Jump to content

Hitchcock


abvfx

Recommended Posts

Hey houdniks,

So i have been working on this project on and off for a while and felt its time to finish it once and for all. At the very latest i want it done before the actual film about his life comes out. :)

So here he is in his current state.

ZBrush+document.jpg

balanced.jpg

And a quick blendshape test.

blendshapeTest_001.jpg

I had to fix a bunch of things on the model like make sure there are no holes, since working with point clouds for SSS can be a real pain. And make sure the UV are all set. There are 6 UV tiles at 4k so im working with a fair amount of resolution.

I will post some pics late of the body and cloth renders. And i have begun the furring process. I have 2 fur setups 1 for the head and 1 for the body. But when the character is complete i would love to implement a full FACS facial setup.

-andrew

  • Like 1
Link to comment
Share on other sites

Thanks cubiccube and Netvudu.

Yeah i feel like there were a lot of things i discovered and some still undiscovered practices when dealing with zbrush and Houdini. Most of which are things you should NOT do :), and really are techniques on salvaging data when the model needs to change.

Lots of use on the Attribute transfer SOP, topogun, point sop, polycap SOP when testing.

This model has gone through about 3 topology changes after sculpting has begun/finished. And the model's final resolution is approx. 20 mill.

One quick question, ive been testing speed between VOPs as a blendshape tool and the Blendshape SOP itself, i cant really differenciate between the two even when running it through the performance monitor. Should VOPs be faster? (It feels like it should be, no?)

Anyway, body is coming up next, then cloth, then eyes

Link to comment
Share on other sites

If you do the "morph setup" on the Blendshape tool (ie. using only Edit SOPs for the second input onwards), then it can be a very efficient because it will only blend the points which change between the neutral face with the current shape. (Such Edit SOPs for it can be created using the sopcreateedit hscript command on any two SOPs of the same point count.)

The other optimization that the BlendShape SOP has is that if you tell it to assume that the input point count never changes, then it won't copy the input geometry after the first cook. This used to be a much bigger deal pre-H12 because copying the input every frame was very slow. The VOP SOP on the other hand currently always copy the input fresh every frame.

If you don't have these turned on, then I'm not sure why the VOP SOP would be slow, did you turn on multithreading?

Link to comment
Share on other sites

  • 5 weeks later...

If you do the "morph setup" on the Blendshape tool (ie. using only Edit SOPs for the second input onwards), then it can be a very efficient because it will only blend the points which change between the neutral face with the current shape. (Such Edit SOPs for it can be created using the sopcreateedit hscript command on any two SOPs of the same point count.)

The other optimization that the BlendShape SOP has is that if you tell it to assume that the input point count never changes, then it won't copy the input geometry after the first cook. This used to be a much bigger deal pre-H12 because copying the input every frame was very slow. The VOP SOP on the other hand currently always copy the input fresh every frame.

If you don't have these turned on, then I'm not sure why the VOP SOP would be slow, did you turn on multithreading?

Yes the sopcreateedit function really has an impact i feel. There is about 20k polys in the face so even so this new spangled geometry engine speeds everything up it is hard not to notice. I do think i will need to make a tool to somewhat automate the process as i plan to have upwards of 30 shapes (basing them off FACS). I will have to look back at the VOP SOP implementation again but i think i didnt have mutlitread on... i know... schoolboy error. :)

Jim c: Thanks. They are just simple morphs from the original base done in zbrush, testing it with the base displacement.

I must confess i somewhat forgot about this thread for a bit. About 2 weeks ago i finished the uv map and all of the diffuse textures. Even though there are 6 4k maps I'm sure more work could go into them but i was using a very old version of Bodypaint and i will be glad not to open these texture in them again. I recently purchased 3d-coat for its texturing tools and Houdini hooks so all further texture development will be done in there and future project. (Can't wait to use ptex).

UV's

UVs.jpg

And a quick shot of the back, not with the SSS settings of the face renders, just a col texture test (still some seems to fix), no displacement. Yes he has tan lines :)

backtexture.jpg

And a quick chest hair test, i will be re-doing it, no displacement. (Ive already fixed the nipples :))

chestHair.jpg

Anyways here are some updates, will be posting more soon

Quick question, is it more computationally expensive if i have to load in texture maps for SSS, Diffuse etc when i could use the color correct VOP to get the same effect. Remember there are 6 maps per channel.

Edited by phrenzy84
Link to comment
Share on other sites

Do you feel more comfortable :P

quicksssTest_braces.jpg

Joking aside, I was thinking of doing a surreal scene set in the 60's of him on the edge of a diving board looking at the camera about to jump into a pool infested with sharks. Lots of Houdini goodness

Then it turned into an exercise but i still might do it. I'm using him as a benchmark for a lot of things at the moment. I had already made him a jacket that is just a quick snapshot of the lower half suit all of this in preparation for testing the cloth capture sop. But one step at a time now :)

Link to comment
Share on other sites

  • 2 weeks later...

So its been a couple of days. Finishing off the shader setup.

But ive been having a bit of trouble working in a linear workflow. Every render feel clamped looking at it in a 2.2. Tweaking is pretty difficult, but i think once i setup the basic comp network things became a bit easier. Still im not happy with the results so far. Gonna have to search and find a good workflow to be able to judge what is going on.

compSetupHi.jpg

Link to comment
Share on other sites

So this is a render i semi-satisfied with. Its using both single and multiple scattering.

newLinear8_COPS.jpg

It needs higher samples, better spec, the displacement/bump isnt optimized and dialled in properly so i will be curious to see how fine wrinkles and pores hold up. And i think i need to colour correct the body textures as he looks like he has come from a tanning booth.

But for the lighting setup(2 area's) its "ok"....for now. I will try to make more improvements on it once i get a few aspects of the asset done.

Edited by phrenzy84
  • Like 1
Link to comment
Share on other sites

  • 1 month later...
  • 2 weeks later...

Next post i will share current SSS setting. I must admit the further i get with SSs the more i dislike it and feel it needs something more. If anyone has any tips on a better results i would love to hear them.

I've turned the quality settings down (displace, environment samples etc) for this quick test of the jacket.

jacketTest.jpg

Also is this image blown out? I can never tell if i my monitor is calibrated correctly so i can evaluate render properly.

Edited by phrenzy84
Link to comment
Share on other sites

  • 7 months later...

Thanks Fabiano,

Quick pass on the hair and eyebrows, (shoulda turned on the skin for the render here, oh well next time i guess :)). Wasn't the most demanding in terms of grooming as you can see but still needed to sculpt the hair in blender as opposed to using point attributes in Houdini, the geometry is dense enough to control local sections but in terms of ultimate control, custom guides are needed. Color is point attribute based though.

Gonna move onto the body hair (which will be all point attribute based) and clean up the other maps.

post-666-0-66397000-1376415547_thumb.jpg

Link to comment
Share on other sites

  • 4 weeks later...

So here a a bunch of final-ish images. He is very deep in the uncanny valley, the one thing that really needs work is the eyes, I haven't even got an liquid eyeline layer or caruncle. The clothes don't have anything else but for a colour map. But i will call it a day for now, can't really stand to look at it, but i learnt a lot. Certainly gonna convert a lot of the work into assets ie the skin shader. And he will be used in my facial/general rigging tests along with cloth sims. When you are working with over 20 x 2-4k textures it can get quite tiresome.

A few questions though, can I work with multi-channel exr/tif textures and convert those to rat, ie One texture would have colour, spec, gloss, displace and sss? What is faster?

The headshot that you see in this post of hitch rendered in 14 minutes (8x8 pixel samples) but i can render him in 3 minutes (4x4) with little change in quality, (yes there is less grain but can it justify the time?). Any other tip to improve the render times?

And im very open to criticism, any tips on what you think would make this better?

Bodyhair render (shoulda fixed those nipples :))

post-666-0-51370700-1378511304_thumb.jpg

Clothes render (didnt know if i shoulda modeled wrinkles or not if i want to make it viable for a cloth sim, also not happy at all with the shader/look of it but oh well :unsure: )

post-666-0-57892600-1378511306_thumb.jpg

Headshot default mantra

post-666-0-83536500-1378511309_thumb.jpg

Couldn't resist doing a few grades

post-666-0-81094800-1378511744_thumb.jpgpost-666-0-24268400-1378511747_thumb.jpgpost-666-0-83377300-1378511750_thumb.jpgpost-666-0-80846500-1378511753_thumb.jpgpost-666-0-46719300-1378511759_thumb.jpgpost-666-0-30041900-1378511762_thumb.jpg

And a base mesh wireframe if anyone wants it.

post-666-0-33936100-1378511314_thumb.jpg

Edited by phrenzy84
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...