Our recent testing of AI-generated video shows clear potential, alongside equally clear limitations when using this technology as a tool in video production.
Although there is a lot of online excitement and discussion that AI will revolutionise video and film, despite the hype, it still has some way to go. Cautiously then, our intention is augmentation, not replacement: AI as another tool, used carefully and creatively within a project if requested.
1. Early-Stage Visualisation
AI excels at: - Mood films and concept visuals - Early-stage product context (cars, machinery, environments) - Exploring multiple visual directions quickly. This means faster exploration of possible scenarios, and communicating concepts and ideas.
2. Filling Gaps that are Expensive to Film
AI can generate shots that would otherwise require: - Large crews - Specialist vehicles - Permits, weather windows or extreme locations. Used carefully, AI can reduce cost and risk without reducing ambition in a video project.
1. Complex details: A Known Failure Point
Testing shows that accurate products and tech are currently one of AI video’s weakest areas.
Specific issues include: - Incorrect geometry and proportions - Impossible internal motion - Misrepresentation of load, friction and rotation - Components morphing or behaving inconsistently between frames
These nuances are subtle but critical for a product video, as even small inaccuracies can undermine credibility.
At present, AI-generated product visuals cannot be relied upon without heavy intervention and should be treated as exploratory references only, not explanatory truth.
2. Technical Inaccuracy
AI does not understand engineering principles and often, laws of physics can go awry. Common issues include: - Incorrect component behaviour - Impossible mechanical motion - Inaccurate product context
Although AI is constantly improving, things can still appear a little odd...
3. Brand Control Issues
AI alone struggles to respect brand nuance.
Common problems include: - Inconsistent colour application - Generic or incorrect materials - Vehicles, products or environments that do not align with brand positioning - Random licence plates, symbols or markings
Policing brand integrity using pure AI is difficult and time-consuming. This is precisely where professional oversight becomes essential.
Without creative direction and post-production control, AI outputs tend to drift off-brand — often subtly, but consistently.
These details matter enormously in professional brand communication.
4. Visual Instability
Raw AI video commonly suffers from: - Flickering details - Morphing geometry - Shifting reflections and surfaces.
This is acceptable in tests, but not broadcast or client-ready.
5. The Illusion of “One-Click Video”
AI tools market speed and simplicity, but: - Outputs still need shaping - Errors still need fixing - Narrative still needs design
Without professional oversight, AI often creates more problems than it solves. In testing, we found that it took a number of attempts (and therefore credits) to get a usable outputs.
A Practical Example: AI + Professional Post-Production
In the supplied example: - The top image shows the raw AI-generated car shot - The bottom image shows the enhanced version after post-production
Professional intervention included: - Correcting and stabilising the car colour - Changing the number plate - Removing the driver - Refining realism and visual consistency
This demonstrates the real value proposition: > AI provides raw material; post-production skill turns it into a usable asset.
Step 1: Guided AI Exploration
Working with us, we can:
Define clear visual goals - Set technical and brand boundaries - Generate AI concepts under creative direction
This avoids random experimentation and keeps outputs aligned with brand values.
Step 2: Professional Curation & Selection
Not all AI outputs are usable. We: - Select viable shots - Discard risky or misleading material - Identify where AI works and where live-action or CGI is better. This will be where sign-off can be tricky, as requests to change even minor parts of an AI video via re-prompting can often lead to other undesirable changes.
Step 3: Post-Production Enhancement
AI footage is treated like raw plates: - Colour correction and brand matching - Retouching and clean-up - Compositing with real footage - Motion graphics overlays
This is where quality and credibility are locked in.
Step 4: Hybrid Final Films
The strongest results combine: - Live-action filming - Motion graphics and illustration - AI-generated elements - Traditional editing and storytelling
The viewer then never sees the seams — only a polished, confident brand message.
Benefits
• Faster concept development
• Reduced production costs in specific areas
• More visual ambition
• Greater flexibility across markets and sectors
Safeguards
• Brand accuracy maintained
• Credibility protected
• No reliance on uncontrolled AI outputs
This AI generated video...