In April 2025, a two-person marketing team from one of our early-access teams, at a growing direct-to-consumer home goods brand, created a 45-second product launch video from brief to first-cut review in 87 minutes. Their materials were a written brief, a 150-word script already prepared for email copy, and three product photos from their asset library. They did not make a conventional storyboard. Instead, they used a scene-by-scene breakdown serving the same function.
This is a detailed look at how the session ran, which preparation enabled it, and when the 90-minute goal is realistic or unrealistic. The purpose is to show what "fast production" means in practice, as a workflow you could test this week rather than a marketing claim.
Preparation Before Starting
The 90-minute timer began when they opened Higgsfield and uploaded the first reference frame. Important preparation came before that, so this was not truly a "zero preparation" workflow.
Before beginning, they had an approved brief specifying visual tone, target formats, a product page hero video and a 9:16 social cut, along with a target duration for each. They also had a script drafted for another purpose that could be adapted to video pacing. Three existing high-resolution product photos showed the product clearly from different angles under consistent lighting.
Preparation likely covered two to three hours over two days, with brief writing in the morning and script adaptation in the afternoon. Starting both tasks from scratch means adding that time. The 90 minutes describes generation, not the entire project. For teams with a brief template and reference frame library from prior sessions, preparation per campaign becomes significantly shorter over time.
Phase One: Brief and Script Alignment (25 Minutes)
After opening the tool, they used the first 25 minutes for no generation. Instead, they converted the brief and script into scene-level inputs for the model.
The 150-word email script contained three parts: a product introduction sentence, a two-sentence use case description, and a call to action. That structure became three video clip segments. They described each as a scene, covering the product's appearance, camera or subject motion, and visual tone. They submitted scene specifications rather than the script as a text prompt.
The script to scene-specification conversion is likely the workflow's most important step. Teams that skip it and use the script word for word as a video prompt often get something more like a slideshow than a video. The model follows scene intent instead of narrative voice.
The three-segment script produced five scene specifications: two introduction scenes, an establishing wide shot and close-up detail shot, two use case scenes, and one CTA scene. The result was five clips, each paired with a scene specification and reference frame.
Phase Two: Assigning Reference Frames (10 Minutes)
Once the five scenes were defined, they distributed their three product reference frames among them. The establishing shot and CTA used one wide-angle reference. The two close-up scenes used a detail reference, while the use case scenes used a lifestyle reference showing the product in context.
Choosing reference frames is the main way to keep clips consistent. To make all five clips appear to come from one shoot, connect as many as possible to the same reference. When different angles or contexts are intentional, assign a separate reference to each and keep the lighting specification consistent for color stability.
This stage took about 10 minutes, mainly to choose the stronger reference for the lifestyle scenes. They needed no new photography.
Phase Three: Generate and Review (40 Minutes)
The five 1080p clips spent approximately 18 to 22 minutes in the queue during generation. While waiting, they prepared review notes covering scenes likely to need a second generation, motion details to inspect, and whether color remained stable enough across clips for assembly.
The first output produced three clips they approved immediately and two requiring a second generation with revised scene specifications. A use case scene moved too quickly for the script's pacing. In a close-up detail shot, camera movement partly hid the product in the clip's second half. Both issues could be corrected through scene specifications rather than reference frame changes.
Generating the two revised scenes again took 10 minutes in the queue. They reviewed and approved both, then proceeded to assembly.
Assemble and Review First Cut
Outside Higgsfield, they placed the five clips in sequence with a standard video editing tool. Assembly took about five minutes. After exporting the first cut, they added a scratch voiceover by reading the original email script aloud on their phone, then sent it for internal review.
First upload to first-cut export took 87 minutes.
Internal review took two rounds, followed by one additional generation. After viewing the cut, the team lead requested a slower camera move for the CTA clip. The final approved cut was ready the following morning. Brief to approved first cut took about five hours over two days.
When 90 Minutes Is Not Possible
The session reached 90 minutes because these conditions aligned: a specific brief, a script adapted into scene-level material beforehand, high-quality reference frames already on hand, and a team with at least two earlier generation sessions and an understanding of how scene specifications affect output.
A team's first session should take two to three hours. The extra time does not mean the tool is slow. Scene translation takes longer while learning how inputs affect outputs. Reference selection takes longer without tested materials to choose from. Review also takes longer before the team knows which generation artifacts deserve adjustment and which are acceptable for the use case.
The 90-minute goal becomes realistic by the third or fourth session if the team builds a brief template and small reference frame library during the first couple of sessions. Teams that begin each session anew without reusing earlier inputs will remain at two to three hours.
The Purpose of This Workflow
This workflow is not presented as a replacement for a full production session for a brand campaign hero film. A 30-second brand spot intended for television broadcast has different quality standards, review procedures, and stakeholder participation from a product video for a DTC landing page and social ads.
The 90-minute workflow is designed for the marketing formats that represent most of a small team's production volume: product demos, explainer videos, social ad creative, product page hero video, and campaign variations. These formats have traditionally taken days from brief to first cut, not because the work itself demands days, but because production coordination does.
Bringing that cycle into one session expands what a content calendar can accommodate. A team reaching first cut in 90 minutes can test two or three campaign directions instead of choosing one. This is not merely a small efficiency improvement. It changes how the team works.
