SCAD X Harbor Picture week 8

SCAD X HARBOR PICTURE

2/27/20253 min read

February 25, 2025

Feedback / Notes

Mentor Hailey

Shot 2: The water droplet doesn't feel like it’s landing on the petal properly.
Shot 3: The droplet spread feels unnatural.
Shots 4 & 5: The front flower should be blurred—render the flower as a foreground layer.

Mentor Luke

Motion blur can help improve.
Use scope to analyze the reference image for accuracy.
Shot 4:
- Apply motion blur to the water for better integration.
- Micro-faceting in interior fluids should prevent discoloration of the label.
Shot 5:
- Render s/t map.
- Check the depth pass.

Mentor Billy

Refer more to reference images for accuracy.
Add more textural details to the flower to enhance realism.

Mentor Kyle

Black point looks dirty—adjust for better contrast.
White point appears gray—increase brightness.
Push the color more for vibrancy.
Shot 2:
- The sky feels too dark—adjust for better balance.
- The flower’s color range needs improvement.
Shot 3:
- Push the sky brightness higher while ensuring the label doesn’t turn gray.
Shot 5:
- Increase the color range in the flower for more variation.
- Ensure the square and logo align properly.
- Blur the sun and try an exposure glow effect for realism.

Adjustments

Apply motion blur.
Continue iterating on the water droplet behavior for improved realism.

Shot 4 - Water Droplets

This week, my goal is to get the water droplet to look right. However, after discussing it with Professor Fowler, we realized that since the angle of the bottle is almost horizontal, the water droplet just splashes instead of maintaining its shape and moving along the bottle. My plan is to adjust the bottle to be almost vertical so that the droplet can retain its shape while moving.

Motion Blur in Arnold

February 27, 2025

vertical bottle

bottle with old angle

I started by using the original source and collision setup but opted for a FLIP solver in DOP instead of Vellum Fluid. To control the velocity within a specific frame range, I implemented a POP Speed Limit and keyframed the surface tension to refine the droplet’s shape. Since the simulation is based on real-world scale, the water droplet is extremely small. If the surface tension exceeds 0.001, the droplet becomes unstable and explodes. To maintain stability, I keep the value between 0 and 0.001. At 0, the droplet appears flat, whereas at 0.001, it retains its form effectively.

Since I want to maintain the rounded shape at the front of the water droplet while also preserving the water residue, relying solely on friction is challenging. Using friction alone tends to flatten the droplet’s shape. To address this, I plan to simulate them separately and later merge them using VDB to achieve the desired effect.

My approach is to first simulate the FLIP and assign it a new attribute called mask with a value of 1. I then apply the same attribute to the bottle but set its value to 0. Using Attribute Transfer, I can blend these values to achieve the desired interaction between the droplet and the surface.

After that, I connect the bottle and the simulation to inputs 1 and 2 of the solver. I use Attribute Transfer to blend values between them and create a float attribute called mask_prev to store the point values from the mask attribute. By applying the max function, I retrieve the highest value and return it. Additionally, I use 1 - chf('decay') to control the overall decay of the mask attribute.

After that, I use Scatter based on the mask attribute to generate points on the surface. However, the result initially appears too flat, so I apply an Attribute Wrangle to adjust the distribution and add more variation. I then merge the points with the original simulation and use VDB to create a rounded shape before converting it back to a polygon for the final output.