Week One
- Apr 16
- 4 min read
Posts arranged newest to oldest. Week of 4/13/26-4/19/26.
As a personal project, I decided to create a spec ad for a poppi soda flavor. I am responsible for all aspects.
4/15/26
Today, I began testing the flip fluid simulation that I will be using for the soda pour.
References for the soda pour:
I also took my own reference video pouring the soda in a glass similar to one I will be using:
One of the first issues I ran into when setting up the simulation was volume loss as the particles poured into the collision object.
To troubleshooted I modified:
Grid scale
Particles seperation
Substeps
None of these settings helped with the volume loss issue, so I began wondering if the issue was the model.
For my collision object, I used the model generated from MeshyAI and used both a surface collision and a volume collision, but this issue was still happening.
Upon modelling a simple glass shape with my own modeling methods, I was able to get the simulation working without volume loss
Since the MeshyAI glass was giving me issues with the simulation, I decided to instead use it as a guide and model my own glass
Here is my current glass model I will be using for the flip simulation:
I then began experimenting with separating out the foam particles from the other fluid particles. To do this, I used a pop wrangle node to seperate out particles that have a higher vorticity.
I then assigned particles in the foam group to have a lower density so that they would float to the top.
I then adjusted the life of the particles and ensured particles reaping was turned on so that these foam particles would die rather quickly.
Moving on to the look of the foam particles, I used an attribute wrangle to create a bounding box and control the size of the particles based on their space in the y axis.
I ran into an issue here as the attribute wrangle worked, but on a few random frames, the pscale of all the particles was jumped up highers, and then returning to their normal pscale a few frames later.
I checked the geometry spreadsheet before applying the attribute wrangle, and found this occuring from the simulation, not the wrangle.
Upon doing more tests, I discovered this was occuring when some of the liquid in the simulation was pouring out of the glass. As long as all the particles stayed together the pscale did not jump up.
Since I want all the liquid to stay in the glass for the aesthetic of this ad, I will be prioritizng ensuring the liquid stays within the glass during my next simulations.
4/14/26
Here is some more reference I've gathered specifically for the layout of the project. All images from Pinterest.com.
The scene will have:
The can
A jar of cherries sitting out
A fun glass where the soda will pour into
Here are specefic lighting references I've found for the project. All images from Pinterest.com.
I appreciation both the more dramatic back lighting and I also enjoy the background gradient. I plan to play around with both types of lighting to see what works when I have the scene set up.
With my idea in mind, I am now starting to begin gathering assets for the project. I want to experiment with using MeshyAI to create models for the project so I've found photos.
I experimented with feeding the tool an image of a fun glass that could be used to hold the soda.
I also wanted to see if the tool would be able to create a jar that will hold the cherries.
Lasly, I gave MeshyAI the soda can photo that was previously generated from Google Gemini, and tasked it with also creating the texture of the model.
I'm pleased with the glass and the jar model created but the can had a lof of faceting and issues with the texture.
I will be bringing the glass and jar model into Houdini to experiment with and test with simulations.
For the can, I will be outsourcing the model from CGTrader.
Ideation
Today, I am officially starting a concept I've been thinking about bringing to life for some time now. Almost every night I unwind at the end of the day with a fun beverage like a prebiotic soda. I am setting out now to create a poppi spec ad. I have begun doing research on the brand and using AI tools to help in creating ideation.
I particularly enjoy this look ad from poppi. Thinking about simulations, I love the look of the pour of the liquid and the splash of the liquid.
I also enjoy the whipped cream look. I think this could be a fun addition to the ad towards the end.
Thinking of flavors, I want to use a flavor that would give me the ability to do whipped cream if I want to in the end. I am also conscious of the colors and aesthetics of each flavor.
The brand olipop has a cherry vanilla flavor, but poppi does not. I prefer the look of the poppi cans so I will be creating a fake cherry vanilla poppi flavor for this spec ad.
Here is the mood board I've put together for the ad. All images from Pinterest.com
For my fake flavor, I looked at real poppi flavors to find one that most closely matched a cherry vanilla flavor.
The shirley temple poppi matched the best with the cherry aesthetic. I then used Google Gemini to come up with a mock design for the cherry vanilla ad.
I'm really happy with the look of this can and will be using this design to texture the can in my full CG scene.
To use this design, I asked Gemini to flatten the texture out so I could extract the elements with photoshop and create a proper texture for the UVs on my can model.
Thinking ahead, I also started to research some tutorials and resources I can reference for my mock advertisement.