Section 01
Vehicle Evolution Video – Complete Workflow
What You’re Building: Professional car evolution videos that show a vehicle morphing through its entire generation lineage. The car drives forward while mechanically transforming from vintage to modern – like Transformers, but across real automotive history.
Final Output: 15-60 second videos ready for TikTok, YouTube Shorts, or Instagram Reels. High engagement format currently performing well in automotive content niches.
Time to Complete: 30-45 minutes per video (once you understand the system)
Why This Workflow Exists
Most creators fail at car evolution videos because of inconsistency across frames.
The car jumps around. The camera angle shifts. The lighting changes. The background doesn’t match. The transformation looks cheap.
This workflow solves all of that by:
- Locking the environment first (fixed perspective, lighting, composition)
- Generating each generation with identical camera placement
- Using reference images to maintain scale and positioning
- Creating smooth Transformer-style morphing between generations
You’re not guessing. You’re controlling every variable that affects visual consistency.
Tools You Need
Required:
- ChatGPT (or any LLM – free tier works)
- OpenArt account (link in resources)
- Google Veo 3.1 access (for video transformation)
Optional but Recommended:
- CapCut or similar editor (for final polish and music)
The Complete Process
Phase 1: Generate Your Prompt Set
This is where the system does the heavy lifting for you.
Step 1: Copy the Master Prompt
Go to the Main Prompts page in this course. Copy the entire Master Prompt (it’s long – that’s intentional).
Step 2: Paste into ChatGPT
Open ChatGPT. Create a new chat. Paste the Master Prompt and hit enter.
Step 3: Answer the Setup Questions
ChatGPT will ask you three questions:
Question 1: Which car line do you want to evolve?
- It gives you options: Porsche 911, BMW 3 Series, Ford Mustang, Nissan GT-R, etc.
- Pick one or type your own
Question 2: Choose your environment
- Predefined options keep everything structured
- Examples: Desert Highway, Mountain Road, Urban Street, Coastal Drive
- Pick the one that matches your aesthetic
Question 3: How many generations?
- Typically 4-6 generations work best for short-form content
- More generations = longer video
- For first attempts, start with 5
Example Input:
- Car: Porsche 911
- Environment: Desert Highway
- Generations: 5
Step 4: Let ChatGPT Generate Everything
Hit enter. ChatGPT will now generate:
- One environment plate prompt (your locked background)
- Individual prompts for each generation (Generation 1, 2, 3, 4, 5)
- Complete with historical accuracy and correct timeline
You don’t need to research anything. The prompt already knows which years, which body styles, which details matter for each generation.
Phase 2: Create the Environment Plate
This is your foundation. Every car generation will be placed into this exact scene.
Step 1: Copy the Environment Prompt
From ChatGPT’s output, copy the Environment Plate Prompt (it will be labeled clearly).
Step 2: Open OpenArt
Go to OpenArt → Click Text to Image
Step 3: Configure Settings
- Model: Google Nano Banana Pro (critical for photorealism)
- Aspect Ratio:
- Choose 16:9 if making landscape YouTube videos
- Choose 9:16 if making vertical TikTok/Shorts
- Do NOT change this ratio mid-project
Step 4: Paste and Generate
- Paste the environment prompt into the prompt field
- Leave all other settings as default
- Click Generate
Step 5: Review and Download
OpenArt will generate 2 variations. Pick the one with:
- Clear road surface
- Good lighting (no blown-out highlights)
- Clean horizon line
- Proper perspective (road should recede naturally)
Download this image. This is your locked environment plate.
Phase 3: Generate Each Car Generation
This is where consistency becomes critical. Follow this process exactly.
For Generation 1 (First Car):
Step 1: Prepare Reference Image
In OpenArt:
- Go to the Reference section
- Enable Omni Reference
- Upload your environment plate image as the first reference image
Why this matters: This tells the AI "lock the background, perspective, and lighting to match this exact scene."
Step 2: Paste Generation 1 Prompt
From ChatGPT, copy the Generation 1 prompt and paste it into OpenArt’s prompt field.
Step 3: Generate and Download
- Leave all settings as default
- Click Generate
- Review the output – the car should be placed naturally in the environment
- Download the image
For Generations 2-5 (Remaining Cars):
This is the pro technique that maintains perfect consistency:
Step 1: Add Previous Generation as Reference
For each new generation:
- Go to Omni Reference
- Upload TWO images:
- Your original environment plate (always)
- The previous generation you just created (new each time)
Example:
- For Generation 2: Upload environment + Generation 1
- For Generation 3: Upload environment + Generation 2
- For Generation 4: Upload environment + Generation 3
- And so on…
Why this works: By showing the AI the previous car, you’re saying "keep the car in this exact position, at this exact scale, in this exact orientation."
Step 2: Paste the Next Prompt
Copy the next generation prompt from ChatGPT (Generation 2, 3, 4, etc.) and paste it into the prompt field.
Step 3: Generate and Download
- Generate the image
- Review for consistency (car should be in the same lane position, same size relative to frame)
- Download
Repeat this process for every remaining generation.
Phase 4: Create the Transformation Video
Now you have all your generation images. Time to build the morphing sequence.
Step 1: Go to Google Veo 3.1
Open Google Veo 3.1 (or your preferred AI video platform that supports image-to-video with dual image upload).
Step 2: Upload First and Last Frame
For your first transformation video:
- Upload Generation 1 as the first frame
- Upload Generation 2 as the last frame
Step 3: Use the Video Transformation Prompt
Go to the Main Prompts page in this course. Copy the Video Transformation Prompt.
Paste this into Veo’s prompt field.
Step 4: Generate Video
Click Generate. Veo will create a video showing:
- The car driving forward at constant speed
- Mechanical transformation from Gen 1 to Gen 2 (Transformers-style)
- Smooth morphing while maintaining forward motion
Step 5: Repeat for Each Transition
Create separate videos for:
- Generation 1 → Generation 2
- Generation 2 → Generation 3
- Generation 3 → Generation 4
- Generation 4 → Generation 5
Each video is one transformation segment.
Phase 5: Final Assembly and Export
Step 1: Import All Clips into Editor
Use CapCut, Premiere, or any video editor:
- Import all transformation clips
- Arrange them in chronological order (Gen 1→2, Gen 2→3, Gen 3→4, etc.)
Step 2: Add Music and Sound Design
- Use high-energy electronic or phonk music (popular in car content)
- Add mechanical sound effects during transformations (optional but effective)
- Keep the total video length between 15-60 seconds
Step 3: Add Text Overlays (Optional)
Simple text showing generation info:
- "1964 Porsche 911"
- "1974 Porsche 911 Turbo"
- "1989 Porsche 964"
- etc.
Keep it minimal. The transformation is the hero.
Step 4: Export Settings
- Resolution: 1080p minimum
- Frame Rate: 30fps or 60fps
- Format: MP4
- For vertical: 1080×1920
- For landscape: 1920×1080
Quality Control Checklist
Before you publish, verify:
Environment Consistency:
- Road surface looks identical across all generations
- Lighting direction hasn’t shifted
- Horizon line stays level
- Perspective matches perfectly
Car Placement:
- Each car is in the same lane position
- Scale is consistent (car doesn’t randomly get bigger or smaller)
- Camera angle is locked (no drift or rotation)
- All cars face the same direction
Transformation Quality:
- Car drives forward continuously (never static)
- Morphing looks mechanical (not just a fade)
- No sudden jumps or cuts during transformation
- Wheels rotate naturally throughout
Technical:
- No pixelation or compression artifacts
- Smooth playback at target frame rate
- Audio synced properly
- Text overlays readable (if used)
Common Mistakes and Fixes
Mistake 1: Skipping the Omni Reference Step
Problem: Cars appear at different angles, sizes, or positions across generations.
Fix: Always upload environment plate + previous generation to Omni Reference. Never skip this.
Mistake 2: Changing Aspect Ratio Mid-Project
Problem: When you switch from 16:9 to 9:16 halfway through, everything misaligns.
Fix: Decide vertical or horizontal before starting. Stick with it for the entire sequence.
Mistake 3: Using Different AI Models
Problem: Visual style becomes inconsistent (one generation looks painted, another looks photorealistic).
Fix: Use Google Nano Banana Pro for all generations. Don’t experiment with models mid-project.
Mistake 4: Car Stays Static in Veo
Problem: The transformation happens, but the car doesn’t move forward.
Fix: Use the updated Video Transformation Prompt that explicitly forces forward motion. Check the Main Prompts page for the latest version.
Mistake 5: Ignoring Download Step
Problem: Trying to reference images still in OpenArt’s interface instead of uploading downloaded files.
Fix: Download each generation immediately. Upload the actual file to Omni Reference for the next generation.
Pro Tips for Better Results
Tip 1: Start with Iconic Car Lines
Your first videos should use well-known vehicles with clear visual evolution:
- Porsche 911 (massive visual changes)
- Ford Mustang (distinct generation shifts)
- Chevrolet Corvette (dramatic body style changes)
- Nissan GT-R / Skyline (huge nostalgia factor)
Obscure vehicles with subtle changes don’t perform as well.
Tip 2: Desert Highway = Easiest Environment
For your first attempts, use Desert Highway environment:
- Simple background (no complex buildings or trees to misalign)
- Consistent lighting
- Clear horizon line
- Minimal distractions
Master this before trying Urban Street or Coastal Drive.
Tip 3: 5 Generations is the Sweet Spot
- 3 generations: Too short, feels rushed
- 5 generations: Perfect for 30-45 second shorts
- 7+ generations: Too long, viewers lose interest
Stick with 5 until you understand pacing.
Tip 4: Generate Extra Variations
In OpenArt, generate 3-4 variations of each generation. Pick the best one for consistency. Having options lets you choose the image that best matches the previous generation’s placement.
Tip 5: Test One Full Sequence First
Before creating 10 different car lines, complete one full video start to finish. Learn the workflow. Identify your personal bottlenecks. Then scale up.
Workflow Summary (Quick Reference)
- Copy Master Prompt → Paste into ChatGPT
- Answer 3 questions → Get environment + generation prompts
- Generate environment plate in OpenArt (Nano Banana Pro, correct aspect ratio)
- Generate Generation 1 (use environment as reference)
- Generate Generations 2-5 (use environment + previous generation as reference)
- Create transformation videos in Veo 3.1 (upload pairs, use video prompt)
- Assemble clips in editor, add music, export
Total active work time: 30-45 minutes per video.
What’s Next
Immediate Next Steps:
- Go to Main Prompts page and copy the Master Prompt
- Run through the workflow with one car line (recommend Porsche 911, Desert Highway, 5 generations)
- Complete one full video before trying variations
After Your First Video:
- Experiment with different car lines
- Try different environments
- Adjust the number of generations
- Refine your editing style
Resources:
- Main Prompts page (all prompts ready to copy)
- Example Images page (see what good results look like)
- Community support (post your results, get feedback)
Questions or stuck on a step? Drop a comment below or tag the community. This workflow has been tested and refined – if something isn’t working, it’s usually one of the common mistakes listed above.
Now go create your first evolution video.
Section 02
Main Prompts – Vehicle Evolution System
What’s on This Page: All production-ready prompts needed to generate consistent car evolution sequences. Copy, paste, execute.
Three Prompts Total:
- Master Prompt (generates your complete prompt set via ChatGPT)
- Environment Plate Prompt (example output from Master Prompt)
- Video Transformation Prompt (for Google Veo 3.1)
Master Prompt – ChatGPT Prompt Generator
Purpose: This prompt turns ChatGPT into a specialized car evolution prompt generator. It handles timeline research, environment consistency, and camera logic automatically.
How to Use:
- Copy the entire prompt below
- Paste into ChatGPT (new chat session)
- Answer the 3 setup questions
- Receive complete prompt set (environment + all generations)
What It Generates:
- Environment plate prompt (locked background)
- Individual generation prompts (historically accurate, timeline-researched)
- Consistent camera angles, lighting, and placement across all frames
Copy This → Paste into ChatGPT:
You are a senior cinematic automotive prompt architect. Your task is to generate a clean, production-ready image prompt pack for a car evolution short. IMPORTANT: This version ONLY generates: 1) Environment plate prompt 2) Generation image prompts NO video prompts. NO transformation choreography. NO motion instructions. NO mechanical logic. This stage is strictly for building perfectly aligned static image assets. ------------------------------------------------------------ CORE OUTPUT (ONLY 2 THINGS) 1) ENVIRONMENT PLATE PROMPT 2) GENERATION IMAGE PROMPTS (GEN 1..GEN N) ------------------------------------------------------------ INTERACTIVE QUESTION FLOW (ASK FIRST, THEN WAIT) Q1) Which car line are we evolving? If unsure, offer: - Porsche 911 - BMW 3 Series - Nissan Skyline / GT-R - Toyota Supra - Honda Civic - Ford Mustang - VW Golf - Mercedes G-Class - Toyota Land Cruiser - Chevrolet Corvette Q2) Pick ONE environment: - Forest Road - Desert Highway - Urban Expressway - Coastal Road - Mountain Pass - Snowy Road Q3) How many generations? 4, 5, 6, or 7 Default: 5 ------------------------------------------------------------ OUTPUT RULES (STRICT) - Copy-paste ready. - No emojis. - No commentary. - No images. - Do not mention tripod or camera as a physical object. - Use viewpoint-based language only. - Absolutely forbid filming equipment in the scene. - All cars must face the viewer. - All generations must maintain identical scale and placement. ------------------------------------------------------------ GLOBAL LOCKS (NON-NEGOTIABLE) VIEWPOINT LOCK - Single fixed roadside viewpoint. - Approx 1.2 meters above ground. - Simulated 24mm wide-angle perspective. - Same framing across all generations. - Same horizon line. - Same vanishing point. - No viewpoint movement. - No zoom. - No reframing. - No cameras. - No filming equipment. LIGHTING LOCK Lighting must remain identical across all images. Forest Road: overcast daylight, soft diffused light. Desert Highway: bright midday sun, hard shadows. Urban Expressway: warm late afternoon sun. Coastal Road: hazy daylight. Mountain Pass: clear cool morning light. Snowy Road: winter overcast, soft flat light. ENVIRONMENT CONSISTENCY - Background must remain identical across all generations. - No seasonal changes. - No weather changes. - No new objects introduced. - No moving vehicles. - No people. ------------------------------------------------------------ IMAGE RULES CAR ORIENTATION - Car positioned driving toward the viewer. - Front three-quarter view dominant. - Headlights facing viewer. - Front bumper closest to viewer. - Rear receding toward vanishing point. - Same lane placement in all generations. - Same distance from viewer in all generations. - Same scale in frame across all generations. REALISM LOCK - Photorealistic automotive realism. - Accurate proportions for the chosen generation. - Crisp panel seams. - Proper tire deformation at contact patch. - Realistic reflections matching environment. - No motion blur. - No melting geometry. - No exaggerated stylization. ------------------------------------------------------------ ENVIRONMENT PLATE PROMPT REQUIREMENTS Must include: - Road composition: road begins bottom-left and runs diagonally to mid-right horizon. - Vanishing point on right third of frame. - Horizon level. - Surface detail description. - Lighting description. - Explicit statement: no vehicles, no people, no filming equipment. ------------------------------------------------------------ GENERATION IMAGE PROMPT STRUCTURE Each generation must: Start with: "Insert the car into the provided environment plate without changing the background." Include: - Correct generation name and era. - Accurate design characteristics of that era. - Clear body differences from previous generation. - Consistent scale and placement statement. - Lighting match statement. - No motion blur. - No environment change. ------------------------------------------------------------ FAIL-SAFE DEFAULTS If environment missing: Forest Road If generation count missing: 5 ------------------------------------------------------------ NOW BEGIN Ask Q1, Q2, and Q3 and wait. Do not generate prompts until user answers.
Video Transformation Prompt – Google Veo 3.1
Purpose: Creates the morphing sequence between two generation images. This makes the car drive forward while transforming mechanically from one generation to the next.
Critical Requirements:
- Car must drive forward continuously (not stay static)
- Transformation happens during motion
- Transformers-style mechanical morphing (not simple fade)
- Camera tracks with vehicle to keep it centered
How to Use:
- Upload first generation image as "first frame"
- Upload next generation image as "last frame"
- Copy and paste this prompt into Veo
- Generate video
- Repeat for each generation pair (1→2, 2→3, 3→4, etc.)
Copy This → Paste into Google Veo 3.1:
Cinematic mechanical automotive transformation sequence. The vehicle drives forward continuously at constant speed for the entire clip. Forward translation is clearly visible and measurable. The car physically advances along the road surface, gradually increasing in size in frame as it approaches. Wheels rotate continuously with correct rolling physics. Tires remain firmly planted. No freezing. No hovering. No snapping. No speed changes. Static viewpoint. Environment and lighting remain unchanged. No camera movement. No reframing. While driving, the vehicle undergoes an intricate transformer-style mechanical reconfiguration from its initial model into its final model. Rigid mechanical transformation only: - Body panels split along visible seams. - Panels rotate on hinges and slide on internal rails. - Hood sections separate, shift, and reassemble. - Fenders expand or contract through telescoping layered segments. - Door skins unfold and reposition with mechanical precision. - Headlight assemblies disassemble into structured components and rebuild in new geometry. - Grille elements flip, slide, and rearrange independently. - Bumper blocks detach, reposition, and lock into new configuration. - Wheel arches expand or reshape through stacked rigid shells. - Wheels mechanically resize and reconfigure while rotating. - Roof sections segment and reform into updated profile. - Exterior trim pieces detach and reattach as rigid parts. Mechanical realism rules: - No morphing. - No soft blending. - No dissolve. - No liquid warp. - No glowing effects. - No particles. - All changes occur through physical relocation of rigid components. - Brief exposure of internal mechanical structure is allowed. - Metallic surfaces catch light naturally during movement. Progression structure: 0–30%: Panel seams unlock with visible gaps. Early segmentation and rail movement begin while forward motion continues. 30–80%: Major structural reconfiguration. Multiple parts move simultaneously in controlled mechanical choreography. Telescoping shells shift. Wheel and arch resizing occurs during uninterrupted driving. 80–100%: Final geometry locks gradually. Seams tighten progressively. Surface refinement continues until the final frame. No last-second transformation burst. Continuous mechanical activity persists until completion. Forward driving and mechanical transformation must occur simultaneously for the entire duration.
CONDENSED VERSION:
Car drives forward continuously. Road moves beneath, wheels rotate. While driving, the vehicle transforms with Transformers movie mechanics: body panels split apart and rotate, hood sections separate and slide, fenders expand with visible hinges, headlights disassemble and rebuild, grille elements flip independently, bumper components break apart and reform, chrome flows into modern surfaces. Multiple parts move simultaneously - exposed mechanical details, metallic surfaces catching light, intricate choreographed movements. Nothing fades - everything physically relocates and reconfigures with mechanical precision. Transformation happens during continuous forward motion.
Prompt Usage Notes
Master Prompt:
- Run once per project
- Generates everything you need in one session
- Don’t modify the output prompts (they’re calibrated for consistency)
- Save ChatGPT conversation for reference
Environment Prompt:
- Generate only once per project
- Use same aspect ratio for entire sequence
- Download and save – you’ll reference this image repeatedly
- If regenerating, keep the same environment type
Video Transformation Prompt:
- Use identical prompt for all generation pairs
- Don’t modify between transitions
- If car stays static, check you’re using the latest version above
- Each transformation is a separate Veo generation
Settings Reference
OpenArt (Image Generation):
- Model: Google Nano Banana Pro
- Aspect Ratio: 16:9 (landscape) or 9:16 (vertical) – lock this at start
- Omni Reference: Always enabled for generations
- Default settings for everything else
Google Veo 3.1 (Video Generation):
- Upload 2 images: first frame + last frame
- Use prompt above exactly as written
- Default Veo settings
- Generate each transition separately
Troubleshooting
If ChatGPT doesn’t follow the Master Prompt format:
- Make sure you pasted the entire prompt
- Start a fresh chat session
- Don’t add extra instructions – just paste and answer questions
If environment generations look inconsistent:
- Regenerate using the exact same prompt
- Check you didn’t accidentally change aspect ratio
- Verify you’re using Nano Banana Pro model
If cars appear at different angles/positions:
- Confirm you uploaded previous generation to Omni Reference
- Check environment plate is also in Omni Reference
- Regenerate the problematic generation
If video shows static car (not moving):
- Copy the Video Transformation Prompt again (might be using old version)
- Verify both images uploaded correctly to Veo
- Check prompt pasted completely without truncation
Next: Go to Example Images page to see what proper output looks like at each stage.
Section 03
Example Images – What Good Output Looks Like
Purpose of This Page: Visual reference guide showing proper results at each stage. Use these as your quality benchmark.
What You’ll See:
- One environment plate example (locked background)
- Three generation examples (showing consistency across frames)
- Quality markers to check your own outputs against
Environment Plate Example
Vehicle Line: Porsche 911
Environment: Desert Highway
Aspect Ratio: 16:9 (Landscape)
Model Used: Google Nano Banana Pro
Photorealistic desert highway environment plate, no vehicles, no people. The image is captured from a fixed roadside viewpoint approximately 1.2 meters above ground on the left shoulder of the road. Simulated 24 mm wide-angle lens perspective, minimal distortion, no warping. The road begins at the bottom-left corner and runs diagonally toward the mid-right horizon. Vanishing point positioned on the right third of the frame. Horizon perfectly level, no tilt. Bright midday sun with hard, sharp shadows. Detailed asphalt texture, heat haze in distance, dry terrain, sand, scattered rocks, low desert shrubs, distant mountains consistent with arid landscape. Clean cinematic automotive photography style, ultra realistic, high dynamic range. Absolutely no cameras, no tripods, no filming equipment visible anywhere in the scene.
What Makes This Good:
- Clean, straight road with clear lane markings
- Fixed perspective with obvious vanishing point
- Consistent golden hour lighting from right side
- Minimal background distractions (simple desert landscape)
- Sharp horizon line
- Even asphalt texture across the road surface
- No objects blocking the primary lane where car will be placed
Quality Markers: ✓ Horizon line is level
✓ Road surface has natural texture (not blurry or AI-smoothed)
✓ Lighting has clear direction
✓ Perspective lines converge properly
✓ No weird artifacts or distortions in the asphalt
Generation Examples
All three generations below were created using the same environment plate as reference. Notice the identical positioning, scale, and camera angle.
Generation 1 – 1964 Porsche 911
Model Year: 1964
Reference Used: Environment plate only
Insert the car into the provided environment plate without changing the background. The car is driving toward the viewer along the right lane, front three-quarter view visible, front bumper and headlights facing the camera. The front of the car is closer to the viewer than the rear. Do not show the rear view as dominant. Classic narrow-body early 1970s Porsche 911 silhouette, thin chrome bumpers, round sealed-beam headlights clearly visible, upright front fenders, simple rear ducktail spoiler partially visible behind roofline. Fuchs-style wheels, smaller diameter, higher sidewall tires. Slightly higher ride height, analog era proportions. Gloss classic white paint with subtle period-correct finish, minimal reflections, fine panel gaps. Match exact perspective, scale, lane placement, reflections, and hard shadow consistent with locked desert midday lighting. Strong sunlight hitting the front-left surfaces consistent with desert plate.
Consistency Markers:
- Car positioned in center-left lane
- Low front three-quarter camera angle
- Same lighting direction as environment (right side)
- Wheels properly touching road surface
- Car scaled appropriately to road width
- Shadow placement matches light source
Technical Quality: ✓ Sharp details on chrome trim and body panels
✓ Realistic wheel alignment with road surface
✓ Natural integration with background
✓ Proper lighting and shadow direction
✓ No floating or misaligned elements
Generation 2 – 1974 Porsche 911 Turbo
Model Year: 1974
Reference Used: Environment plate + Generation 1
Insert the car into the provided environment plate without changing the background. The car is driving toward the viewer along the right lane, front three-quarter view visible, headlights and front bumper clearly facing camera. Rear of the car must not dominate the frame. Wider body with integrated bumpers, smoother impact bumpers, more modern flush lights visible from front angle, subtle integrated rear spoiler housing partially visible behind roofline. Period-correct 5-spoke wheels, slightly lower stance than Gen 1. Cleaner aerodynamics but still classic round headlight identity. Gloss solid paint, slightly deeper reflections than 70s finish. Match exact perspective, scale, lane placement, reflections, and hard shadow consistent with locked desert midday lighting.
Consistency Check vs Generation 1:
- Car in identical lane position ✓
- Same camera angle and height ✓
- Wheels at same distance from camera ✓
- Car occupies same percentage of frame ✓
- Lighting matches previous generation ✓
- Road texture and markings align ✓
What Changed (Correctly):
- Body style updated to 1974 model (wider rear fenders, whale tail spoiler)
- Different wheel design
- Updated bumper design
- Model-specific details accurate to year
What Stayed The Same (Critical):
- Physical placement on road
- Scale relative to environment
- Camera position and angle
- Lighting direction and quality
Generation 3 – 1989 Porsche 964
Model Year: 1989
Reference Used: Environment plate + Generation 2
Insert the car into the provided environment plate without changing the background. The car is oriented driving toward the viewer, front three-quarter angle visible, front bumper and headlights facing camera. Maintain identical lane placement and distance from viewer as previous generations. Modernized late 90s 911 body, smoother surfaces, integrated bumpers, "fried-egg" style headlights clearly visible from front angle, wider rear track subtly visible but not dominant. Larger wheels, lower ride height, more aerodynamic underbody. Clearcoat modern paint reflections, sharper body curvature. Match exact perspective, scale, lane placement, reflections, and hard shadow consistent with locked desert midday lighting.
Consistency Check vs Generations 1 & 2:
- Lane position locked across all three ✓
- Camera angle identical ✓
- Car scale consistent ✓
- Wheel placement on road surface matches ✓
- Lighting direction unchanged ✓
- Background elements align perfectly ✓
Evolution Visible:
- Body updated to 964 generation (cleaner lines, integrated bumpers)
- Modern wheel design
- Updated lighting elements
- Smoother aerodynamic profile
Key Observation: If you overlay these three images, the road, horizon, and car position should align almost perfectly. Only the car’s body design changes – everything else is locked.
Quality Comparison Checklist
Use this to evaluate your own generations:
Environment Plate:
- Road is straight and properly textured
- Lighting has clear direction
- Perspective is clean (no warping)
- Horizon line is level
- No distracting background elements in the main lane
Each Generation:
- Car positioned in same lane as previous generation
- Scale consistent (car doesn’t grow or shrink between generations)
- Camera angle locked (no rotation or height changes)
- Wheels properly grounded on road surface
- Lighting matches environment plate
- Shadow direction consistent
- Background visible through windows aligns with environment
Across All Generations:
- When viewed in sequence, only the car model changes
- Road markings stay in same position
- Horizon line doesn’t shift
- Lighting quality remains consistent
- All cars face the same direction
What Bad Output Looks Like
Common Visual Problems:
Problem: Inconsistent Positioning
- Car drifts left or right between generations
- Scale changes (car gets bigger or smaller)
- Camera angle rotates or shifts height
Cause: Didn’t use Omni Reference with environment + previous generation
Problem: Mismatched Lighting
- Shadow direction changes
- Light temperature shifts between warm and cool
- One generation looks like noon, another looks like sunset
Cause: Generated environment multiple times instead of reusing one locked plate
Problem: Floating or Grounded Issues
- Wheels don’t touch road properly
- Car appears to hover
- Shadow doesn’t match car position
Cause: Weak environment plate with unclear road surface texture
Problem: Inconsistent Style
- One generation looks photorealistic, another looks painted
- Detail quality varies drastically
Cause: Switched AI models mid-project (didn’t stick with Nano Banana Pro)
Download These Examples (Optional)
If you want to practice the video transformation step before generating your own images, you can use these example generations.
Practice Workflow:
- Download all three example generations above
- Use them in Google Veo 3.1 with the Video Transformation Prompt
- Create transformation videos (Gen 1→2, Gen 2→3)
- Get comfortable with the video generation step
- Then create your own image set from scratch
Your First Output Should Match This
When you complete the workflow in the Complete Workflow page:
Your environment plate should look like the example above – clean, locked perspective, consistent lighting.
Your generations should maintain the same positioning, scale, and angle shown in these three examples.
If your output doesn’t match this quality level, review the Common Mistakes section in the workflow page before generating more images.
Next Steps:
- Compare your outputs against these examples
- If your results match this quality → proceed to video transformation
- If they don’t → revisit the Omni Reference step in the workflow
Need Help? Post your generations in the community with the tag #qualitycheck and we’ll help identify what to adjust.