Fur Render Enhancement

   63520   89   10
User Avatar
Member
1533 posts
Joined: March 2020
Offline
circusmonkey
Do you know of an example movie where the old school technique has not been used ? Id love to check it out .
rob
“but in my opinion there's a slow shift away from spending hours baking out countless bits of information before we can even get to render. ”
meaning slooow, so no i don't have any examples, but i know ILM, Sony and Weta seem to be using more ray-traced techniques.

jason
HOD fx and lighting @ blackginger
https://vimeo.com/jasonslabber [vimeo.com]
User Avatar
Member
175 posts
Joined: July 2005
Offline
Houdini is a well established consistent environment, which gives you “creative freedom” you Never Even Smell in any other 3d app … at some cost.

It wasn't too religious, was it? :wink:
User Avatar
Member
1533 posts
Joined: March 2020
Offline
lol amen to that!

j
HOD fx and lighting @ blackginger
https://vimeo.com/jasonslabber [vimeo.com]
User Avatar
Member
350 posts
Joined: Jan. 2008
Offline
Hi Nick,

lor
Honestly guys, I don't see what everyone is getting so upset about.
I just wanted to register a message of support.

It seems like you got a lot of flak simply for suggesting that the default out-of-the-box fur and rendering experience should be improved. I agree wholeheartedly - it should. While, at the moment, a TD can get some amazing results out of fur and Mantra, personally I think the defaults should be a lot better (this goes for a number of areas in Houdini, not just fur). There's a huge number of users who simply want to throw down a few presets, change a few parameters, and have a pretty picture as a result. In many areas, I'm one of them myself.

I think a lot of what you see is a bit of a defensive reaction along the lines of “defend Houdini, it's great as it is!” - i.e. for some reason, your comments seem to have been perceived as an attack. To me, they're extremely valuable suggestions for improvement - this is exactly what we need even more of.

And just to conclude with a disclaimer, all of the above is my own personal opinion, and is not related to SESI's official position in any way. I just wanted to let you know that we are listening and taking notes, though sometimes silently.
Oleg Samus
Software Developer
Side Effects Software Inc.
User Avatar
Member
1533 posts
Joined: July 2005
Offline
I think a lot of what you see is a bit of a defensive reaction along the lines of “defend Houdini, it's great as it is!” - i.e. for some reason, your comments seem to have been perceived as an attack. To me, they're extremely valuable suggestions for improvement - this is exactly what we need even more of.

Hello Oleg, et al,

I think the defensive reaction that you are seeing is primarily based on painful past experiences.

It usually goes something like this:

- A ‘feature’ is needed in production.
- Someone spots a fancy demo that promises the world.
- Minimum testing takes place.
- A recommendation and purchases are made.
- Production ensues. Problems arise (and you know they do).
- Artists shrug their shoulders.
- TDs, whose job it is to overcome (and possibly build their own) the limitation of these system don't sleep to get the job shipped.

This, imo, is where Houdini shines. That last step.

As for presets, again, no disrespect to anyone, I haven't seen a single one that worked (for production scenarios) since I started roaming the Houdini multi-verse (prisms).

Again, please don't misunderstand me… I'm not suggesting that your RFEs should be ignored, or are worthless, but I *AM* suggesting that I would rather see focus and emphasis put towards low level tools that will help production TDs CREATE presets that WILL work within your respective pipelines.

It just comes across a little bit like ‘I want running water without all those pesky pipes’.

G
User Avatar
Member
732 posts
Joined: Dec. 2006
Offline
A quick word of support for IOR as well, (and he is my co-worker, so I'd be a danged fool to enrage him!).

Everyone on this forum loves Houdini and we want to see it succeed. Part of this is convincing producers and other pencil pushers that Houdini is a viable solution. And this might mean, as much as we hate it, “Okay, you've got a day to show me what Houdini fur looks like.”

After half a day, you have a stubbly ball that looks like My First Fur Render VS Cute Furby Ball, what do you think is going to happen? No one is going to listen to your protests that six months down the line they're going to be sorry.

So yes, reasonable, aesthetically appealing presets are not just frippery for idiots. I have been using Houdini since Alpha, and this is where I often see Houdini fall down. They create some great set of tools, but fail to showcase it in such a way that said tool ever gets their due. And I have sympathy for SESI because I know they're a small company and they don't have 20 demo artists hanging around making presets.

Finally, if someone has a conception that a tool is hard to use, just shouting at them isn't going to help. If someone who is a member of this forum believes this to be true, imagine how those who are not on this forum feel about this same tool.
Sean Lewkiw
CG Supervisor
Machine FX - Cinesite MTL
User Avatar
Member
292 posts
Joined: Dec. 2007
Offline
Good Morning folks.

I appreciate the calmer remarks. For a second I thought I was going to get eaten by a swarm of piranhas. I don't mind getting told I'm dead wrong, which I am as often as anyone, but it's nice to have a little courtesy.

It's more clear to me why there was a negative reaction, so thanks for that clarification. As to the “water without pesky pipes” remark (a good analogy) let me say this. It is obvious to me that Houdini is a powerful tool in the hands of a skilled TD, and I would never suggest removing any of that power. I do believe in artist functionality being built on top of that power. And I do believe it is SESI's job to do that, because if SESI plans to expand it's market share to artists, artists need to be able to go from one studio to the next and know how the tools work. If artist tools are custom at each studio, then an artist will have to train on the tools every time they start a new gig. That's just way too expensive. Some custom tools will work well, some won't based on the skill and experience of the TD. Output and bidding become unpredictable. Consistency of artist tools (and TD tools, for that matter) is a Very Good Thing in a world where margins shrink yearly.

And it may very well be that the piping needs a look long before the water flows. What I say to that is, “well let's call in the plumber”.

No doubt Houdini TDs will always be in high demand to create custom tools. But Basic Functionality™ should be readily available to artists (non-TDs). SESI has already made good movement in this direction. My remarks on fur tools are intended to be merely a furthering of this discussion which is already happening.

I enjoyed Keyframe's painfully accurate description of many typical productions. I am also aiming to change this experience at our studio, as are others. And I absolutely agree that where Houdini shines must be carefully guarded and nurtured, not sacrificed for black-box tools.

Nick
User Avatar
Member
292 posts
Joined: Dec. 2007
Offline
Oh, and just to clarify/reiterate where I'm coming from, my main concern regarding fur ATM is shading/rendering, not shaping/styling or motion. I believe we can accomplish all the styling and motion we need right now with current tools, although they could always be better. I am not convinced we can shade or light fur adequately, or render such fur in a reasonable time, nor have I seen any really beautiful photoreal fur from Mantra yet. I've seen some good fur, but not photoreal.

I'd love to crack that nut. I've been working on this off and on for about a year, but have been unsuccessful so far.

Sean you are wise. I enrage so easily.

Nick
Edited by - July 29, 2010 14:50:54
User Avatar
Member
168 posts
Joined:
Offline
oh my ..

yep, that changes everything…….
Albert
User Avatar
Member
678 posts
Joined: July 2005
Offline
lor
Oh, and just to clarify/reiterate where I'm coming from, my main concern regarding fur ATM is shading/rendering, not shaping/styling or motion. I believe we can accomplish all the styling and motion we need right now with current tools, although they could always be better. I am not convinced we can shade or light fur adequately, or render such fur in a reasonable time, nor have I seen any really beautiful photoreal fur from Mantra yet. I've seen some good fur, but not photoreal.

I'd love to crack that nut. I've been working on this off and on for about a year, but have been unsuccessful so far.

Nick

Man, if you play with this one year and only thing you could achieve is shity ball with ugly fur as hell, while others could achieve definetelly better results I say you should stay far away from Houdini, it's not application for you. One year and you couldn't come even close to result in link I gave you. I don't care if I ofended you right now but personally I would be ashamed if, as a proffesional artist that you claim you are, I could achieve only this.
User Avatar
Member
292 posts
Joined: Dec. 2007
Offline
Wow! Well I guess we know who the professionals are now. :shock:

I wonder where you got the idea that it took me a year to make a ball with a basic fur shader assigned at default settings, especially after I clearly explained that it was a quick set-up for the purpose of a simple example.

Someone needs a hug!
Edited by - July 29, 2010 15:10:40
User Avatar
Member
1533 posts
Joined: July 2005
Offline
Thanks for the feedback Sean and Nick. Your perspective is much clearer now.

I think where we differ, philosophically speaking, is that I don't believe that high-level artist tools can ever be effectively supplied by a software vendor (then again, if anyone could, it would most likely be SESI).

How can we ask them to deliver production-level assets, if they don't do production? When you consider the parameters required to handle the ridiculous client demands, the mind truly boggles.

The other ‘discrepancy’ I'm seeing is that for some strange reason (it seems to me) people believe that rendering ‘should just happen’.

For instance, I've never heard anyone asking for ready made (production level) animation and deformation character rigs. I've never heard anyone asking for (production level) explosion presets. I've never heard of people asking for a parametric walk cycles – and you should see how animators freak when you show them Endorphin. People expect to create custom solutions for each of those cases, and that's never to the detriment of any software package.

Maybe (again, i'm just pontificating here), (production level) fur isn't a point-and-shoot operation at this stage? Just like volume rendering wasn't accessible to all but a select-few (with the necessary support staff), not 5 years ago.

Maybe we all need to do a better job of explaining to assorted pencil pushers the difference between a demo, and production pipeline. I've been in this situation, and I feel for you. I really do.

I know for a fact that the people (lord, I wish I was one of them, sadly, i'm not) with the necessary skillsets to accomplish these tasks are out there – and I don't like to see their abilities marginalized because people perceive that it ‘should just happen’.

I've obviously slid into philosophical territory here, and for that I apologize.

Thank you all for your feedback. I've learned lots.

Best,

G
User Avatar
Member
678 posts
Joined: July 2005
Offline
lor
Wow! Well I guess we know who the professionals are now. :shock:

I wonder where you got the idea that it took me a year to make a ball with a basic fur shader assigned at default settings, especially after I clearly explained that it was a quick set-up for the purpose of a simple example.

Someone needs a hug!

I never sad you worked on a ball one year, just that you play with fur in H on and off one year. And you writed that you played with ball 4 hours and still you failed.
So please, show this ball after those 4 hours so we could see what it looked after you worked on it.
BTW. after year of on and off playing you really heavn't done any presets for yourself to achieve better look of fur faster ?

Hug. :wink:
User Avatar
Member
292 posts
Joined: Dec. 2007
Offline
Keyframe your points are well taken. I think it _has_ to start philosophically, as that drives everything. Without a philosophical approach, you'll meander, so I certainly don't have a problem with it.

I'm very interested in your point about SESI providing production solutions if they don't work in production. And that is an excellent point. In many cases they can't. Enter the production TD. However there is a set of Basic Functionality™ that should be in there, much of which currently is not. I guess this is due to Houdini's long past as a hardcore TD tool. I think as more and more people recognize the power of the software, they want to use it in more and more ways. I believe this is what is driving the call for artist-level functionality. SESI recently spent about 6 hours at our facility asking questions to this end. I believe high-level artist tools can be achieved by SESI in an evolutionary way. And if it's not fast enough, we always have our Houdini TDs. In the end, TDs will be kept for more important tasks than simply making functional artist interfaces.

I'm not sure I'm clearly understanding your remarks about nobody asking for production level anim rigs, explosion presets or parametric walk cycles. Because these things, or presets for them, already exist, and they are extremely useful in giving artists and TDs a head start on builds and shots. Could you clarify?

Animation can certainly never be replaced by, say, mocap. Just as photography will never replace painting and lighting artists will never be replaced by HDRI solutions. Lighting and shading tools should always offer the opportunity to forget about reality and do whatever you want. However many of us, especially those in VFX are driven to photoreal standards. And within that fairly restrictive parameter there are only so many ways a glass ball can look. Or a wooden chair, or calico cat fur. Real world shading attributes are simple and few, meaning that real materials and real lighting are actually very simple. I believe that a fairly limited set of parameters can be used by artists to create most materials…say 95% of all materials a VFX lighting/shading artist will ever need. SESI has already demonstrated this is possible with their in-progress Mantra Surface Shader.

Maybe it will also help to clarify my position if I explain that I became a supervisor for the express purpose of explaining reality to pencil-pushers, and to protect the crew against ludicrous client demands. I have never admired the job of a VFX Sup, but I just couldn't take it anymore. After years of being forced into positions of failure by supervisors who had never worked behind the box, I realized it was my responsibility to either try and change things, or quit. Well I'm not much of a quitter.

So speaking philosophically, I guess SESI will decide where their philosophy lies regarding this discussion, and then make the decisions they think are best for the community as a whole. And then the companies who need to make photoreal imagery for smaller and smaller margins will examine their own philosophy and decide if it coincides with SESI's. It's great that we have a forum where everybody can make their feelings known to the developers. Ain't free speech grand? :twisted:

Hopefully that all makes sense.

Nick
Edited by - July 29, 2010 19:29:14
User Avatar
Member
732 posts
Joined: Dec. 2006
Offline
Swann_
Man, if you play with this one year and only thing you could achieve is shity ball with ugly fur as hell, while others could achieve definetelly better results I say you should stay far away from Houdini, it's not application for you. One year and you couldn't come even close to result in link I gave you. I don't care if I ofended you right now but personally I would be ashamed if, as a proffesional artist that you claim you are, I could achieve only this.
Very classy Swann.
Sean Lewkiw
CG Supervisor
Machine FX - Cinesite MTL
User Avatar
Member
2624 posts
Joined: Aug. 2006
Offline
Well on positive note , I looked at the fur tutorials last night , ( never looked before ) , found a new user will come to a dead stop as theres no fur material to apply in H11 , so maybe a good start would be to get 2 basic fur / hair materials rolling ….


“but in my opinion there's a slow shift away from spending hours baking out countless bits of information before we can even get to render. ”
meaning slooow, so no i don't have any examples, but i know ILM, Sony and Weta seem to be using more ray-traced techniques.

jason

I only say this as I always thought if your rendering hair or feathers , your basically rendering a curve , so if you have lots of curves x lots of characters , when you render you have the whole scene loaded into memory with raytracing. which can become very heavy to process.
I do know from the current show that I am working on that we use raytracing coupled with an LOD for super close up shots where detail is required but other than that its depth map shadows all the way. Maybe some firms are using raytracing for single characters or where scenes are light enough to get away with it.

r
Gone fishing
User Avatar
Member
12604 posts
Joined: July 2005
Offline
circusmonkey
Well on positive note , I looked at the fur tutorials last night , ( never looked before ) , found a new user will come to a dead stop as theres no fur material to apply in H11 , so maybe a good start would be to get 2 basic fur / hair materials rolling ….

Damn.. .another casualty of the recent clean-up of VEX shaders. We do have many people here missing the old trusty VEX Plastic/Metal/Clay's too.
Jason Iversen, Technology Supervisor & FX Pipeline/R+D Lead @ Weta FX
also, http://www.odforce.net [www.odforce.net]
User Avatar
Member
1533 posts
Joined: March 2020
Offline
hi

for me it's not about having presets at all, i've been rendering for a good few years,
i'd say from a artist or even an audience point of view, we've reached a really high level of image quality, even at a commercials level with very short turnaround time, we can't just render something that looks ok anymore, it needs to feel well “physically correct”.

what i guess i'm trying to say is, the fur/hair shader/procedural needs to be optimized for PBR
and of course PBR needs to be further optimized. and from what iv's seen where SESI has taken PBR between version 10 and version 11, i'm pretty confident they can do this in future versions

jason
HOD fx and lighting @ blackginger
https://vimeo.com/jasonslabber [vimeo.com]
User Avatar
Member
279 posts
Joined: July 2005
Offline
Just to enter the fray here since this is an interesting discussion .

I'll admit that initially I scoffed at your mantra furball Nick since I knew that it is actually easy to get something decent out of mantra. Then I tried it in v11 and …er… well the scoffing stopped pretty quickly since my first attempt looked pretty much the same as yours.

Having said that though with a bit of perseverance I managed to get the image below. It took about 20 minutes to setup and 8 minutes to render. Ok it's not quite as nice as the Maya render, but again… I've got infinite freedom from here to improve on it.
Anyway, while digging through the thing trying to get it to look good I've discovered some bugs (and an old one found by other guys here):

1) I can't render with the skin and fur at the same time. It took 45 minutes to do 4% of the render with the skin turned on (clay shader). Making the skin a matte brought the render down to 8 minutes.
2) You'll notice the few pieces of fur on the left shooting off to inifinity…. eww..
3) There's no spec on the fur since that part of the shader is broken.

So bugs aside, here's what I think needs to happen. With the new mantra surface paradigm (I said it!) simplifying a lot of things, we need a similar improvement to the fur. It's no use changing your workflow to throw away terms like specular and the like for everything, only to have to dredge them out again for a fur render. If that happens then it will be quicker and easier for the beginner to render something that looks good.

My .2c
M

Addendum: While writing this I fiddled with my mantra node and found that rendering fur with micropoly and ‘KD-Tree’ Ray Tracing Acceleration is deadly. Switching it to Bounding Volume Hierarchy makes the fur render as you would expect.
Addendum 2: I've submitted all these bugs to SESI.

Attachments:
furball_11.0.446.jpg (137.0 KB)

User Avatar
Member
2624 posts
Joined: Aug. 2006
Offline
Having said that though with a bit of perseverance I managed to get the image below. It took about 20 minutes to setup and 8 minutes to render. Ok it's not quite as nice as the Maya render, but again… I've got infinite freedom from here to improve on it.

I think it might help nick if you could post the scene up …..
Gone fishing
  • Quick Links