Class 11: Midterm Check-In

I have a lot of concerns, I will list them here: Several things: firstly, figuring out why textures keep disappearing randomly on only some objects when I submit to the farm (gonna also add texture consistency as well, mainly the wood/stairs roughness and general shaders like that). Secondly, continuing to get lookdev consistency across my files. Thirdly, animation refinement/tweaks (not affecting the timing though). Fourthly, getting a lighting look that I like without it being super over exposed (I think I'm on the right track right now, but it's very difficult to tweak without having to re render everything; that continuous camera move is making things so hard!!!). Render times might become an issue when going through the denoising process; going to look into how to reduce GPU render times as well. Also getting the car to hit the paper weight again (not sure what happened here, but I am quite upset about it!) also some weird camera line ups; need t double check the accuracy of the camera/transforms between the second and third parts.

HOWEVER, I am very excited with the way this project is starting to shape up :)

Class 10: Finishing Textures in Part1

All I really had left to do in the first part of the sequence as far as texturing goes was the marble machine, marbles, stairs and some other basic props, so I really wanted to knock that out so I could focus on setting up the second part and getting some textures in that section for our class 11 check in. Brittany was able to get me some track support models to replace my proxy geometry there.

Then I started texturing. The marble machine itself was pretty straightforward, as well as the stairs. I started running into issues when I started to texture the marbles themselves. I thought it would be really fun to have my friends create some marble shaders that I could assign to the marbles randomly.

I made sure that the naming conventions and file types were the same, with only the number changing, so I could properly create a randomization string. I first ran into issues with Redshift, as when I assigned the material and used a string value override, it could not find a diffuse color parameter. I remembered from a previous project that I did that I needed to create my own diffuse color parameter and reference it in the texture file plugged into the RSMaterial node. This set up fixed my issue.

Normally when I randomize textures in this fashion, I use the primitive number. This worked when I tested it, as the base color of my texture maps were visible, but the actual patterns on them were missing. I quickly realized I probably needed to resize my UVs on the marbles. To do this, I needed to access the UV channel in my file (the first frame of my marble simulation in this case), so I did this by adding an unpack node and tweaking the UVs. When using a test material of a UV grid, the UVs were easily adjusted.

However, when I did this with the multiple textures, each individual primitive became accessible, so rather than the material node reading the whole sphere from the file (packed primitive) as one primitive, it was reading each face on each sphere as a primitive. This resulted in a grid-like look with each face having a randomized texture and it obviously was not the effect I wanted. I knew the previous file had a piece attribute from the RBD simulation, and that contained the information I needed, but I wasn't quite sure how to access it. After opening a separate test file and refreshing my knowledge on classes and packed primitives, I found a solution. After unpacking and adjusting the UVs, I add a connectivity node with the attribute type set to integer rather than string (this will allow the class name to have just a number rather than piece#), then I appended an assemble node after to pack the primitives created by the connectivity node. This essentially re-packed the spheres as their own separate entities as I had before (a simple pack node would pack all of the spheres together into one primitive, which gave me no access to the individual marbles for the texture variation). Then, I was able to use the randomization based on primitive number while having full access to the UVs for adjustment.

Obviously in these stills there are still several issues I need to address before I feel comfortable with moving forward (lighting mainly, UV stretching, minor tweaks with specular and subsurface materials) but all the materials are assigned in this scene, so with these minor things fixed I can finally start on the second section!

Class 9: Modeling Downstairs

Over the weekend, I decided to fully flesh out the kitchen elements that the Rube Goldberg machine relies on or interacts with. This involved mainly cabinetry and countertops (just something a bit more detailed than the proxy geometry from last quarter. I also finished the overall layout of the lower floor so that the windows that provide supplemental lighting as the sequence moves downstairs are set in place. This will aid in the coming week/weekend when I continue lighting iterations.

I obviously still need to put in a bit more detail in the background where the camera is, but I'm thinking (hoping) I can use cards with projected geometry and hide the fact that they are two-dimensional with depth of field.