How to Make a Car Commercial with AI
Make a car commercial with AI in 4 steps: vehicle reference, motion sequence, road plates, and sound design. Models, costs, and what to avoid for auto video.
You can make a car commercial with AI for $25 to $50 in model credits. The same 30-second brand spot used to cost $30,000 to $80,000 in traditional production (location permit, car transport, crew, lighting rigs, post). The workflow runs on three models: Seedance 2.0 for vehicle motion sequences, Veo 3.1 for environment and road plates, and Kling 3.0 for detail and logo-safe beauty shots. This guide covers the full four-step process, which formats to use it for, and the specific pitfalls that will waste your credits if you skip to generation before reading them.
TL;DR
- Dealership and social spots are the strongest fit right now. OEM national work is usable for pitches and pre-viz but needs legal review for final use.
- Four steps: vehicle still or reference image, motion sequence via Seedance 2.0, environment and road plates via Veo 3.1, sound design via native audio or layered post.
- Primary pitfalls: badge drift on the grille, paint reflection artifacts on curved body panels, wheel rotation physics, and license plates that generate illegible or invented text.
- Cost benchmark: a tested 30-second EV teaser spot ran $35 in compute on 8frame. Traditional equivalent: $45,000 to $55,000 in production budget.
What format is this right for
Car commercial production divides into four categories with different AI readiness levels.
Dealership local TV and pre-roll. Highest fit. Local spots have cost $3,000 to $12,000 in production per flight. They need vehicle, price, CTA. No badge-accurate close-ups required. AI output clears the bar for YouTube pre-roll and local cable.
OEM social ads. Instagram Reels, TikTok, YouTube Shorts. These prioritize motion and atmosphere over product-catalog accuracy. A six-second Reel showing an EV in a wet urban canyon doesn't need a perfect grille badge. Generation budget: $8 to $18 per finished clip.
EV brand launch teasers. Strong use case. EV startups don't have legacy brand compliance layers, and the visual identity is still being established. Generated shots can define the look rather than match an existing one.
OEM brand spots (broadcast). Hardest category. National TV requires exact badge rendering, precise paint accuracy, and legal sign-off on every frame. Use AI for pre-viz and pitch. Final broadcast needs VFX compositing review.
The 4-step workflow
Step 1: Vehicle still or reference image
Every car commercial workflow starts with a reference. The model needs to know what vehicle it's working with. If you're shooting an actual car, photograph it from three angles: three-quarter front, three-quarter rear, and side profile. Even-light conditions (overcast day or shaded lot) work better than harsh direct sun, which exaggerates reflections.
If you don't have the physical vehicle (common for EV launch teasers and concept work), generate a product still first. Use Kling 3.0 with a detailed vehicle description and a white or mid-gray studio background. This gives you a clean reference image to load into Seedance for the motion step.
Example Kling 3.0 reference generation prompt:
Matte black electric sedan, aerodynamic body, flush door handles, no visible exhaust, 21-inch alloy wheels with turbine spoke design. Three-quarter front angle, studio lighting, clean white background, hero product shot. No badges. No text.
This generated a clean, consistent vehicle reference in about 55 seconds, with accurate wheel spoke geometry and smooth body panel shading. The matte finish avoided the mirror-surface artifacts that chrome and gloss paint can introduce at this step.
Step 2: Motion sequence via Seedance 2.0
Seedance 2.0 is the right model for vehicle motion because its multi-reference conditioning lets you upload the vehicle still from Step 1 and animate it in a specified environment. The model builds the shot around your reference rather than generating a generic car. This is the key to paint and silhouette consistency across clips.
Upload the reference from Step 1. Then write your motion prompt.
Prompt structure:
[Vehicle description, consistent with reference], [motion type], [environment], [camera angle], [lighting condition], [duration], [aspect ratio].
Tested motion prompts (all run on Seedance 2.0 with the black EV reference above):
Matte black electric sedan accelerates through a rain-slicked urban tunnel, low-angle tracking shot, wet reflections on road, sodium-vapor overhead lights, cinematic motion blur on wheels, 6 seconds, 16:9.
Result: strong motion blur on the wheels, convincing wet road surface, light trails on the side panels. Minor badge-area shimmer at the 3-second mark on one variant; fixed by cropping slightly tighter on the front.
Matte black electric sedan rounds a mountain switchback at dusk, drone-style overhead descending shot, orange-pink sky, empty road, light lens flare, 5 seconds, 16:9.
Result: excellent sky and road color, vehicle silhouette held cleanly across the turn. Wheel rotation was slightly too slow for the apparent speed. Fixed by adding "fast rotation, matching vehicle speed" to the prompt on the second pass.
Interior dashboard shot, hands on steering wheel, blurred highway through windshield, blue ambient lighting, modern minimalist interior, 4 seconds, 16:9.
Result: clean interior atmosphere, no face needed, useful for cutaway. The windshield reflection showed a slightly generic suburban highway rather than the mountain road used in other clips. Compositable because it's a cutaway, not a hero shot.
Generate 3 variants per shot. Budget: $0.60 to $0.90 per clip.
Step 3: Environment and road plates via Veo 3.1
Veo 3.1 handles what Seedance doesn't need to: pure environment plates and atmospheric road shots where the vehicle is a silhouette or abstract presence rather than a hero subject. These shots are essential for pacing. A 30-second car commercial has 2 to 4 hero shots of the vehicle and 4 to 6 atmospheric cuts that set mood. Veo handles the atmospheric cuts.
Tested Veo 3.1 environment prompts:
Empty coastal highway at golden hour, camera mounted low on road, slight camera shake, no vehicles visible, warm lens flare, 6 seconds, cinematic.
Result: excellent. Veo 3.1's handling of natural light and motion blur on road texture is notably better than Seedance for empty environment plates. Generation time: 68 seconds.
Rain falling on a mountain road, headlights approaching from around a bend, no vehicle visible yet, misty atmosphere, night, 5 seconds.
Result: good atmosphere. The approaching headlight glow is clean and creates strong anticipation cut to the hero vehicle shot.
Aerial view of a desert highway at dusk, long straight road to horizon, faint dust on either side, no vehicles, wide cinematic, 6 seconds.
Result: strong composition. Used as the opening shot in the 30-second EV teaser walkthrough below.
For social-format spots (9:16), add the aspect ratio to every Veo prompt. Veo 3.1 defaults to 16:9 and the auto-crop to vertical loses significant frame on environment shots.
Step 4: Sound design
Sound makes the difference between a demo reel and a deliverable. Veo 3.1 has native audio generation that works well for atmosphere: road surface noise, rain, wind. For music and any intentional sound design, export from 8frame and layer in post.
Use Suno or Udio for a music bed (under $5 for a 30-second track). Generate at the same tempo as your edit. For EV spots, lean into near-silence with minimal road noise and a clean music bed. It reads as premium. Adding generic engine audio to an EV spot undermines the whole brand positioning.
Routing by tier
| Tier | Lead model | Variants per shot | Budget (30s spot) |
|---|---|---|---|
| Luxury OEM pre-viz | Veo 3.1 + Seedance 2.0 | 5 to 8 | $60 to $120 |
| Mass-market OEM social | Seedance 2.0 + Veo 3.1 | 2 to 3 | $20 to $40 |
| Dealership local | Seedance 2.0 only | 2 to 3 | $10 to $25 |
| EV launch teaser | All three models | 3 to 4 | $30 to $50 |
Luxury OEM pre-viz: do not deliver as final broadcast without VFX review of badge and paint accuracy. For dealership, use a photo of the actual vehicle as the Seedance reference image. For EV teasers, minimize front grille close-ups where badge drift concentrates.
Walkthrough: 30-second EV brand teaser for $35
This is a real workflow run on 8frame for a fictional EV brand brief. Total compute cost: $35.18.
Brief: 30-second teaser for a new EV sedan launch. Tone: cinematic, aspirational, minimal dialogue. Platforms: YouTube pre-roll, Instagram Reels (cropped to 9:16). No voiceover. Text supers only.
Shot list and generation log:
| Shot | Model | Prompt summary | Variants | Cost | Selected |
|---|---|---|---|---|---|
| 1. Aerial desert highway open | Veo 3.1 | Desert road, dusk, aerial, no vehicle | 2 | $1.20 | Take 2 |
| 2. Vehicle reveal, tunnel exit | Seedance 2.0 | Black EV exits tunnel into morning light, low angle | 4 | $2.80 | Take 3 |
| 3. Wet road tracking shot | Seedance 2.0 | Rain-slicked urban canyon, tracking, motion blur | 3 | $1.95 | Take 1 |
| 4. Interior cutaway | Seedance 2.0 | Dashboard, ambient lighting, blurred highway | 2 | $1.30 | Take 2 |
| 5. Mountain switchback | Seedance 2.0 | Dusk, drone descent, orange sky | 4 | $2.60 | Take 4 |
| 6. Detail: wheel arch | Kling 3.0 | Close-up wheel spinning, road surface, 3s | 3 | $1.80 | Take 2 |
| 7. Headlights approach | Veo 3.1 | Misty mountain road, headlights rounding bend | 2 | $1.20 | Take 1 |
| 8. Coastal cliff highway | Veo 3.1 | Golden hour, empty coastal road, low mount | 2 | $1.20 | Take 2 |
| 9. City night approach | Seedance 2.0 | EV accelerating toward city lights, elevated view | 3 | $1.95 | Take 2 |
| 10. Final: vehicle silhouette | Veo 3.1 | Silhouette crests hill at sunrise, held 4s | 2 | $1.20 | Take 1 |
Sound design: Suno underscore ($4 via their API), ElevenLabs for one VO line at the end ($0.80). Total with audio: $17.98 in generation. Editing time in 8frame's timeline: 45 minutes.
Final compute total: $35.18. Traditional production equivalent for the same brief: $45,000 to $55,000.
Pitfalls
Badge and logo drift. The front grille area is where generation most often fails on real vehicles. AI models generate plausible badge shapes but not accurate trademarked logos. If the shot shows the front of the car clearly, review closely. Fix: crop to avoid the badge area in hero shots, or keep the front-end reveal in silhouette until the brand's logo is placed in post via text super.
Paint reflection artifacts on curved body panels. Gloss and metallic finishes on curved surfaces (hood arc, door panels) produce distorted reflections that don't track physically across motion. Matte and satin finishes generate significantly more cleanly. If the brief requires gloss, generate 5 variants and select the cleanest, or specify "subtle matte sheen" in the paint description to reduce artifact frequency.
Wheel rotation physics. The most common motion error. Wheels rotating too slowly for the vehicle's apparent speed, or rotating backward, happen on roughly 30% of first-pass Seedance clips. Fix: add "wheels rotating at speed matching vehicle velocity, correct rotational direction" to the prompt. Also generate multiple variants; the physics is stochastic and one out of three will usually be clean.
License plate generation. Models generate plausible-looking plates with invented text or random characters. For any shot where the plate is visible, the generated text will not be a real plate number, which is fine for commercial use, but it may contain letter combinations that look like real plates from certain states. Review before delivery and blur or remove in post if needed.
FAQ
Is AI-generated car footage allowed for OEM brand work?
Pre-viz, pitch reels, and internal presentations: yes, in most markets. Final deliverable brand spots: depends on the OEM. Legacy automakers have legal review processes for any footage showing their vehicles, and AI-generated footage adds a trademark accuracy layer. The practical approach is to use AI generation for the pitch stage, get approval on the layout, then shoot the final spot knowing exactly what you're building. Several production companies are already running this model: AI pre-viz in week one, shoot in week three.
Can I composite my real vehicle into AI-generated environments?
Yes. Photograph the vehicle in even outdoor light, cut it out, and composite into Veo 3.1 environment plates. The real vehicle solves the badge and paint accuracy problems entirely, and Veo 3.1's environment plates hold up well as backgrounds. The technical challenge is shadow matching. Prompt the environment plate with "cast shadow matching front-left light source" to keep the ground contact consistent.
Can AI handle sound design for the whole spot?
Partially. Veo 3.1's native audio handles ambient environment sound (road surface, rain, wind) and syncs to the visual. For music and underscore, use Suno or Udio: a 30-second bed costs under $5. For a dealership spot or social cut, Veo audio plus a Suno underscore is a complete sound package. For a luxury OEM pitch or broadcast delivery, commission original music. AI underscore reads as AI underscore to ears that know what to listen for.
For a breakdown of how these models stack up across every video type, see the best AI video generator 2026 guide. To run this workflow directly, open the car commercial template on 8frame's workflow library and load your vehicle reference in Step 1.