top of page

GenAI videos

5 days ago
3 min read

Updated: 4 days ago

49ing

Context

Branding materials


Selected visual language

49ing is an AI-assisted ice hockey analytics platform that boosts young players' performance through data. The ice hockey robotic figure I generated was selected as their mascot to visually emphasize the 'AI inside' product's value proposition.



GenAI video process

Including real-life references to escape GenAI standards.

Google Flow

The tool's current strength is understanding intent


Initial prompts


"Create a 10-second clip of an ice hockey winning moment, starring an ice hockey robot team. Use the image to style the robot players."

Turns out 'winning moment' does not refer to the actual goal.


"A futuristic white mechanical robot plays ice hockey. The robot approaches the net, scores a goal, strikes a winning pose, and turns to face its cheering robot teammates. The setting is an ice rink with a dark, smoky atmosphere, illuminated by dramatic cool blue lighting and glowing ethereal nodes in the background."


The goalie disappears from the gate; I turn to real-world ice hockey scenes. See final prompt.

Final prompt

A futuristic white mechanical robot plays ice hockey. The robot approaches the net, scores a goal, strikes a winning pose, and turns to face its cheering robot teammates. The setting is an ice rink with a dark, smoky atmosphere, illuminated by dramatic cool blue lighting and glowing ethereal nodes in the background. Model the scene, camera angle, robot style, and atmosphere after the enclosed images.



Generating Variations


'MidJourney'

The tool's current strength is keeping style consistency


Initial prompt

A futuristic white mechanical robot plays ice hockey. The robot approaches the net, scores a goal, strikes a winning pose, and turns to face its cheering robot teammates. The setting is an ice rink with a dark, smoky atmosphere, illuminated by dramatic cool blue lighting and glowing ethereal nodes in the background. Model the scene, camera angle, robot style, and atmosphere after the enclosed images.


Combining tools

Midjourney video generation can be based on a start and end image. I returned to Google Flow to generate an end image. These precise camera movements were not possible in 2025.


Google Flow prompt: Use the enclosed image style to show how the robot scores a goal; use frame 4 as a reference.
Google Flow prompt: Use the enclosed image style to show how the robot scores a goal; use frame 4 as a reference.

Google Flow prompt: Turn the camera so the player scoring the goal is in the center, and the goalkeeper is on the left
Google Flow prompt: Turn the camera so the player scoring the goal is in the center, and the goalkeeper is on the left
Google Flow prompt: Turn the robot scoring the goal so they face the camera, a bit farther away.
Google Flow prompt: Turn the robot scoring the goal so they face the camera, a bit farther away.

Back to Midjourney

The start and end images yielded a morphing video. I opt for a new detailed prompt.



The deep understanding of what is feasible in an ice hockey context, which Google Flow achieves so easily, is beyond Midjourney's capacity. However, I have enough brand-aligned Midjourney-generated video clips to edit a final sequence aligned with both the visual style and ice hockey realities.

Gemini detailed prompt

"Cinematic video of a futuristic white-armored robot hockey player on a dark, foggy ice rink with cool blue lighting and glowing data network lines. The camera follows the robot as it guides a visible black puck toward the goal. A robotic goalkeeper dives to block the shot, but the puck enters the net. The scoring robot strikes a triumphant winning pose, then turns away from the camera toward approaching robot teammates who skate in and cheer."



YouTube low res with sound

 
 
bottom of page