Thursday, March 21, 2019

My ArnoldGPU (betta) test with Yeti

As a groomer, I always interest how works new render features with hair and fur, so I did the test on my Ringil workstation new version of MtoA 3.2.0, Maya 2018.5 and Yeti 3.1.7
2xXeon 2683v3 (28core all turbo 2.78Ghz) vs AORUS GTX 1080 Ti Xtreme Edition 11G, Nvidia drivers 419.35 and 419.67.


The render time with approximately the same level of grain (CPU/GPU): 4m38sec/3m34s
I understand that Arnold GPU is the betta and PeregrinLabs not yet release Yeti for new Arnold core, but: no support Maya light links, Hair shader need tweaks to "CPU" look (incorrectly solution tweaking by increasing Indirect Spec)

  "CPU" looks  (419.35 driver) 4m38sec

"GPU" original version (419.67 driver) 3m27sec

"GPU" corrected version  (419.35 driver) 3m34sec

After updating NVidia driver to 419.67 the render time little decrease: 3m28sec (3% faster) and after overclocked GPU on 25Mhz (from 1999.5 Mhz up to 2025) render time: 3m25sec (4.3% faster than on very first result)
Also if you have small black dots (I had this issue on 419.35 version of display driver) just reopen ArnoldRenderView

Although as for betta looks hopeful. Keep rocking Arnold Autodesk team


Thursday, March 14, 2019

Tutorial: Nontrivial use of the Shader node in Yeti 3

Short tips and tricks video about use Shader node in Yeti3
Method in action







Maya example scene (Maya 2018, Yeti 3.1.X)   

I hope this information will be useful to you

Friday, September 14, 2018

Cheetah Family

My latest work is devoted to these wonderful animals, which soon may be lost.
I tried to achieve photorealistic similarity to real animals in their habitat, and I hope you will like the result.



All grooming work and grass scattering I did in Yeti 2.2.8.
Yeti tree of adult cheetah

Shading and rendering in Arnold (core 5.x.x) in which I used aiStandartHair and melanin centric texturing:
Fur shader of adult cheetah

crypto_object pass

viewport

Fusion CC tree



More info coming soon...

Statistic:
Total fur density ~19.5M fibers
Total blades of grass - 108453
Ram used during render ~39Gb
Render time- 9h51m on Ringil workstation
Also, you can buy the model of the adult cheetah here   or 

Monday, May 21, 2018

Tutorial: RGBA additive masks


A short tutorial dedicated to a very simple and very useful technique for creating RGBA additive masks

Friday, December 1, 2017

The Baby Cheetah in studio

 few variation CC





HI-res:  





render.ru and posted in Render Magazine 01/2018


CG Society Award




Statistic:
Fur density - 3736698 fibers
Body fur consist of 7 layers and have only 3 fur system
Render time on Ringil workstation res 3032x4096px- 3h56m
Usage memory when rendering 4k res ~ 9Gb
Body mesh: 3udims, 39362poly
Groom textures: 2 additive RGB textures (2k and 8k), 3 luma texture (2*8k, 1*1k)
SSS textures: 5*8k
Fur textures: 4*8k
Soft: Maya 2017, mtoa 2.1.0.1, Yeti 2.2.4



Tuesday, November 28, 2017

WIP: The Baby Cheetah


some screenshots of my current project with goal to obtain photorealistic model the baby cheetah




Grooms

Thursday, September 14, 2017

Yeti and Arnold or V-Ray motion blur bug fix

If you need to get a working motion blur using V-Ray or Arnold and you have had such problems:

V-ray (normal/with bug)

you need to make a few changes.
If you use in your Yeti tree any expressions with Position attribute (P) then you need:
For Yeti 2.1.xx
1. for all objects from which you get P attribute you need create texture reference objects;
2. replace all P on Pref
For Yeti 2.2
1. the same step as for Yeti 2.1.xx
2. P -> Pundef

If you have this bug and you use VRay, then render time increase approximately x3.3

Friday, September 1, 2017

nvoglv64.dll crash Maya 2017 - Solution to the problem

If during the rendering by Arnold (using IPR Arnold RenderView) your Maya dies and the screen blinks and you find such an error


 then you need to set for Maya app. in Nvidia Control Panel: 3D Settings\Manage 3D settings:

OpenGL Rendering GPU - << select your card in list >>
Power management mode - Prefer maximum performance
Vertical sync - Use the 3D application setting
Maximum pre-rendered frames - Use the 3D application setting


in my cases it was with Maya 2017 Update 4 and Nvidia driver version 385.41

if I correctly understood this is due to the fact that during the time of work in Maya, the video card resets the frequencies and operating mode to the minimum and when the load is needed to update the viewport the driver does not have time to boost the video card and as a result the library crashes.

If this did not help you then try these methods:

  • Try to download and install other version video driver, but before install you need complete remove previous drivers and other utilities. For this task use Display Driver Uninstaller (DDU) . Also check  and uninstall Vulkan Run Time Libraries. Then install Nvidia Drivers and in custom mode select only drivers and PhysX. Do not install any Ansel and GeForce Experience 
  •  Other tips - change the Graphic device timeout in registry editor.

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers

With the GraphicsDrivers key selected, on the Edit menu, click New, and then select the following registry value from the drop-down menu specific to

For 64 bit Windows
a. Select QWORD (64-bit) value.
b. Type TdrDelay as the Name and click Enter.
c. Double-click TdrDelay and add 8 for the Value data and clickOK.

or create *.reg file and paste content below

Windows Registry Editor Version 5.00

[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers]

"TdrDelay"=hex(b):08,00,00,00,00,00,00,00



Sunday, April 16, 2017

Intel Xeon E5-2683 v3 in Ringil workstation

My new workstation Ringil, created based on engineering samples Xeon E5-2683 v3. Some of the components I used from the previous computer and the current characteristics are - 2*Intel Xeon E5-2683 v3, Asus Z10PE-D16 WS, 2*32 GB DDR4 2133 MHz Kingston (KVR21R15D4/32) Reg. ECC 16-15-15-36@1066, SSD Samsung 960 EVO, Gigabyte GeForce GTX 1080 Ti Aorus Xtreme Edition 11G‎ (on photo old GTX 680 4G). I plan to install more 64 GB RAM. Briefly, the machine is very different from the desktop, for our CG tasks a very profitable investment considering the price of such processors (~ $350). But as a user who used desktop computers for a long time, it is necessary to get used to the quirks of server-based motherboards, and almost insignificant (about 3-5%) overclocking on the bus of the processors themselves.
Link to the table of comparison some tests using Maya, Arnold, RealFlow, Zbrush and synthetic tests





















After BIOS modification and all core turbo on (all cores works on 3GHz)

Arnold rendering (102Mhz*28):

Real power consumption: 
Idle ~300W, Rendering (2.75GHz) ~500W, Rendering and GPU (GTX1080Ti) full load ~750W