I started this project about two months ago and it is finally finished
I began this project as a remake of an old project I did over two years ago, which had about 2 million polys and was really sloppily made the goal with this project was to try to keep everything clean(er), and use a lot less polygons to create the meshes
I also wanted to use a bump map to get in a lot of the smaller details, like pieces of text and icons in the gun’s body to keep down on polycount
Adding the ability to make a seamless FK/IK switch in my rigs has been something I have wanted to add to my rigging toolbelt for a while, and I’ve finally gotten around to learning how to do both an FK/IK switch with just constraints and a custom attribute, as well as a full on seamless FKIK switch
I also added a check for if the objects had namespaces or not, which doubled the length of the code as I have no idea how I would check/apply the attributes properly with any other method and it is way easier to just check when doing the switch :<
Currently I dont think it works perfectly for animation, but a fix for that should just be to just add keys before doing the switch
I wanted to try this study project as I was aware that my lighting and colour grading knowledge for 3d was completely lacking, and use some films as the reference for the lighting and colour grading I got some tips from a lighting artist in a chat I’m in to light in Arnold how the lighting teams would in real life, and then colour grade for the final look after the fact, which I tried to do for this project
The Great Gatsby
A low warm light to light up one side of the model’s face, and a low cool light behind to give the model a slight rim light at the back
Also got given the good idea to light the scene like it would actually be lit in real life and then give the dark shadows and bright highlights while colour-grading in post, and this really helped with how my renders looked
I’m also using my custom LPE AOVs in these lighting tests, and rendering out a Diffuse, Specular, and Direct Emission (basically just the HDR as a background) pass to have more control while colour-grading
The second movie lighting study that I did was from Ex Machina
This lighting setup was pretty close to the last one, but way less intense slightly orange light on the left, low intensity blue rim light, bright background, and a lot of the character’s face (slightly) hidden in shadow
This setup was the only one that i used Subsurface Scattering, partly because it gives a nice subtle red tint to the (camera left) side of the nose, and because Arnold 5.4 dropped with Randomwalk V2 on GPU and I wanted to try that out
My partner has this little dragon holding a shiny ball model sitting on her desk, and I’ve wanted to try out photogrammetry using this model for a couple months now and finally got around to it
The final model is about 450k faces
I used MeshRoom to do the point cloud/meshing/texturing, and it took about an hour for the full compute, after a few hours of me messing about trying to get it to work properly.
The ball that the model is holding is way too reflective to give a good model, but fortunately it’s a super simple shape so I just deleted the faces, placed a sphere where the dragon’s hands are, and added a reflective material to the sphere as a replacement
Some of the mesh turned out really good, like the scales on the body
Some parts really lacked mesh detail though. Especially on the back, underside of the tail, under the chin and under the hands
In future scans i’ve got to make sure that I’m getting every part of the model in high res while I’m taking the photos, as I definitely missed some parts of the model on this attempt
The mesh is made using a six sided tube, with slightly random scales set on the sizes and widths for each, copied to scattered points. Then a Boolean is added to each mesh with a slight change in rotation on that to give the crystals different rotations on the top cap. the VDBfromPolygons makes the entire poly set a VDB Volume, and then converted back to polygons as a single mesh. Finally a PolyReduce node to take the mesh down from 1.3mil polys to 20thousand polygons.
I’ve been messing with using Houdini over the past couple weeks, mostly for Pyro sims (explosions are cool, sue me), and have been trying to export the volumes to Maya for rendering with Arnold. I’ve finally got it working to a point that I’m mostly happy enough with, but will definitely need changes in the future.
Adding collissions to the simulation gave me over an hour of issues with the smoke sim clipping through the collision object. Initially I tried just adding more Substeps to the PyroSolver node, and making the collision object thicker, but neither worked. Luckily everything was ‘quickly’ fixed completely by adding a GasEnforceBoundary node to the Advection input of the PyroSolver. I also changed the StaticSolver to an RBDSolver, but I’m not sure that helped fixed the issue. (problem screencaps shown below)
Currently it’s still not a perfect process between Houdini and MtoA. Firstly is the emission colours. I have absolutely no idea how to have a gradient ramp based on the intensity of the Heat channel emission, so it has to be one single colour. (BlackBody doesn’t seem to work either, and I have no idea how to export that)
Arnold GPU Beta also doesn’t support OpenVDB volumes yet, so rendering with that just straight up doesn’t show the volumes. Having to render volumes with CPU feels like slow pain. luckily none of these test frames took more than a minute each at [3,2,2,1,1,1 // 10,2,2,4,0,10]
Another of the issues currently is that turning on Motion Blur while rendering gives the aiVolume a blue tinge through the entire bounding box, making the renders basically unusable.
I have no idea if this is an aiVolume Container issue, an AiStandardVolume issue, or a Houdini issue from exporting either the Heat or Vel.x, Vel.y, or Vel.z channels
Unfortunately, rendering out Motion Vectors and compositing the blur in Fusion doesn’t seem to work as nothing is rendered out in the MV Pass, so I can’t do that 😦
Overall, it’s definitely fun as hell to do simulations in Houdini and exporting them to Maya for rendering, but I’m definitely going to need a load more practice before I feel comfortable