SynnalAnalyse your own →
ErikDoesVFX

Did AI Just Kill VFX?

1,016 comments analysedPublished 5 Jun 2026Analysed 10 Aug 2026

What the video argues

VFX vs. AI — one Kong scene, two very different processes

Erik recreates a Kong: Skull Island sequence by hand: location filming, a 100-hour Houdini explosion, a friend's Skull Crawler animation, real water splashes, and a ragdoll simulation. The process takes months and fills nearly 2 TB of storage.

Running the same scene through AI video tools produces results that are visually inconsistent and hard to control. The video argues that the joy of making something from scratch is itself the point, and that AI cannot replicate it.

VFX recreationAI video comparisonpractical production
Derived from transcript · 17-min video

The channel

ErikDoesVFX
Genre
VFX filmmakingcreative processindie film
Subscribers424K
Total videos139
This video · views1.5M0.64× channel median
Likes44K29 per 1K views
Views per subscriber3.6×channel median 5.5×
Video age2 monthspublished 5 Jun 2026
Identified sponsor.store domains (domain registrar)identified from video · 02:30

Timeline

The video makes six moves: introduce the challenge, establish the VFX workflow, reveal the store launch, tackle the hardest shots, run the AI experiment, then compare results.

0:00–1:47
Claim

Recreating a famous Kong VFX sequence by hand is a meaningful test of whether AI video tools can replace skilled VFX work.

TopicsVFX recreation · AI comparison
71→
1:47–3:30
Example

Building the explosion from scratch in Houdini requires multiple simulation layers, weeks of work, and nearly 100 hours of render time.

TopicsHoudini simulation · explosion workflow
152→
3:30–6:15
Definition

Using 2D layered VFX assets during compositing is a standard production shortcut, and the creator launches a store to sell high-quality drag-and-drop assets.

TopicsVFX assets · compositing workflow
81→
6:15–8:26
Example

Animating the Skull Crawler requires rigging, 3D tracking, water simulation, and mixing CG with real filmed water splashes to achieve convincing results.

Topicscreature animation · water simulation
183→
8:26–13:33
Example

The final shot demands a ragdoll simulation, branch rotoscoping, and fire assets built from real bioethanol burns, totalling months of iterative work across multiple software tools.

Topicsragdoll simulation · compositing
203→
13:33–17:07
Reframe

AI tools cannot hold continuity across shots or respond to a specific creative vision, making them a poor substitute for intentional, craft-driven VFX work.

TopicsAI video generation · creative control
325→

What the room actually said

Reception across all 2,500 comments.

Vibe check

AI's shot-to-shot continuity failures are the defining complaint.→
The closing speech about craft and struggle is the most-quoted moment.→
The scale of production effort surprises even experienced viewers.→
Positive · 72%Friction · 16%Neutral · 12%

The room is enthusiastically pro-craft. AI's continuity failures are the dominant shared observation, and the final speech about the value of struggle landed harder than the VFX itself.

Engagement over time

Activity in the last week, hour by hour. Recent buckets are the most reliable — comments are sampled, weighted toward recent.

Hourly · grouped by day
4 Aug
5 Aug
6 Aug
7 Aug
8 Aug
9 Aug
Today
LessMore

Peak: 4 Aug, 17:00 · 2 comments

What's driving discussion

Comments cluster around five clear themes. AI criticism dominates; craft appreciation and future-of-AI anxiety are close behind.

Named topics · 88%Others · 12%
PositiveFrictionNeutral
AI continuity and control failures284· 32%→
Craft appreciation and process244· 27%→
AI trajectory and job anxiety183· 20%→
Technical critique of VFX shot122· 14%→
Humor and entertainment61· 7%→

Top comments

The three comments the audience pushed hardest, by likes and replies.

@kiwishamoo6494 · 4.7K likes · 181 replies→
I love the way AI has TERRIBLE continuity between shotsClick to open this thread →

Agreement spectrum

Audience agreement tracks the video's argument closely. The AI-is-soulless conclusion draws the most consensus; the fairness of the comparison draws the most friction.

AgreeNeutralDisagree
Building the explosion from scratch in Houdini requires multiple simulation layers, weeks of work, and nearly 100 hours of render time.152→
The final shot demands a ragdoll simulation, branch rotoscoping, and fire assets built from real bioethanol burns, totalling months of iterative work across multiple software tools.203→
Animating the Skull Crawler requires rigging, 3D tracking, water simulation, and mixing CG with real filmed water splashes to achieve convincing results.183→
Recreating a famous Kong VFX sequence by hand is a meaningful test of whether AI video tools can replace skilled VFX work.71→
Where it splits
AI tools cannot hold continuity across shots or respond to a specific creative vision, making them a poor substitute for intentional, craft-driven VFX work.325→

Alternate takes

A handful of comments push back not on the conclusion but on the framing, arguing the comparison is unfair by design or that AI is best understood as a tool inside the existing workflow.

AI continuity and control failuresAI trajectory and job anxiety

I don't think this is a fair or apples-to-apples comparison. You're comparing a highly polished, high-effort production created with extremely expensive tools against a low-effort AI experiment made with a moderately priced tool. Naturally, low-effort AI output will look like slop, just as low-effort VFX work would look like slop compared to a professionally produced film. A better benchmark would be something like this short film by Dan Keift: https://youtu.be/ZghLm9MXVIY That's the level AI is actually competing against. It was meticulously crafted, has a clear creative vision, an original concept, and strong direction. Even though AI was part of the process, the filmmaker's intent is visible throughout the final product. If the goal is to evaluate AI's capabilities honestly, then the comparison should be fair in terms of effort, craftsmanship, and creative direction on both sides. Comparing a carefully produced traditional project to a quick AI-generated test doesn't prove much beyond the fact that effort matters. And we already know what modern AI systems are capable of. Tools like Seedance 2.0 can produce results that would have required entire teams of CG artists and simulation specialists not long ago. The real question isn't whether AI can create impressive work.. it's whether creators are willing to put the same level of thought, direction, and effort into using it.

@itsrazvinotrizvi · 17 likes→
craft appreciation and processTechnical critique of VFX shot

9:30 Fun fact: Resolve's magic mask uses AI. This is the kind of tools I want to see when it comes to AI - tools that remove the absolutely mind numbing tedious work such as masking or UV unwrapping (for 3D artists)

@DaddyMouse · 16 likes→
AI trajectory and job anxiety

Reminds me of what a lot of people probably said in 1850: Cameras are awful, they threaten the work of portrait painters, and the photographs don't even look that good. And if you move in any way during the exposure, the photo is ruined. That doesn't happen if you get a real portrait painter to paint you. I don't think this photography thing is going anywhere, it's just a cheap gimmick to try and replace true artists.

@captainyossarian388 · 1 likes→