Call for Blender artists for Blender to Renderman test!

#1

Note to the mods - not putting this in the WIP/Collab threads because there is no clear idea what is to be made.

Anywho…

Us over at the Blender to RenderMan project are kind of stuck in situation where we want to showcase the ability to PROVE that the tools that have been developed to export animation from Blender into Renderman can work, yet at the same time the software developers are saying it’s not ready yet. This actually got started on the BlenderArtists forum and on on one hand I can agree that taking on a big project would not be a good idea, on the other hand I say we pump out something small and quick that can prove what we all have been working at can actually work. I do agree on some points, think big and it falls apart. I also think if we build up towards something big then it would be easier to work with - so starting small first is the best option at this time. So far over the past 2 years only one animation was made and that was by the developer of Mosaic - a very good camera fly through btw - what is needed is something with a bit more emotion in it.

I keep thinking about the Pixar shorts, where in the earlier days they were meant to show the technology but with that artistic touch. A lot of the earlier stuff in comparison looks primitive compared to the later shorts, yet even Wally Bee and Luxo are still amazing pieces of animtion. This is something that I believe we can do, I simply cannot do it alone though because I am horrible at character modeling and animation. Which is why I am coming here for help as opposed to the BlenderArtist forums - I think there might be a better chance of obtaining help from a couple of people here, the BA forums tend to be a bit Blender specific biased (just my opinion folks, not fact).

  1. Artistic benefits. If we can’t show that the result
    couldn’t be done as well or better in Blender without the aid of
    RenderMan
    the effort will do more damage than good. It would need to focus
    squarely
    on the strengths of RenderMan (motion blur, depth of field, high level
    primitives,
    displacement, shading, etc.)

  2. Pipeline benefits. RenderMan is designed specifically to sit in a
    larger
    pipeline, showing how this works and how it made the project easier
    to
    manage and complete would be valuable.

  3. Technology benefits. Things that can only be done in RenderMan,
    such
    as procedurals, extreme scene size…

The above list came from Paul Gregory - the primary developer of Aqsis. He is right on most counts there, I believe though that if we try something BIG then things get overly complicated and it will fall apart. MY idea is that we develop a 60 second short (or even less) of a character animation, with some of the more advanced features of Blender such as a cloth and physics simulation (fluid sims tend to come out a bit rough right now), while at the same time showing RenderMan specific features such as displacement shaders, motion blur, DoF and so on. I want to keep it simple for this first test, if it’s kept simple then there is a better chance of it actually being done and fairly quickly too. There is no reason to redo open movie projects for this, that is too big for this idea, at the same time a walk cycle is not enough. Something with character.

This is a serious idea, all I would need is the help from a talented character animator / modeler. For an environment I think I could handle it as well as the shading, lighting and rendering. It’s just that there is doubt from all sides, from the Blender fans to the BtoR devs… I feel alone almost lol (even though the devs are fully supportive of the idea - they just don’t want to get burned if it doesn’t look as well as it should - I agree with that), so I want to put this together and get it out there. The more content that is out there, even if it doesn’t show all the bells and whistles of Renderman, the better. There is no reason to come out with a bang for something small, learning to walk before you run is essential and as far as I know there is not one single character Blender to RenderMan animation (not even a walk cycle).

So this is a call to artists! Anyone who wants to help, anyone at all is welcome. We CAN do this, I would do it myself but again I am horrible at character modelling/rigging and animation. Other than that part the rest I can handle. We can have this done in a month and then at least A goal has been done so that bigger ideas can come forward. That has been the stopping point is that there has not been goals here by artists - only by developers. Developer goals are great and in fact that has been driving the BtoR project greatly - I just think it’s time for some artist goals as well.

#2

nevermind…

#3

The obvious bottle-neck for this project is the lack of a decent blender character rig/rigger. If you guys or someone from the blender community could come up with a decent rig it would be easier to get animators on board, they are always on the look out for a good rig to play with.

Designing and modelling a character is also not that big of a problem, just get someone who can build a good rig in Blender.

Just my two cents.

#4

Addiso,

I’m curious what you think of the Mancandy rig. Both the DVD and rig are freely available online with a little searching.

Cheers!

#5

you can use my character if you want, its free and the rig can be adjusted as pleased.

http://home.deds.nl/~mp/elf/elf.htm

#6

Well then :smiley:

Uh yeah we are totally in need of a modeler and animator, still. There is talk about this now, at first I wasn’t too sure in anyone was interested at all but seems that there is. I had the idea of using Mancandy as well, however there seems to be a urge amongst those involved that having something else would do better. So other rigs or models would be looked at for sure. Unless we happen to get a modeler/rigger. Like I said I would do it myself but that is one area that I am so not good at at all.

And that is one messed up looking creature there ysrvy :)… I’ll take a look at it.

#7

I am currently working on a 2 minute film and I could be interested in chipping something in.
(See: Blenderart Issue 21 for the short “Game Theory”). We’re set to begin “shooting” in 2 to 4 weeks.

I don’t know anything about Renderers other than Blender Internal, and I had only mixed results with Yafray. :stuck_out_tongue:

Is this something I could install? Do all the Material/settings and things within Blender carry over to your RenderMan implementation?

We are thinking about porting over things to a home-made Server Farm running on Loki, and if your RenderMan implementation is mobile, we might see about trying to get it to run there too.

No guarantees… This is just brainstorming.

Cheers. :slight_smile:

#8

Yes in fact on my website (see my signature) the latest post is about a downloadable ISO CD image of the very software we use, and in certain cases - develop. It’s all open source, of course, save for a few usefull programs here and there which are free to distribute and use (have author permission to distribute - like ShaderMan for instance). Not sure about Loki able to run RenderMan software, but there is Dr.Queue on the ISO as well if you want to check it out.

Anyways - resources are resources and even if you personally don’t contribute to this short you are free to use it and give feedback. The Blender to RenderMan website is for the very purpose because some of the software still has bugs so the more people we get interested in any way the better it is for the software developers. Over the past couple of weeks we’ve seen a lot of new people who had no clue about RenderMan get interested in this and since there is no risk of losing your investment (other than time) if you do find out that this is in fact not the thing you would do then there is nothing lost (again other than time). Either way if you do contribute something (anything at all) your effort will not go un noticed. Just head over to my site and go to the Forum Group on the right hand side.

#9

Judging from the demoreel on the site you linked, my first reaction was that however great the rig may be under the hood, the character has really little apeal. Surrealistic proportions, stuck in the nomansland between toony and semi-realistic.

Concerning the rig - again only judging from the reel - the lack of modelled facial extremes (blend shapes/morphs) is also a concern as it only seems to be using “clusters” , that can lead to really limited facial animation.

#10

Hi,

    Just some (random) ideas:
    
    0) re-render Big Buck Bunny / Elephants Dream - probably not creative at all, but definitely would be huge stress test for the pipeline
    
    1) render missing scene from Elephants Dream [1]- there are two shots that has not been rendered because original Orange team didn't have enough time to do it, so they left it out. Movie still worked without it, but AFAIK noone from the community has ever rendered it, mostly because it was huge task in terms of computational power and it kind of got forgotten. That would be something huge if someone step up, and finally render these shots. Plus a good chance entire Blender community would absolutely love it, since noone has really seen it yet.

[1] http://www.youtube.com/watch?v=X8C-IVK1syc#t=8m38 (sorry thats the only info I could find…)

    2) check out these characters, they seem to be pretty advanced and resource intensive
    
    [http://www.jpbouza.com.ar/ESP2/descargas/blenrig-2/id/en](http://www.jpbouza.com.ar/ESP2/descargas/blenrig-2/id/en)
    BlenderArtists thread: [http://blenderartists.org/forum/showthread.php?t=131461](http://blenderartists.org/forum/showthread.php?t=131461)
    
    [http://www.jpbouza.com.ar/ESP2/descargas/blenrig-3/id/en](http://www.jpbouza.com.ar/ESP2/descargas/blenrig-3/id/en)
    BlenderArtists thread: [http://blenderartists.org/forum/showthread.php?t=138117](http://blenderartists.org/forum/showthread.php?t=138117)

Ok, I guess it doesn’t really help much. Anyway, all the best with your project!

#11

We’re going to download the ISO and try to evaluate it for use with “Game Theory”.
Our plans for a Renderfarm are close to falling through so I hope this Renderman thing is FAST on one PC. HAHAHA.

I also hope the guys who made this are up for tech supporting someone like me whose only knowledge of 3rd party Renderers was back when Yafray was still in vogue.

#12

The idea has crossed several people’s minds, in fact a long debate about the subject of rendering parts or even the whole open movie projects happened on BlenderArtists forums recently. In fact that very debate lead to our current project. However as it was mentioned there, even though it would be a great stress test for the tools, not only do we not have the resources to do final renders of that magnitude but there are so many factors needed to convert them into RenderMan ready models that we would spend a good portion of our time redoing everything that the Blender Foundation had done. At this time it would not be a good idea to do such a task when there hasn’t been any animation done period under this pipeline. Not to say that we wouldn’t do this in the future though.

However due to the nature of the test, we have decided going with an original concept and deisgn is our best bet. In fact we are lucky enough to aquire the help from a talented modeller from the UK and as of yesterday have started to work on some of our ideas. We still need some help though since we still don’t have an animator. So far there are about 5 people involved in some way or another, this test is going to be short but we are shooting for the beauty factor here and I am giving a summer limit. If it goes on past August we have failed.

Anyways thanks for the help here, certainly we get better reception on the CGTalk forums than BlenderArtists - some of them over there are pretty ruthless when it comes to infringing on Blender territory. So yeah any and all help is still appriciated. :slight_smile:

#13

I have reached a similar realization to the one above but for my own movie project.

The idea that the switch to Renderman is not a simple “Install and Press Render” for projects already with built-up assets (like my project!) means that it is a daunting task to use it.

Because MOSAIC also seems to have a few “disconnects” with Blender’s native materials and such (ex: Materials in MOSAIC are different in some ways to what we are accustomed to in Blender and the same colors can yield different results in Aqsis and Pixie) we are also in danger of going through a period of re-discovering how to do shaders and materials.

Again, similar to the post just above, re-learning things can be a big obstacle to the adoption of Aqsis and Pixie into projects where the creators expect to be pushing things forward and relying less on rediscovery.

Also the lack of a farm and resource to do something worthy of a Renderman result for a movie longer than 60 seconds. That is another factor.

The solution I have been thinking of… Though I have yet to flesh it out… Is that I would want to use Aqsis and Pixie on my NEXT project instead and nominate Ted and Eric to be Technical Directors who will then handle infrastructure, shaders, lighting, and the farm rendering. :applause: Ted and Eric apparently might have a Dr Queue farm and a perfectly working Blender-Renderman Set. Ted and Eric also have shown they know quite a lot about their MOSAIC and Shaders…Why should I (or any other Writer/Director/Creator) have to re-invent anything if it’s possible to have the very experts on board to have a hand in the desired result for a project conceptualized from the start to use Aqsis/Pixie? (That is, while the subject is on doing a fail-safe on-the-limit output of Aqsis/Pixie with as little waste as possible).

The hitch though is the need to make a workflow where the project could rest on the infrastructure on Ted and Eric’s side while allowing someone like say me to animate the work and send files. So it is just a pipedream right now. :wip:

#14

The character was never intended to have appeal. It’s intended as a learning tool to teach some advanced rigging techniques.

#15

wich is why mancandy may be great for learning,
but might not be the best suited rig for this project,
there are plenty of good free rigs out there… :slight_smile:

#16

Yeah this current project that has JUST started is more or less designed around RenderMan, as opposed to just “fixing” prior content. It is a lot faster to build and texture a model around the RenderMan aspect, even though the geometry pretty much is not a problem, the shading and texturing area does tend to react differently - not a lot but enough to notice. Planning for RenderMan is an advantage in this new project, at least for those of us that have used RenderMan for years, so take for instance our modeler Dan - he has little experience with RenderMan but for his work it’s not really needed at his stage so even though he has not used it much prior it doesn’t affect his workflow. However for my area, shading and texturing, I am using Aqsis only, for previews and rendering. This way I know exactly what needs to change, what looks good and so on.

Some of the very early tests made this past week for various stages of the pipeline are also previewed in Aqsis. This is not only a project to make a Blender to RenderMan short, it’s also a test of the entire pipeline itself. So that also means I am documenting these tests, taking screenshots, making log notes and so forth so that in the future this can be used as a sort of manual or something like that.

Once we get our short done at least I will love to help you out with the Tech Dir. part! For your thing, from what I have read up on and seen, it would not be a benefit to use RenderMan for final renders - as you have said it would require a lot of re-learning and fixing of things already made, not practical. Glad that you did look into it though with an open mind :smiley:

Anyways so far this is going well, we are not going to release much publicly on other forums about this yet because it is so early in the process that if we toot our own horn too early karma is going to rear it’s head and things will go wrong hehe. Actually we just want to get things done first before posting stuff online. It’s not even been a week yet since the initial concept started.

#17

And yes, I have observed the same thing you have Ted, Modellers, Writers, Storyboarders, even to an extent Texture Painters… They don’t reallyl need a setup of MOSAIC/Aqsis/Pixie.

They can actually complete their tasks in their native environment.

The key would be linking YOUR infrastructure as the nexus where the end-result is created. :slight_smile:

I will email you soon in early 2010 about how this might work. :slight_smile:

#18

Well we are coming across some challenges of our own here, things that have not really fully been tested until very recently. Maybe this thread in itself is a CGTalk WIP? Anyways, a lot of the problems that have been anticipated between Blender and RenderMan are coming through now, however due to some learning of WHY this happens, we’ve been coming up with fixes. One problem in particular was high poly count, recently a 1.4 million polygon scene with a lot of SubD modifiers was used to test just how long a single frame would take and the result was not pleasent. The system I use here almost crashed, took forever and I had to kill the process, however I knew it was due to my small RAM amount (640MB) more so than the actual scene. Paul Gregory, the Aqsis developer (who is behind us all the way here), took a look at the scene and offered advice on how to fix it. Dan, the awesome modeler that he is, realized that it was not due to the high poly count by itself - it was more due to the proximity of the vertices and just a bit of re-thinking how to model for RenderMan made it less CPU intense.

Already I have updated Aqsis to the latest Windows build of 1.5.0 (4/26) so that Paul and Chris of the Aqsis dev team can get accurate bug reports and help fix issues as they come along. This makes it worth it for them as well since this project seems to be the first large scale effort to truly test out the software. We may be a small team of 5 people or so but so far in one week we have put together a CVS network, a backup FTP server, around 20 different models, several custom shaders, documentation and a pretty good communication line between Gmail, a project wiki and IRC. The basic foundation is there and it’s all done remotely - yet feels like it’s over a LAN. Now we just have to keep up the flow and by the end of summer this could be done.

Gonna show off some goods here - some of the things we are working on.

(That is the main character by the way - thanks to Dan we now have this wonderfull creature)

Some of the custom shader code includes a concrete shader that can be used inside Blender to make a wide variety of concrete types, plus it also has a nice Wetness function that can make it appear as if it is dry or wet (or moist for that matter). There is a Grime function that adds Turbulence on top of the base concrete color, which also can be mixed in with a texture map, thus it can break up repeating texture patterns. That is not even including a displacement shader which is next in line. Below are some quick test renders, between the two renders was the span of about 2 minutes to change parameters.

(ok grant you the second render doesn’t look much like concrete but it was just a test to see how quick one can edit the shader with the functions I built in)

Well that is all we are gonna reveal for now here, though anyone is free to check it out! We are keeping it open to the public for everyone to watch.

http://groups.google.com/group/blendertorenderman/browse_thread/thread/cda38fd17d5b9fd2

#19

Definitely Ted, there will be a need to ensure that the system works in a farm environment.

Unless there are quantum leaps in Processor Technology, the only feasible way to complete even a short film in adequate time given to not just Rendering, but for Re-Takes, Corrections, and Pre-Production design is to have it in a Render Farm environment.

I have no capability for this, so I hope when the time comes, your team has this ability.
Even a small one may create big gains, so tests in this area using RenderMan are definitely things you should look into.

My other question is a bit… well…more of a “studio/end-user question”. In learning about Aqsis and Pixie, I have read that “RenderMan’s original DNA is actually the forebear of many rendering systems”. So that means to say that both BI and Aqsis/Pixie can land at the same point in time. BI has the extra benefit of having these “fruit coded” movie projects.

Having said that, is there reason to believe that Blender Internal for 2.5 won’t approximate RenderMan results or equate to it?

If there is an arms race with Blender Internal, then that also adds fuel to the fire for ensuring there is a farm available for projects of higher than 3 minutes in length that wish to get Renderman results. :stuck_out_tongue:

#20

There are a few online render farms, they currently don’t run Aqsis/ Pixie, but i think if enough voices are heard from the comminuty (here and perhaps at Ba) then i think that is something they would look at.

Take respower for example, very cheap unlimted rendering, i think it’s something along the lines of 1500GHz, which many not be huge, but it really does take rendering times down (I did a 1550 frame render in blender, with raytraced AO at HD resolution, it took about 3 hours for the whole lot, which was probably way more than X100 times faster than my machine i had at the time).

About this project, i think it will benefit a very large portion of blender users, for example if a studio were looking at using blender with a renderman compliant renderer, at the moment it’s not really possible as MOSAIC has never been production proven (Well not to my knowledge), the same goes for Aqsis, if there is an effort to prove to everybody that blender can be linked in an efficient and robust pipeline to a reyes renderer, then that will surely be a good thing.

I’ll be honest, i’m a little confused about the way it works, but i’m slowly learning how to use MOSAIC and how to get the same results in Aqsis, as the blender internal.

I’ll just add, about the SDS surfaces, i had issues with broken meshes, this was due to, imo edge loops being to close, but i’ve been informed by paul (the Aqsis dev.) that this was an issue with 1.42, this issue is (i think) fixed in 1.5.x, although i haven’t managed to download a copy and test yet.

I think the single biggest diffrence between Aqsis and the internal engine (imo, which may be wrong) is that you have these amazingly powefull shaders which can replace textures maps, you have micro-polygon displacement and when rendering large scenes it seems to handle them quite well, i’m going to stress test it later on today with a scene of about 5 to 10 million polygons (depending on how much blender can handle).

I’ll be sure to document the tests.