The idea behind this personal brief was to create a realistic animal in CG. I chose the Basset Hound because of its interesting shape and personality. I wanted to capture a playful moment in a photo studio where the dog was been photographed.
References were a must for this project, since I am trying to reproduce something realistic. I tried to find a wide range of references from Interesting photos, pose to anatomic photos and closeups of different body parts. Since I wanted to do fur, I also found some nice Hi-res images which showed how the hair flowed and grew.
The modelling process was done in Zbrush, using an anatomic image as a strict reference to get the right proportions firstly, then begin to define the major body parts. The zbrush workflow is detailed in the images here. Once I was happy the modelling, it was retopologized, UV’d and textured.
Using the retopo’d model I posed the dog in the action of turn its head, been attracted by something which caught its attention for a second. The flopping ears added a nice touch to give twisting movement and the opposing legs gave the model a nice forward movement.
I used Joe Alter’s Shave & Haircur for the fur in maya, and Zbrush to style the fur. S & H has the ability to use CV curves as comb / fur guides, which is very useful to style the hair in another external application.
S & H uses two texture maps for the length and growth areas, these maps are important to give your fur a realistic look as fur does not grow evenly on an animal. References are a must and also think about areas of high friction or crease, generally these have less hair.
I created and styled a rough flow guide for the fur which was exported as curves into Maya, again using references to see how the hair flowed. There is no need to import a massive amount curves into Maya, I would say aim for 1/10 of the final and all over coverage.
Import the guides, pay attention to the instructions in the image as these are important steps to make sure the guides work properly with S & H. Curves should be rebuilt to give a nice curve / flow as fur. This can be a CPU intensive process, so be patient. :)
Import the model and make sure it is in real world scale units. Most modern renderers are scale dependent, so you need to make sure your model is in the right scale. Position and scale the guide curves into position also. I would recommend you freeze transform the model and curve guides at this stage, this will avoid problems later.
Grow the fur by selecting your model / shave menu / create new hair. Use the default hair preset. Depending on how big your model is, you will need to tweak the default settings. At this early stage I made sure there was no Kink, Frizz, Dynamics, multistrand or clumps, this just makes the styling process easier to manage. You may need to adjust your hair display settings to show more hair in the viewport.
Try to set a scale that is close to the longest length the hair will grow on the model. Here are my default settings, here is the default hair after tweaking.
Now we want to comb the fur to the style we want, we do this using the imported guide curves. Select all the curves then ctrl select the shavenode, then in the shave menu, “comb using curves”. The trick here is to select the curves only and not the parent group, you can easily do this by hiding all other objects in the viewport and with only burbs curves showing. If everything went well, we should see the S&H hair now styled. You can hide the guide curves now to speed up the viewport. You may need to increase the scale to see the result.
Plug in your two cut and density textures you made earlier then update texture, the fur should now be in a close style as your final. You may need additional tweaking of scale, frizz, kink. The Hair Passes amount will also make a difference to the quality of your render, I set this to 4. For a denser coverage I used multi strand, which is less computational heavy then increase actual hair count. Here is my multistrand setting:
I am using Arnold as the renderer, which comes with its own hair shader that works great with S&H its called the aiHair shader.
The image shows how to connect the aihair shader to the shavenode.
Getting a good render for the fur will depend on the aiHair shader setting and also your lighting setup. I highly suggest reading the Arnold manual for what each of the setting does, here are my aiHair settings.
It's important to note the additional Uparam, Vparam settings, this added flag will tell arnold to render the fur based on the UV texture. Which I simply used the diffuse texture for the dog.
The shading setup for the dogs body was fairly straight forward, I used a aiStandard shader with a facing ratio node to give it a slight peach fuzz / fresnel effect.
Here is a test render with the basic shader without fur.
For the eye I used a pretty simple setup, modelling just the Cornea, Iris and lens. Shading wise the only tricky thing to remember is for anything that requires transparency or refraction, make sure to untick the “Opaque” option in the Arnold tab of the geo shape node. This will tell Arnold to calculate transparency on that object. Don’t forget to use bump mapping on the cornea and iris, this makes a big difference in how your eyes reflect light. I didn’t use any SSS on the eyes.
For lighting I used Arnold’s aiSkyDomeLight and ai_sky as an environment map. Because I wanted to replicate a photo studio shoot it was just a matter of finding a suitable indoor HDR image. Rotate your skydome and ai-sky to a position that suits, using IPR as a quick way to preview the final render. I also made a simple backdrop and floor with a standard shader on it. This gives a great photo studio look and acts as bounce lighting. When lighting, I switch off the fur to get quick previews and set my AA samples to 1 or 2 as I setup and key my cameras. For my final render the single HDR was enough to give great result and very fast to render. If however you wanted more control on the lighting a good trick I learnt from the senior lighter at RSP is to clone out the very hot spots in the HDR and use it just for the ambient lighting, then hand placing your key light to taste. You might like to experiment using this method for better control, just remember to save your edited HDRi in 32bit float linear color.
Here is my setting for the aiDomeLight with a EXR HDRI mapped.
Rendering with Arnold is a simple process, the trick is to find a balance between speed and quality, the Arnold website has great resources to guide you. For my scene the majority of the work will be computing the diffuse lighting and glossy fur, so both of these samples were set higher then refraction, which was only used for the eyes. Generally speaking if you are getting noise upping the sampling with help. AA acts as a multiplier for all the sample types.
This controls how many bounces of each type of rays Arnold will compute, as I wanted a nice looking fur with full GI I set this to 2, generally a value of 1 should give good results. The reflection and refraction are set to 2,4 for mainly the eyes, and glossy set to 1 because we dont need to see the bounce in glossy fur. Generally speaking if you are getting render errors, black spots, check the raydepth.
I’m a fan of getting a final image from the renderer / 3D Package without too much post work, so I did not seperate out each AOVs. You can of course do this easily with Arnold, what AOVs I did use were the Beauty, Ambient Occlusion and Zdepth. Creating a AO Pass is simple, add a custom AOV with the aiAmbientOccusion shader assigned to it. ( See Image )
Photoshop was used to composite the final image, I work in linear as much as I can before switching over to 16bit for color grading and look. Unfortunately photoshop does not handle linear workflow well, I recorded a tutorial to address this before:
Finally Magic Bullet PhotoLooks was used to add a bit of punch to the image, adding bloom, chroma etc.
I hope you enjoyed this tutorial on how I created the Basset Hound from start to finish. I hope to make several other creatures soon to incorporate into a creature showreel. I am a passionate VFX artist that enjoys lighting and compositing, I’ve been learning VFX for 3+ years and looking forward to receiving my first feature film credit. Recently I completed an intensive 10 week internship with Rising Sun Pictures where I trained in lighting, rotoscope, compositing and tracking. You can find more of my work and showreel on my homepage:
Thank you to Autodesk Area for sharing my works.