Make a Cinematic Ad End-to-End: Stage 1: Building the Assets

2026-08-10 · Academy lesson · Adil · recording, 8:01 · original

Topic: Assets, references & consistency · mode: Tutorial · tags: course chapter, series, one creator's experience

In the first stage of the course Make a Cinematic Ad End-to-End, Adil builds every asset for the ad inside a Collab project before any video is made. He makes a product sheet of the can in GPT Image 2, a character sheet of himself from 20 photos and an empty New York street in Soul Cinema, then tests the character and street together in Seedance 2.0 before locking them. Two robots, a futuristic ball and a glowing goal follow, generated in Soul Cinema from prompts written by Claude.

Part of another source

This recording is almost entirely part of I Made a Cinematic Ad Using Claude Fable 5 + Higgsfield AI (Full Workflow) (2026-06-15); what is said in it is covered on that page.

What is said

• 00:00:00 — The can's product sheet in GPT Image 2 inside a Collab project
• 00:01:21 — Adil's own likeness: character sheet in Soul Cinema from 20 photos
• 00:02:30 — The location: an empty New York street in Soul Cinema
• 00:03:20 — Testing character and location together in Seedance 2.0 before locking them
• 00:04:30 — The hero robot and the rival robot from Claude prompts in Soul Cinema
• 00:06:50 — Props: the futuristic ball and the glowing goal

Threads in this video

• restate · How do you keep a location, product or prop the same across shots? — Adil makes a product sheet of the can (front, back and top shots) in GPT Image 2, so that the video model knows the can from all angles and does not hallucinate it. (00:00:00)
• refine · How do you keep a location, product or prop the same across shots? — Adil says props matter as much as characters in long videos, since they too have to stay consistent across all scenes and shots. (00:06:34)
• restate · Which image model should be used for keyframes, edits, text and outfits? — Adil calls GPT Image 2 the best model right now for image editing. (00:00:00)
• refine · Which image model should be used for keyframes, edits, text and outfits? — Adil calls the location the most important image: a sloppy or plasticky one ruins the videos whatever the prompt, since the video takes textures and lighting from it; Soul Cinema gives movie-still images by default. (00:02:11)
• restate · How do you keep a character the same across shots? — For a character with his own likeness Adil calls Soul Cinema hands down the best model, keeping all facial features consistent: he selects Character and drops in 20 images of himself, the more the better. (00:00:46)
• restate · What does an AI clip, shot or film cost? — Adil says eight Soul Cinema images cost only one credit, so five credits buy many outfits to choose from: the character sheet is a step to experiment a lot. (00:01:46)
• restate · How do you avoid the AI (plastic) look? — Adil calls the location the most important image: a sloppy or plasticky one ruins the videos whatever the prompt, since the video takes textures and lighting from it; Soul Cinema gives movie-still images by default. (00:02:11)
• restate · How do you get a cinematic look in a generated image or shot? — In the Soul Cinema location prompt ("cinematic anamorphic lens, shallow depth of field, film grain"), the anamorphic lens gives the wide cinematic feel and film grain removes the too-clean digital look. (00:02:53)
• restate · How do you draft and lock a shot before rendering the final? — Before locking key assets Adil tests them together, calling this stage critical: he attaches the character and the location to Claude and asks for a test-shot prompt with dynamic camera movement and a creative angle such as a Dutch angle. (00:03:20)
• restate · Which LLM should be used to write video prompts? — Adil has Claude write the asset prompts because it writes detailed, structured prompts for him, which saves time and credits by avoiding bad generations. (00:04:17)

Mentioned

Central: Claude · GPT Image 2 · Soul Cinema

Also: Works: TOPUP football ad

Concepts: AI look · character sheet · dutch angle · product sheet

Places: New York

Products: Collab · Enhancer · Seedance 2.0

Only in the description or credits: Higgsfield Academy · Make a Cinematic Ad End-to-End

← Previous: Make a Cinematic Ad End-to-End: Intro: One Website, One Cinematic Ad · Next → Build an AI Ad Agency with Claude + Higgsfield: Pressure-test the economics