CATEGORY: MUSIC VIDEO
CLIENT: FOUNDATION ENTERTAINMENT
PROJECT: J’RIVA - FRANK OCEAN
POSITION: PRODUCER / DIRECTOR
PROJECT CREDITS
Director/Producer: Martyn Watts (BRONKO)
Artist: J'Riva
Label: Foundation Entertainment
Director of Photography: Jacob Caron
Steadicam Operator: Ilya Gittelson
AI Workflow Design: Martyn Watts
Colorist: Martyn Watts
Transforming the ordinary into the extraordinary: How hybrid AI workflows turn a shoestring budget and 4-hour shoot into cinematic storytelling
PROJECT OVERVIEW
When emerging artist J'Riva approached us for his debut music video "Frank Ocean," the challenge was clear: deliver a visually stunning, metaphor-rich narrative on an independent budget with an aggressive 4-week timeline. The creative vision—J'Riva surrounded by swarms of bees, ultimately ascending into the sky—demanded VFX-level execution that would typically require animal wranglers, extensive green screen work, and weeks of compositing.
Instead, we developed a hybrid AI workflow that strategically integrated AI-generated enhancements into traditionally shot footage. Filming at a rock sanctuary in Simi Valley with a skeleton crew, we captured 20 setups in just 4 hours before losing daylight. In post-production, we used Google Veo 3.1, Nano Banana, Runway ML, and Kling Omni to generate photorealistic bees crawling across J'Riva's tattooed skin, create swarming sequences, and build a fully AI-generated ascension finale—proving that AI doesn't need to replace traditional production, but can amplify it to achieve what budgets and timelines wouldn't otherwise allow.
THE CHALLENGE
Resilience as Visual Metaphor on an Independent Budget
J'Riva's "Frank Ocean" explores resilience through surrealist metaphor: a man surrounded by swarms of bees, enduring their stings while maintaining unwavering composure, ultimately ascending skyward—transformed, not defeated, by adversity.
The creative treatment demanded:
Photorealistic bees progressively swarming J'Riva throughout the video
Intimate close-ups of bees crawling across his tattooed skin
Wide shots showing hundreds of bees encircling him as he performs
A climactic final shot: J'Riva rising into the sky, carried by the swarm
The constraints:
Shoestring budget: First video for an emerging artist on Foundation Entertainment—no room for traditional VFX studio costs
Compressed timeline: 4 weeks from brief to delivery
Production window: Shooting two music videos in one day; this was the second, with only 4 hours before losing sunlight
Safety/logistics: Live bees on set = insurance nightmares, animal wrangling costs, unpredictable behavior, talent safety concerns
Traditional VFX approach would require:
Green screen shoot with extensive rotoscoping
Manual animation of bee swarms (40-60 hours of compositing)
Motion tracking each bee onto J'Riva's skin (100+ hours)
Budget estimate: $25K-$35K in post-production alone
Our solution: Shoot real footage on location with natural light and a skeleton crew, then use hybrid AI workflows to seamlessly integrate photorealistic bees in post—treating AI as a strategic VFX tool rather than a full production replacement.
THE SOLUTION
Hybrid AI Integration: Real Footage + AI Enhancement
Rather than building everything in AI or relying on studio-controlled environments, we developed a hybrid methodology that strategically deployed AI where it delivered maximum impact with minimum friction.
PRODUCTION TIMELINE
Week 1: Treatment development, location scout, tech scout (sun path mapping), AI styleframe iteration
Week 1 (Day 7): Production shoot (4 hours, 20 setups, Simi Valley rock sanctuary)
Week 2-4: Post-production (AI enhancement, sky replacement, compositing, color grading)
Step 1: Treatment Development & Pre-Production (Week 1)
Creative Vision:
J'Riva's journey unfolds as a meditation on resilience:
Opens with J'Riva alone atop a rocky hill overlooking the city at dusk
A single bee appears, then more arrive progressively
Bees land on his arms, shoulders, neck—he continues performing, undeterred
Final payoff: Completely enveloped by bees, J'Riva levitates skyward, ascending above the city
Color Palette: Muted tones, warm amber hues, surreal turquoise skies—evoking introspection and transformation
Pre-Production Planning:
This was an outdoor location shoot—requiring:
Location scout: Identified rock sanctuary in Simi Valley with city overlook
Tech scout: Mapped sun path to maximize golden hour lighting within 4-hour window
Shot list prioritization: 20 setups sequenced by lighting progression (we were racing the sunset)
Crew logistics: Skeleton crew (DP Jacob Caron, Steadicam operator Ilya Gittelson, minimal support)
AI Styleframe Process:
Generated styleframes showing bee placement, swarm density, and J'Riva's ascension—similar iteration count to previous projects (1,200+ variations) but streamlined by lessons learned.
Step 2: Production - Racing the Sun (Week 1, Day 7)
Location: Rock sanctuary, Simi Valley, CA
Shoot Window: 4 hours before sunset
Setups Completed: 20
Camera: Sony FX6
Crew: DP (Jacob Caron), Steadicam (Ilya Gittelson), Director/Producer (Martyn Watts)
On-Set Workflow:
No bees on set—J'Riva performed against empty air, reacting to imagined swarms
Shot with natural light, no green screen, no LED walls
Prioritized wide establishing shots first (longest shadow angles), then moved to close-ups as light changed
Steadicam movement added cinematic fluidity to performance shots
Challenge: Gray, overcast sky on shoot day—not the surreal turquoise/amber palette we'd storyboarded
Key advantage of shooting traditionally first:
J'Riva's authentic performance energy
Natural lighting interactions on skin and tattoos
Real camera movement (handheld, Steadicam) added organic feel
Faster shoot (4 hours vs. full day in studio)
Step 3: Post-Production - Hybrid AI Enhancement (Week 2-4)
Timeline: 3 weeks post-production
Approach: Run traditional footage through AI tools to add elements impossible/prohibitive to capture practically
AI Enhancement Workflow:
A. Bee Animation on Skin
Captured J'Riva's performance footage with clean tattoos visible
Fed footage into Nano Banana (AI enhancement tool) to generate initial bee placement
Refined animation using Google Veo 3.1 for realistic crawling movement, wing flutter, and organic behavior
AI recognized tattoo patterns and kept bees tracked to skin topology frame-by-frame
B. Swarm Compositing
Generated wide shots of bee swarms using Runway ML
Composited swarms around J'Riva using traditional VFX techniques
Blended AI-generated swarms with practical footage using Magic Keyer (10% of shots required AI-assisted rotoscoping)
C. Sky Replacement - Novel AI Solution
Challenge: Shoot day delivered gray, flat skies—not the surreal turquoise/amber aesthetic we'd designed
Traditional approach: DaVinci Resolve's sky replacement tool (tested, results unsatisfactory)
AI-Native Solution:
Imported footage into Kling Omni
Used object removal feature to isolate J'Riva from shot, separating background
Replaced sky using custom sky map image (1080p source)
Uprezed footage back to 4K using Topaz AI
Reinserted J'Riva on top of new sky-replaced background plate using DaVinci Resolve Fusion
Applied color correction, film grain, and tilt-shift lens effect for selective focus
Why this worked: Kling Omni's object-removal AI was more sophisticated than Resolve's masking—cleaner separation, better edge work
Step 4: Fully AI-Generated Finale
Challenge: Final shot required J'Riva to ascend into the sky, levitating above the city—impossible to capture practically without expensive wire rigs or helicopter shots
AI Solution:
Created Kling Omni Elements 3D model of J'Riva using 4-5 reference angles from the shoot day
This ensured tattoo consistency across the AI-generated ascension sequence (critical artist requirement)
Generated fully AI shot: J'Riva surrounded by a swarm, rising into the twilight sky above city lights
Final composite approximated real footage aesthetic (lighting, grain, color palette matched)
Why Elements mattered: Without the 3D character mapping, J'Riva's distinctive tattoos would have shifted/morphed across frames—breaking immersion. Elements locked his visual identity.
STANDOUT AI SHOTS
Shot 1: Bees Crawling Over Tattoos
Narrative Context: As J'Riva sits on the rock performing, bees begin landing on his arms and shoulders, crawling across his intricate tattoo work.
Challenge:
Live bees = safety/insurance nightmare
Training bees to crawl specific paths across skin = impossible
Traditional VFX: Frame-by-frame manual animation = 60+ hours
Hybrid AI Approach:
Shot J'Riva's clean performance with no bees
Fed footage into Nano Banana to generate initial bee placement and pathing
Refined animation with Google Veo 3.1 for realistic wing flutter, leg movement, and antennae twitches
AI tracked bees to skin topology automatically, respecting tattoo contours and body movement
Why It Worked:
3-hour turnaround per shot (vs. 8-10 hours manual animation)
Photorealistic—indistinguishable from live insects
J'Riva's authentic performance energy is preserved
Shot 2: Swarm Encircles J'Riva
Narrative Context: As the song builds, hundreds of bees swarm around J'Riva, creating a hypnotic, choreographed cloud of insects.
Challenge:
Hundreds of individual bees with independent flight paths
Depth, perspective, lighting, and interaction with the environment
Traditional particle simulation = expensive rendering time
Hybrid AI Approach:
Generated AI swarm elements using Runway ML at multiple depths (foreground, midground, background)
Composited onto real footage using traditional VFX layering
AI handled naturalistic flight patterns (avoiding robotic/uniform movement)
Added motion blur and depth-of-field in post to integrate with Steadicam footage
Why It Worked:
AI-generated organic, unpredictable swarm behavior faster than manual particle systems
Real camera movement (Steadicam) added authenticity
Hybrid approach: AI for generation, traditional compositing for integration
Shot 3: J'Riva Ascends Into the Sky
Narrative Context: Finale shot—J'Riva, completely enveloped by bees, levitates above the rocky hilltop as the swarm carries him skyward over the city lights.
Challenge:
Impossible to shoot practically without wire rigs ($15K+ rigging, helicopter aerials)
Needed to maintain J'Riva's tattoo consistency across movement
Required seamless blend with real footage aesthetic
Fully AI-Generated Solution:
Created a Kling Omni Elements 3D model of J'Riva using reference stills from the shoot
Generated ascension sequence with Elements ensuring tattoo/facial consistency
AI rendered twilight sky, city lights backdrop, swarm envelope
Matched color grading, grain, and lens characteristics to real footage
Why It Worked:
Elements 3D mapping = tattoo consistency (J'Riva's specific requirement met)
Final shot approximates real footage so closely that viewers assume it's wire-work VFX
$500 in API costs vs. $20K+ practical aerial shoot
RESULTS & IMPACT
Production Metrics:
Timeline: 4 weeks (concept to delivery)
Shoot Time: 4 hours (20 setups)
Budget: 300% cost reduction vs. traditional VFX approach
Post-Production Efficiency: AI-enhanced workflows completed in 3 weeks vs. 8-10 weeks manual VFX
Client Feedback:
"Martyn captured the essence of the song perfectly and executed the video exactly as described in the treatment. The bees looked completely real—I was shocked when he told me there were no actual insects on set. The way he transformed what could have been a simple performance video into this surreal, metaphor-rich story shows true creative vision. For a debut video on our budget, this is beyond what we imagined possible."
— J'Riva, Foundation Entertainment
Creative Impact:
Proved AI integration scales to shoestring budgets without compromising cinematic quality
Demonstrated outdoor location shoots + AI enhancement = viable alternative to studio-controlled approaches
Showcased speed advantage: 4-week turnaround enabled timely release aligned with single drop
TECHNICAL SPECIFICATIONS
Camera: Sony FX6
Location: Rock sanctuary, Simi Valley, CA (outdoor, natural light)
AI Enhancement Tools: Google Veo 3.1, Nano Banana, Runway ML, Kling Omni (Elements, object removal), Higgsfield
Post-Production: DaVinci Resolve (editing, Fusion compositing, color grading), Magic Keyer (AI-assisted rotoscoping)
Upscaling: Topaz Video AI
Effects: Tilt-shift lens effect for selective focus, film grain overlay
Delivery Format: 4K, Rec.709
