What happens
when attention
comes apart?

I refined the film story, improved the concept images, generated detailed 3D models, and began assembling an outdoor scene in Unreal Engine. The source log does not give a completion percentage.

• Refined the film story and its narrative sequence.
• Improved the concept images with Dreamina, then used the revised images as references for high-detail models in Tripo 3D.
• Acquired desert and rock asset packs, shaped terrain and applied textures in Unreal Engine, and tested a preliminary assembly using the monster and candy house models.

1. Complete the candy house interior design and build; consider using UE MCP.
2. Design and build the pipe models; consider using online assets.

It is unclear whether the interior and exterior scenes should share one Unreal Engine project. If they do, I need to determine whether the current candy house model can support an interior scene.

Animating the muscle movement inside the underground monster’s mouth is technically difficult.

I could build the interior and exterior atmospheres as separate Unreal Engine projects and connect them through a button or film editing.

After placing the models, I will test whether AI tools can help animate the mouth. If that does not work, I will consider rigging or omit the animation.

I completed a local browser prototype for a scene in Devour: liking content raises syrup inside the room. I also connected a finger-heart gesture to the like action and tested the interaction on my computer. The prototype has not yet been integrated into the final artwork or published as part of this website.

• Chose Codex, p5.js, and ml5.js HandPose to explore gesture-based interaction. In a p5.js study, I tracked the thumb and index finger and tested a finger-heart gesture.
• Built a separate browser scene with a computer, a like button, and rising syrup. Each like raises the syrup by 10%; ten likes lead to an ending screen. I added mute and restart controls.
• Added a low, slow syrup-flow sound that grows louder with each like while keeping the like notification volume unchanged.
• Connected webcam gesture recognition to the scene's existing like action. I corrected a model-loading error and a case where the camera showed video but reported “No hand.” I then confirmed that the interaction worked locally.

1. Test gesture recognition under different camera angles, lighting conditions, and hand distances; adjust thresholds if needed.
2. Decide how the interactive scene will connect to the final project's visual style and narrative.
3. Record a demonstration video of no more than two minutes and complete the short tool-experience report required for Week 07.

Codex initially increased the like-notification volume instead of the syrup-flow sound; I clarified which sound should change. The first gesture integration also hit a detectStart is not a function error and then failed to detect a hand despite showing camera video. Those issues were corrected, but recognition can still depend on hand position and camera angle. The prototype remains a local study rather than a finished project scene.

I will test the open-hand-to-finger-heart sequence under varied conditions, tune the rules, and keep mouse controls as a fallback. The prototype and troubleshooting notes will inform the final scene and the tool-experience report.

Concept design is in progress. I completed base coloring for the project map and line art with base colors for three key elements: the monster, the trees, and the candy house. The source log does not give a completion percentage.

• Added base colors to the project map.
• Produced line art and base-color studies for the monster, mushroom-like trees, and candy house.
• Designed the candy house with a sticky, distorted form to create a horror atmosphere.
• Combined tree trunks with mushroom forms in the tree concepts.

1. Continue refining the concept designs.
2. Test generating 3D models with Tripo3D.
3. Explore an interior design for the candy house.

The concepts do not yet feel clearly connected or memorable. They need stronger visual symbols that link them to fragmented attention in the short-form video era.

I plan to create a mood board to define a consistent visual style and shared elements, gather more references, and refine how the concepts connect to the project's central idea.