
and, now i learn that a model that the 32-bit-application zbrush handles with grace (even on my airbook), is a complete desaster in blenders multires on a capable machine, i cannot sculpt on the model anymore in blender - i can well do sculpt on the exact same model in zbrush, a 32-bit application,i can rotate the model as quickly as i want in zbrush and use brush sizes on it that almost fill the screen, no problem, all is smooth even if i do multiple undo’s in a row, and despite my abundance of ram i cannot do the same in blender, rotation of such a multires is a pain, brushes work with hopping even though the brush size is comparedly small, undo’s better forget them on such a model, and now blender even fails to render that damned thing!! … how are we supposed to work with this??? for high detail models that should be rendered large, and not like a postage stamp, we need models of this size in multires…

sorry if i sound angry, but so much wasted time and effort for… nothing… makes me angry… now i need waste more time, and figure out a work around, since i WANT render in cycles…

maybe not the right forum, but i think this topic needs another new discussion to make aware of the problems multires has, since something has to happen…

2

Yes, i can UNDERSTAND your frustation but here i think the developers don’t hear you . they talk " i want blender take seriously " but what happen they develop one feature then it’s ok no more until the next year .

Look at the Sculpt in blender now we have dynotopo and what happen nobody develop this year ( and what happen what they talk to take it seriously? ) nothing it’s a lie .

if they want Blender take seriously in the industry the SHOULD work in those areas Blender has a great Sculpt but need more work

We needs

-layers
-transpose
-etc etc.
But what happen with the developer now is working in other place

3

As far as I have read, the last time Nicholas Bishop has been contacted was late last year, perhaps it would be worth a shot to fire off another PM to see if he’ll be able to find some time to fix some of the multires issues.

Is the issue with rendering recent by any chance? If so, then it could easily be the result of the sculpt mode becoming an orphaned area that no one is interested in developing. In this case, it would be even more urgent to find a developer to start working on it because otherwise things might be better off to just remove the multires modifier entirely and instead focus on things like retopology for the Dyntopo system, which is interesting because it seemed like the sculpt mode had a ton of activity for a couple of years at one point. Perhaps if it was removed it will return sooner or later in the 2.7x series in a much stronger form.

It’s like how the game engine was in danger of wholesale removal before Ben2610 showed up and fixed all of the major bugs, there is a chance that Lukas Tonne’s arrival to becoming a paid developer could free up time for others to look at issues such as these, but it’s like as Ton was reported to have said before, needing more developers and more financial resources to ensure a more even development coverage for all areas.

Also, I’m not sure what the effects will be, but there is talk of moving the Blender development scene over from using SVN to GIT, this might encourage more development as GIT is superior in some areas like merging with trunk.

4

Well, nobody is going to fix this for you any time soon (and even then it’s hard if the problem cannot be reproduced), so I suggest you try working around it. Have you tried applying the modifier?

Yes, i can UNDERSTAND your frustation but here i think the developers don’t hear you . they talk " i want blender take seriously " but what happen they develop one feature then it’s ok no more until the next year .

You seem to believe that there is one group of developers that all work on everything and “want blender taken seriously”. That’s not the case. The sculpt code is from volunteers, not the (few) core developers. Volunteers sometimes move on with their lives. They’re not required to support their imperfect code indefinitely.

5

@Zalamander I may be wrong but I think Breacht not Nichloas is the one that did the data structure that mulitires currently uses around the time of Sintel. So one of the core developers was involved in that area of development. This happened at the beginning of Sintel, I’m sure google can Still find that blog post.

@Ace Dragon you must really not sculpt because otherwise you would have picked up that Dynatopo has an even lower and more restrictive polycount than multires its good for concept sketching but if you really want to add a ton of detail you have to retopo your model and use multires. 500K-700K triangles and dynatopo grands to a halt for most people. So removing multires and relying on dynatopo is a bad idea we need both working fluidly.

EDIT:
Sintel post detailing sculpt developments at the time. I confused my facts Nichloas implemented the data structure but Brecht also worked on it. The question is does he have time to work on it on top of Cycles and bug hunting.

http://www.sintel.org/news/sculpting-development/

6

But if Multires rots to the point where it becomes completely broken, everyone may have to swallow the bitter idea that the idea of professional sculpting in Blender is something that was never meant to be. Of course this will mean that Blender will get the embarrassing distinction as the only 3D app. that removed a major feature due to lack of maintenance, but the maintenance thing is something that has historically plagued the entire concept of FOSS and can only be mitigated by strong leadership and a lot of financial resources that allows for paid developer positions. Just think of where Blender could’ve ended up if it was just a organically flowing project driven strictly by volunteer effort. (hint, a lot of the professionals on this forum may not even be here).

Also then, you will no longer see threads like this because they will move from complaining about a feature to a feature request, I really don’t like the idea of throwing away all of the hours of work done by Nicholas Bishop, but broken features that aren’t maintained means there might be some hard choices to be made down the line.

7

@Ace its not broken enough to warrant removing it, I also think it is a super idea that needs more development. From cartoons to realistic work people are sculpting first and than building retopology on top. BA is one of the few places where I see people avoiding sculpting so much even when doing things like a head model people are going poly by poly. I often think if you use dyntopo or mulitires can you cut your modeling time down so much. multires works its just temperamental and likes to throw spanners down your way.

I think an open movie with a more modern character and creature modeling workflow that has sculpting at its core will get things fixed.

8

In what I’ve been seeing, looking at it from this case would make it seem like there’s an uphill battle in terms of encouraging more sculpt work (and ultimately more dev. interest), in part because of a combination of the vast upgrade of the modeling tools and the stagnation of core structural development for sculpting.

Now this might be true in part that more people would take to sculpting if it was developed more (multires getting full vertex color support and the like) along with automatic retopology tools, but it currently seems to be an easier thing to do as of now to just introduce new tools for box and poly-by-poly modeling.

As for me, I generally have a preference for just using the modeling tools because it caters to a more analytical way of thinking and allows for careful tweaks to make everything perfect, I’m actually not all that stellar with a brush compared to Doris and Michalis.

9

How is it “completely broken”? Do so many people really have problems with it? It might just be an isolated case, and then there isn’t really anything to do about it, unless it can be reproduced. If you want something fixed, you have to at least file a reproducible bug, not complain about it on the forums.

Maybe people have to swallow the “bitter reality” that sometimes software fails and nobody will take responsibility. It happens in commercial software, as well, but at least you have a right to complain about it there. So, if it works better in ZBrush (which implements everything differently), use Zbrush in the future.

10

Not sure if i understand correctly but if the problem comes when rendering with Cycles,i’t not a Multires problem but more a Cycles problem.
I also remember trying to render huge multires meshes with cycles without luck.
As far as I know Brecht did a lot of work for multires when Sintel was done,so in the future I think he can support it in Cycles.

11

Well, I think this highlights an opportunity for Blender to (once again) differentiate itself from the poor reputation other FOSS graphics applications have. They can ignore the reports & comments about a badly performing area of functionality or they can listen to the users that are trying to make the application work in the real world.

It’s hard to argue that Michalis and Doris are not amongst the more talented/serious users of Blender’s sculpting tools. Just take a look at their sketchbooks and their comments/contributions to threads discussing the feature. Ignoring them basically writes the feature off as a loss. They’re not here to make trouble, they’re not trolling for outrage, they’re passionate about sculpting in Blender and, as such, passionate about the bugs that make it unusable for serious work in Blender.

I hope the Blender developers don’t treat them as David Revoy was by another FOSS project. When your brightest & most passionate evangelists decide to give up on your project - it’s a very bad sign for the future. Like it or not, studios (small indie through to large commercial) pay attention to the talented artists supporting/promoting a software package. When they stop supporting/promoting a package - so too do their followers. It’s human nature.

12

@doris: perhaps the problem is due to the undo stack: when you change the level in multires, blender may store the precedent level to be able to undo. You may try to reduce the number of undo available.

13

it is good that the discussion restarted.

i will add/reply to the comments in no particular order.
1)
of course, yes i can apply the multires modifier and then render, but the project is not finished, and if i apply now, i cannot sculpt seriously on it anymore. now the performance is bad, but i still can do something. with multires applied i can forget it…if i apply the modifier now, i cannot edit the uv’s any longer, who wants manage uv’s on a 5 million mesh, no applying the multires does break the workflow even more, it is not the correct solution (even though i often do it for final rendering when i need paint the vertices…but it is wrong from the workflow point of view)… i wanted do a a complete preview render as i often do to see how all looks, right in the middle of the project, applying multires now is wrong when blender cannot recalculate the subdivisions …
2)
my undo level is only 2, i had reduced that already since i rarely do undos and in blender they give trouble anyway with larger meshes and dynatopo
3)
i dont know if the renderproblem occurs in blender internal, i have never used blender internal for rendering, i do not know how to use it, and my interest was always cycles, that was what has drawn me to blender in the first point…
4)
it was said we should produce a file that reproduces the error… i did so with the spikes error, it was a very easy mesh with just 2 faces more than the cube, that produces the spike, but then that discussion swapped to the observation that the mesh was broken and it was my fault that multires makes spikes,however i had tried to show how the spikes occur everytime, zbrush btw did not had issues with this mesh, zbrush ignores users errors… now, how can i give you a 1 gig file to show the error, as this rendering issue only occurs on big multires… there is nothing special in this file, its a portrait bust as i have done many
5)
of course i could do everything in zbrush, but(!) blenders dynatopo is far superior as a sculpting tool than zbrush dynamesh. blenders sculptbrush “claytubes” is far superior to zbrush equivalent, and this is the brush that does 95% of my sculpting work,when you only box model, or polymodel, you cannot see the huge difference, but when you do sculpt, you will recognize the far superior concept and behaviour of dynatopo. and, i like to use the best tool for the hardest part of sculpting which is the concept and blocking out stage…
6)
if multires gets removed, i will probably use blender only for rendering. to me the main assets of blender is cycles and dynatopo. i love both very much. i have learned to use the fetures i need, and am generally happy with everything,but if multires gets deleted completely, then i could not even work with the workarounds i am using right now. blender would become a tool which i could use only for sketching/concepting and then rendering… the rest i would have to do in zbrush…well, of course i can do that, but blender has some other points i favor over zbrush, one of them is quite generally that blender does let me adjust everything via precise numbers, in zbrush many things you have to eyeball, its hard in zbrush to pose a tool at an exact coordinate, since this is not what zbrush supports easily…

blender and zbrush make a very good combo, blender could stand alone, if multires gets now the attention it needs. in the one year i am using blender now, i have done many projects completely in blender, for what i do blender has a good toolset, however the workflow issues need be resolved now … and, i do not mean necessarily new features for multires, vertex paint in multires is not a new feature i would think, as all it needs is allow the user to access the vertex data outside of sculpting mode…but i wish multires gets cleaned up and stabilzed. that it works like it should in a production with no issues generally… and, of course, there is a huge list of feature i would love to see multires had, but as said, my main concern is that i now wish it become a reliable tool just with the things it can now, no more…

14

I dont necessarily agree with your assessment of what zbrush isnt doing as well as blender, but I do agree with the conclusion that Blender could (and should) be able to stand alone as a great sculpting application…and to do that the development has to continue and improve. Unfortunately I dont think its even on anyones radar as far as development is concerned… There are a lot of 3d modeling applications out there, but not a lot of sculpting ones. This is a good area for Blender to shine in.

15

You have to understand that Blender internal and Cycles are 2 different renders,something it’s supported in Cycles,something not(Blender internal supports everything because features were designed to work with it,differently from Cycles),before everything is supported we have to wait years(I hope much less).
And,I understand you are passionate about sculpting in Blender and you have an idea about how it should work,but if you use Blender since 1 year,I think you haven’t understood yet how open source development works,you 'll have to wait a long time before something get fixed(for example I’m still waiting since 5 years some fixes in the particle system for fur),if you want to use Blender you have to be prepared to have a lot of frustration.

16

renderdemon, yes. i understand that cycles cannot render everything blender internal can… but this bust in question is simply the multires mesh with a basic material on it i asked cycles to render, same as i did in many other projects. i know cycles can render this, in principle. it is a multires issue here, that created the huge data out of “nothing”… and yes i do have very clear ideas how the sculpting area of blender should integrate into blender to allow for the best experience on a project involving sculpting…

yes, saint heaven, sculpting is a very good area blender could shine in, i agree so much! this is what i wish to support with my work i show here, showing what blender already can, dynatopo is in its baby shoes, but a genius thing. but dynatopo needs multires afterwards, otherwise it would not make that much sense to develop dynatopo further,

17

very good topic!

18

Indeed!!

I’ve been using multires lately after more than a year and have run into a plethora of problems. But, hearing crazy talk about multires just being dropped makes me smirk, at least so I don’t get angry.

The main problem is we don’t have a developer able to spend time on multires, right?

19

Yes, that’s basically the problem, sculpt mode could certainly use a dedicaced developer for it.
I hope the donations will increase with that Valve - Blender thing ,so in the end the Blender Foundation will be able to hire one. It would help too if the next open movie require sculpting, as it would expose first hand what are the problems of multires and dyntopo to the devs participating in the movie.

20

I contacted nic today, just asking his opinion on the future of multires and dyntopo. Lets see what he has to say, if he has time to respond.