Jump to content

Recommended Posts

Dear odforce community,

we (WE CAN DANCE Animation Studio) are working hard on our Christmas Animation Short for 2012 - "Santa & Klaus". After several years of half baked attempts this is the first time we really committed to producing a true character animation piece. Everything is being done in Houdini except for the animation which we unfortunately had to do in Maya due to the limited amount of Houdini Rigging and Animation Artists. We are however interested in developing a full Houdini Character Animation pipeline in the future if we are sure to have enough skilled artists at our disposal (feel free to contact us).

With this animation piece we are venturing in several new areas that we haven't done before like skin shading and fur as well as the overall sheer amount of work that has to be done to finish something like this.

I would like to use this thread for three things. First to show you regular updates on our progress. Second to get some feedback (which however might not be implemented as our deadline is rather tight but is definitely appreciated and discussed). And third to get technical help if we might get stuck in an area we aren't very experienced in yet. I will try to post updates and pictures as regularly as possible.

Posts in this thread might be a little bit on the technical side while the artier (is this a word?) side can be seen at http://santa-und-klaus.tumblr.com

We hope you find this thread interesting and entertaining.

Best regards,

WE CAN DANCE Animation Studio

  • Like 1
Link to comment
Share on other sites

To get you started here it goes.

I'm working on the fur right now and ran into several problems. Facts first:

- face rendered with PBR

- fur rendered with MP

- 2 spotlights with DSM

- using the old fur shader and not the new hair model (it gave me better results with spotlights)

It's very hard for me to get a really white fur without losing all the details or getting blown out areas. For now we decided to fix it in comp but I wonder if there is a better way?

Do you think it is better to use the new hair model? I was only able to get decent results with area lights with that.

post-6793-0-49406500-1352544369_thumb.jp

post-6793-0-32696900-1352544380_thumb.jp

post-6793-0-16973400-1352568861_thumb.jp

Edited by dennis.weil
  • Like 1
Link to comment
Share on other sites

Did some more work on Santas Hair. I improved the whiteness of the hair a little bit by carefully balancing the lights and the shaders. I hope that it will hold up when we will light him for a particular shot. The burnt white areas were caused by overlaying transparent hairs.

post-6793-0-29148600-1352748970_thumb.jp

Link to comment
Share on other sites

Here's a little trick that I used while working on the characters eyes. The problem with spheres (or possibly other geometry) is that you get artifacts at the poles when subdividing. Thats because of the high valence points (points with more than 4 adjacent edges). My workaround for this is not to use rendertime subdivision as it recomputes the normals which results in artifacts. The default behaviour of the subdivide SOP is the same. However, if you use the subdivide SOP with the "recompute point normals" toggle switched OFF you get perfectly smooth interpolated normals. Just make sure you have created your normals first with a facet or something similar.

If someone knows a trick to get this behaviour with rendertime subdivisions I would be glad to hear that.

post-6793-0-66024700-1352817233_thumb.jp

post-6793-0-24058700-1352817241_thumb.jp

post-6793-0-13178100-1352817249_thumb.jp

post-6793-0-26766000-1352817260_thumb.jp

post-6793-0-16489700-1352817266_thumb.jp

pole_problem.hip

Edited by dennis.weil
Link to comment
Share on other sites

Hi there Dennis,

Great renders and project. For the sphere subd pole tricks I just usually make sure the poles are in quads. I'm attaching two comparisons before and after. Sorry for not giving you some proper shader setup to show reflections but if you see the first image the sphere with triangle pole got artifacts due render time subd. The second image is after I collapse the edges so that the pole has only quads, if we apply using rendertime subd then we can have smoother pole for the sphere. Mostly this will work out, so you need to check it time to time, if not I just use nurbs sphere instead.

Hope it helps.

Cheers...

post-6390-0-19054500-1353296806_thumb.jp

post-6390-0-94929400-1353296818_thumb.jp

Link to comment
Share on other sites

Donny, thanks for the tip. I remember trying this in Maya and it didn't quite render as clean as I hoped it would. Maybe I should try it again in Houdini.

I'm really behind my schedule with updating this thread, but there's so much work to do that I simply don't find the time. Still, here's a much needed update on Santa. As you can see there are still some rendering errors which we will need to fix.

Is there some tip on getting more texturing and bump mapping detail into PBR renders? I'm always struggling with this as the textures seem to be really blurred. I guess this is due to the fixed half pixel derivative calculations in raytracing renders as this problem is not present in micropoly mode. Is there anything we could do about this?

http://player.vimeo.com/video/54000762

post-6793-0-66100500-1353483312_thumb.jp

post-6793-0-59469100-1353483349_thumb.jp

post-6793-0-26417900-1353487582_thumb.jp

Edited by dennis.weil
Link to comment
Share on other sites

Great characters so far!

Is there some tip on getting more texturing and bump mapping detail into PBR renders? I'm always struggling with this as the textures seem to be really blurred. I guess this is due to the fixed half pixel derivative calculations in raytracing renders as this problem is not present in micropoly mode. Is there anything we could do about this?

Could you show a test case? Normally I see opposite thing :blink: .

Link to comment
Share on other sites

Thanks guys :)

Could you show a test case? Normally I see opposite thing :blink: .

Actually the problem has been the default box filter setting for the texture filtering that gave me bad results. I reduced the filter width drastically and even used the sinc sharpening filter on some textures to get the detail I want. I know this is prone to aliasing and moire effects, but I haven't had any issues with it. Because we are rendering with DOF and MB our pixel samples are already quite high which results in great AA even with extremely sharp textures. Which settings do you generally use to get the texture detail?

We are using a gaussian pixel filter with a 1.5 1.5 width to get the image even sharper. I'd rather soften the image in comp than not having all the detail in the first place.

Are there some other pitfalls with these settings that we need to be aware of before starting to render the shots?

Link to comment
Share on other sites

  • 2 weeks later...

Hey Guys,

sorry for not updating regularly, but we're in rendering hell right now. And we learned a painful lesson: Never ever use the png format for your textures, especially not for really big ones. Our background artists thought it was so convenient to just save a 10000 x 3000 PNG from Photoshop and use it directly in a shader. Our rendering times went up from 30 minutes to 5 hours a frame. Strangely it only happened when the backgrounds were set to matte or phantom. When we rendered them with primary visibility all was well.

My guess is that this is due to the way mantra keeps texture maps in memory for primary visible objects versus matted or phantomed objects. Anyone has some more info on this issue?

Interestingly the render times imporoved to 3 hours when we switched from kd tree to bvh as our ray tracing acceleration structure. Again I guess it is due to the way things are handled and kept in memory.

I'm hoping to post some renderings today.

-dennis

Link to comment
Share on other sites

Hey Guys,

sorry for not updating regularly, but we're in rendering hell right now. And we learned a painful lesson: Never ever use the png format for your textures, especially not for really big ones. Our background artists thought it was so convenient to just save a 10000 x 3000 PNG from Photoshop and use it directly in a shader. Our rendering times went up from 30 minutes to 5 hours a frame. Strangely it only happened when the backgrounds were set to matte or phantom. When we rendered them with primary visibility all was well.

My guess is that this is due to the way mantra keeps texture maps in memory for primary visible objects versus matted or phantomed objects. Anyone has some more info on this issue?

Interestingly the render times imporoved to 3 hours when we switched from kd tree to bvh as our ray tracing acceleration structure. Again I guess it is due to the way things are handled and kept in memory.

I'm hoping to post some renderings today.

-dennis

You should get a big speedup if you convert your textures to .rat format. Highly recommended - especially with huge texture maps like that.

Link to comment
Share on other sites

Thanks Donny, glad you like our project.

We are rendering in several layers like background, floor, characters (each of them individually) as needed. Additionally we are generating a lot of AOVs. For that we are creating light groups which contain several related lights (fireplace, tree, fill) and create diffuse, reflection and refraction AOVs for each of these groups (and a separate indirect reflection AOV for object to object reflections). Unfortunately we couldn't get per light (group) exports working for sss so this is an all in one AOV, but I guess this is by design of the physical sss model. Additional AOVs are a unique material id as well as an object id which are used extensively in compositing and heavily reduce the requests to render additional masks. We render everything with MB and DOF because it makes things so much prettier and easier to handle.

Nuke is our primary compositing software because unfortunately COPs are not powerful and fast enough to do some serious comp work.

Best,

Dennis

Link to comment
Share on other sites

Hey guys, we're having a really bad and urgent problem with our hair and I'm hoping someone might be able to help us.

When rendering the hair with motion blur it gets a lot brighter when there is a lot of blur. Has anyone experienced something similar? Would be great if anyone could help us out because we are running out of ideas.

The image attached shows three consecutive frames from the animation.

post-6793-0-19930400-1355091151_thumb.jp

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...