Poster 01LIGHT

Poster 02CIVILIZATION

Poster 03ALCHEMY

LIGHT selected. Open the film using its title or the open-film link.

Selected / AEIROpen film

Scroll

Dmytro Uhnivenko / Kyiv, Ukraine

AI director. Creative director. AI artist.

Brand films, commercials and visual worlds—from concept to final edit. Based in Ukraine, working with teams worldwide.

Meet your creative partner

02 / Method

A film is not assembled.It is uncovered.

Four decisions turn an abstract brief into something with weather, weight and a pulse.

01The pulse

Find the pressure.

Before the image, find the human tension inside the brief.

02The world

Give it physical laws.

Light, weather, scale and camera behavior become one coherent world.

03The direction

Direct the unseen.

AI is a production department, not an aesthetic. Every choice still needs intent.

04The cut

Leave an afterimage.

Edit until the explanation disappears and only the feeling remains.

03 / Practice

One idea. One visual system. One finished film.

As an AI artist and designer, I combine filmmaking, graphic design, motion, 3D and generative AI production—from concept and visual development to final edit.

01
AI filmmakingContinuity from shot to sequence.
02
Concept & storytellingFind the tension before the frame.
03
Graphic & visual systemsOne language across every frame.
04
Motion designRhythm, timing and behavior.
05
3D & game artSpace, material and scale.
06
Edit & post-productionAttention, emotion and final form.

Production system

The film chooses the pipeline.

For every idea, I build a focused workflow—from research and visual development to edit, sound and final delivery.

Toolkit / 36 tools

07 production fields

Interlude / Director’s maxim

The frame is only evidence. Direction happened before it.

— DU

04 / Signals

Signals worth keeping.

A filmmaker’s field journal for generative cinema: production shifts, films worth studying and open calls worth entering. No daily noise.

Open all signals

A silver sailing vessel with a petrol-blue sail beneath oversized ODYSSEY lettering. AI-generated editorial illustration, not a frame from the film.

Films / Analysis26 Oct 2026

Musk promised an AI Odyssey. The first episode has arrived.

A seven-minute episode, a July promise and a revealing earlier experiment. What the Grok Imagine Odyssey actually shows about making films with AI.

Read article

Tools / Analysis25 Oct 2026

CapCut is coming to ChatGPT. The first edit starts with a conversation.

An idea in ChatGPT, a draft video, then CapCut’s web editor. The announced integration changes where editing begins. Its November beta will have to show how well the handoff works.

Read article

Tools / Analysis24 Oct 2026

Kimi K3.1 Pro: the real test starts after the demo

An icy harbour, a working 3D scene and an unusually honest comparison with Claude. Kimi’s new release makes a bigger claim than a better answer: work you can keep developing. Here is what the evidence supports.

Read article

Two dark sculptural audio tracks overlap beneath large DESSA typography, with a lime interruption passing between them. Editorial illustration of simultaneous listening and speaking.

Tools / Analysis24 Oct 2026

Dessa learned to talk by listening. The interesting part is when it interrupts

Block says its new voice model learned from human conversation alone. Behind the expressive demos are three separate claims: speech-only training, listening while speaking, and local inference. Here is what each actually means.

Read article

An AI-generated editorial portrait of Jensen Huang beside AFTER DOOM lettering and a pale green sheet with an illustrative rising line. The line is a metaphor, not financial data; this is not an event photograph.

Field note23 Oct 2026

Jensen Huang rejects AI doom. Optimism still has questions to answer

Jensen Huang has criticised predictions of AI catastrophe. His argument opens a more useful question: how can confidence in the technology coexist with evidence, safeguards and uncertainty about its consequences?

Read article

A human hand receives floating silver documents wrapped in a cobalt ribbon beneath WORK lettering. AI-generated editorial metaphor for delegated work and human responsibility.

Field note23 Oct 2026

Uber puts AI on 30,000 desks. Who owns the result?

Uber is rolling out ChatGPT Work to more than 30,000 employees after a 1,000-person pilot. The announcement marks a large deployment; evidence of its effect on productivity and jobs is still to come.

Read article

A stylised Optimus robot holds a red inspection tag against large OPTIMUS lettering and a faint perspective grid. AI-generated editorial illustration based on a Tesla reference, not a factory photograph.

Tools / Analysis23 Oct 2026

Optimus reaches the production line. When does it become a worker?

Tesla reports an early production ramp for Optimus in Fremont. What this milestone tells us about the robot — and why it does not yet tell us when humanoids will become ordinary workers.

Read article

A sculptural silver letter A and cyan paper plane stand over printed shapes beneath large CANVA CRAFT lettering. AI-generated editorial illustration, not Canva Craft output.

Tools / Analysis23 Oct 2026

Beautiful is not enough. Canva Craft wants AI to understand design.

Canva’s new image model plans a composition before rendering it. The ambition reaches beyond attractive pictures: hierarchy, readable text and the ability to keep working on the result.

Read article

A floating silver cinematic frame and terracotta audio waveform beneath large CHIMERA lettering on pearl grey. A silhouette stands at a sunlit doorway inside the frame. Conceptual editorial illustration.

Tools / Analysis22 Oct 2026

Chimera gives your character a voice. Who directs the performance?

Chimera combines Eleven v4 speech with Seedance 2.5 video. The appeal is a shorter path from a character image and a line of dialogue to a speaking performance; the limits are those of a short generated clip.

Read article

The Claude Duet icon, with its terracotta circle and black looping line, floats beside silver message cards and a small approval gate. Editorial illustration of proactive assistance and human choice.

Tools / Analysis22 Oct 2026

Claude Duet writes first. Who decides what needs you?

Claude Duet can bring work to you before you ask. Anthropic’s preview raises a practical question: how much initiative is useful when software can read connected apps and act on your behalf?

Read article

Silver pages open toward an amber orbit and sphere beneath AGI? lettering. A conceptual illustration of the distance between accumulated knowledge and discovery.

Field note22 Oct 2026

Two or three years to AGI? The harder question is how we will recognise it

Demis Hassabis has shortened his AGI forecast. His more revealing question is whether AI can make a discovery it was never taught—and what evidence would convince us that it had.

Read article

A silver portrait frame unfolds into a room as a performer steps beyond it; a lavender bottle anchors the scene. Conceptual editorial illustration of HeyGen Video.

Tools / Analysis22 Oct 2026

The avatar has left the frame. HeyGen wants to make the whole scene

A face, a product, a room—and everything that happens between them. HeyGen Video expands the assignment from a speaking presenter to a complete shot. That makes continuity, sound and physical action more important, not less.

Read article

Two floating silver rooms share a doorway; a small amber key appears in the first and is missing from the second. An editorial metaphor for story continuity across a cut.

Tools / Analysis21 Oct 2026

One prompt, ten minutes of film? VideoGen wants to take responsibility for the cut

Model 1 promises a whole production from a brief. The interesting question is what survives from one shot to the next—and what happens when the director changes their mind.

Read article

Two silver and cobalt sculptural forms face one another across a bright interval. Separate ELEVEN REALTIME typography frames an editorial metaphor for taking turns in conversation.

Tools / Analysis21 Oct 2026

Eleven Realtime: a beautiful voice still needs to know when to stop

ElevenLabs is moving from generating speech to managing a conversation. Mini and Max combine listening, response and voice; speed is only one part of making that exchange work.

Read article

Large black FLUX 3 MAX lettering behind a floating silver frame showing a Western cowboy. A yellow-backed film ribbon reveals an alternate take with an astronaut, on pearl grey.

Tools / Analysis20 Oct 2026

FLUX 3 Max: what faster video generation changes

Ten seconds of video in roughly twelve seconds of generation. Inside the BFL × fal launch: a surreal Western, native sound, ten keyframes, the real meaning of “tied for #1”, and what faster iteration buys a director.

Read article

A floating silver windmill world with jade walking figures and an orbit illustrates a persistent scene, character and motion. Original editorial illustration, not JST-2 output.

Tools / Analysis20 Oct 2026

Viggle JST-2: a world that outlasts the first frame

A scene in about a second is the hook. Returning to the same world from another angle is the bigger story. Inside JST-2: the launch film, developer evaluations of characters and motion, Gaussian splats, API prices, and the limits behind “physics-native”.

Read article

A monochrome worker and trolley remain continuous across a split between a practical gray stage and a finished mainframe room. Original editorial reconstruction in a floating silver film plate.

Films / Analysis19 Oct 2026

Real actors. A rebuilt world. How Greptile’s AI-finished film was made.

A gray stage becomes a vintage computer room. Osmo’s breakdown reveals the real actors, Beeble mattes, replacement environments and film-aging pass behind Greptile’s launch film—and the decisions that make the combination work.

Read article

A single terracotta ownership tile floats between a silver space rocket and a child’s school chair. Editorial metaphor for Musk’s proposed SpaceX share gift.

Field note19 Oct 2026

One SpaceX share for every child. Is this how universal high income begins?

Musk’s promise now has a company, a recipient group and a year: SpaceX, children in America, 2027. We check the $10.5 billion estimate—and the distance between owning a share and receiving an income.

Read article

A terracotta ownership share hovers between a silver AI processor and an empty silver envelope. Editorial metaphor for the unresolved transfer from AI wealth to personal income.

Field note19 Oct 2026

Universal high income: who gets paid when AI does the work?

Musk proposes giving shares directly to the public. Anthropic announces an equity-backed public benefit fund. Between those ideas and a reliable income lies the real story: ownership, cash and who makes the rules.

Read article

An amber bridge connects a white game-level blockout to a detailed silver architectural miniature. Editorial metaphor for previously unaffordable games, not an Epic product render.

Field note19 Oct 2026

Tim Sweeney says the games we couldn’t afford to make are coming.

A 5.2-million-view prediction, Epic’s AI-assisted Launcher Preview and experimental tools inside Unreal. The interesting question is which previously impossible games can now be finished—and what still stands between a prototype and a release.

Read article

A floating glass room holds a burgundy chair and its reflection. Original editorial metaphor for access and documentary trust.

Films / Analysis19 Oct 2026

Nearly three hours with Elizabeth Holmes. How A24 filled 2,000 screenings.

You Can See Everything brings Nathan Fielder, Lance Oppenheim and Elizabeth Holmes into a very unusual room. Behind its $15 million opening and 99% Rotten Tomatoes score is a filmmaking question: does seeing more mean understanding more?

Read article

Three floating silver frames follow one jade-green basketball across successive moments; the central ball projects forward. Conceptual editorial illustration of video understanding, not output from Glimpse.

Tools / Analysis17 Oct 2026

Glimpse watched the video again. The “cut” was a camera move.

Glimpse revisits a video before answering. Its launch examples pose a deceptively difficult question: did the subject move, did the camera move, or did the shot change?

Read article

A silver film strip becomes a burgundy-lined abstract award loop, in front of $1M lettering. Editorial illustration, not the official trophy.

Films / Analysis16 Oct 2026

A $1 million AI film prize. The real brief is in the rules.

A $1 million prize, a five-to-seven-minute film and a 70% Seedance requirement. The AI Film Award 2027’s rules define a more specific challenge than its headline suggests.

Read article

Silver film frames and a raised vertical crop with a cobalt-blue back illustrate selecting and reframing footage.

Tools / Analysis16 Oct 2026

Meta put AI to work in DaVinci Resolve. How much of the edit can you trust?

Meta demonstrates its open-weight model working in DaVinci Resolve. The example connects media understanding to software actions; the hardware figures show what local use demands.

Read article

A monochrome door photograph floats above an ochre layer that peels into a sound waveform: changing the soundtrack while keeping the picture. Editorial AI illustration.

Tools / Analysis16 Oct 2026

The shot works. The sound doesn’t. Meet Mirelo SFX 2.0.

Keep the picture you finally got right. Mirelo’s new sound model promises to replace unwanted music and rebuild the scene around its dialogue. The useful question is what survives that second pass.

Read article

A monumental silver numeral 4 floats beside large GEMINI and ARGON typography, with one cobalt-blue ring passing through it, on a pearl-grey background. Conceptual editorial AI illustration, not an Argon output.

Tools / Analysis14 Oct 2026

Gemini 4 Argon is out. The best demo has two camera angles.

One room, two camera angles—and a model that can help build the scene behind them. Google’s public Argon rollout brings ambitious creative workflows, a shared million-token budget and benchmarks that tell a more complicated story than “best AI”.

Read article

A silver paper plane flies through a sculptural frame opening into a miniature room, following one teal route. WORLD MODEL typography sits behind it. Conceptual editorial illustration, not a ByteDance interface or output.

Tools / Analysis13 Oct 2026

ByteDance’s world model: what if you could step inside the shot?

A paper plane crosses an AI-generated room. A spoon appears on command. Then the demo turns to cinema. ByteDance’s world-model showcase raises a better question than whether AI can make a prettier video: can you direct what happens next?

Read article

A flat silver kitchen reference unfolds into a floating three-dimensional miniature with one cobalt-blue pot, against MUSE WONDER typography. Conceptual editorial illustration, not a screenshot of Meta’s software.

Tools / Analysis13 Oct 2026

Meta Muse Wonder: the AI demo you can actually play

Meta’s new AI turns image references into a playable kitchen in its launch demo. Behind the charming result: a coding agent, a million-token context and an important distinction between building a world and understanding it.

Read article

Floating silver versions of an invented scene converge on one burgundy-accented selected frame in front of THE ARC lettering. A conceptual illustration of selection and iteration, not footage from the series.

Films / Analysis12 Oct 2026

The Arc: a beautiful AI shot still has to survive the edit

A finished AI series makes a useful production question concrete: how much work sits between a promising generation and a shot you can keep? Credits, continuity and the cost of another attempt.

Read article

An original silver clapperboard contains a fictional empty-chair and telephone scene. A lifted glass layer marked AI represents a lighting adjustment. Conceptual editorial illustration; no actor likeness or film still.

Films / Analysis11 Oct 2026

Ben Affleck’s Animals hit No.1 on Netflix in the U.S. Where was the AI?

A chart-topping thriller and an AI admission. What Affleck has actually disclosed, what the ranking measures, and why the most interesting effect may be the one you never notice.

Read article

Silver SUBMIT and SAMPLE nodes connect to one video tile showing a cup being lifted. Two measurement brackets end at that same output: 10.8 s and 17.3–17.5 s. Conceptual editorial AI illustration, not a ComfyStreamerH3 test result.

Tools / Analysis11 Oct 2026

Real-time AI video has more than one clock

A fifteen-second clip, two different timings. ComfyStreamerH3 shows why the speed of generation and the wait for a completed video need separate measurements.

Read article

Tools / Analysis10 Oct 2026

ElevenLabs Brand Kit: giving AI the same brand brief

Brand Kit keeps a brand’s visual references and rules together, then makes them available across ElevenCreative tools. The change is in the brief: less rebuilding the same context for every request.

Read article

A silver film frame with an engraved contour imprint floats over teal SYNTHID typography; a narrow teal beam reveals the imprint

Tools / Analysis10 Oct 2026

Google opens SynthID Detector. What does a watermark prove?

Google is widening access to its watermark detector. A positive result can identify a supported AI signal; an absent signal cannot certify that a file was made entirely by a human.

Read article

A floating silver control dial emerges from an opal glass plane in front of giant cobalt-blue GPT-6 typography, with a cool blue accent

Tools / Analysis9 Oct 2026

GPT-6 Intelligent UI: the answer is becoming an interface

OpenAI is adding controls to ChatGPT answers. The useful question is what readers can change—and which assumptions the interface makes visible.

Read article

Three luminous silver planes float above terracotta MOTION lettering on a pale pearl-grey background; one small terracotta square lifts from the front plane

Tools / Analysis9 Oct 2026

Claude Motion turns an explanation into editable animation

Anthropic’s new animation tool works with code, text, charts and shapes. Its appeal lies in what can still be changed after the first version.

Read article

A brushed-silver banana floats below a frosted diffuser above golden-ochre NANO BANANA lettering, with a small 2.1 and a pale pearl-grey background

Tools / Analysis7 Oct 2026

Nano Banana 2.1: ‘Just change the light.’ Is AI finally listening?

The image works. You ask for one small change—and risk losing the face, the product or the detail you liked. Google’s new image model promises tighter control. Its first published comparisons show both the progress and the catch.

Read article

Editorial line drawing of Andrew Garfield as Sam Altman and Yura Borisov as Ilya Sutskever in conversation above a thin floating table, with separate ARTIFICIAL typography and an OpenAI signature

Films / Analysis6 Oct 2026

Artificial at NYFF: OpenAI’s story becomes a screen drama

Andrew Garfield plays Sam Altman; Yura Borisov plays Ilya Sutskever. After Artificial’s world premiere, a director’s look at the human conflict behind the AI story, the first reactions and what the official record can establish.

Read article

Conceptual silver portrait emerging from a floating metal video frame, with large GRIFFIN lettering on a milk-white and periwinkle gradient

Tools / Analysis5 Oct 2026

26 of 54 people mistook an AI for a human on a video call. What Griffin showed

A one-minute call. Then the reveal: the person on screen was AI. Inside Tavus’s 54-person study, its Griffin demonstrations and the timing gaps in NVIDIA’s benchmark.

Read article

Conceptual illustration of two silver stunt performers at a library window, with floating glass fragments, warm lemon backlight and separate KLING 4.0 typography

Tools / Analysis3 Oct 2026

Kling 4.0: AI now has to sustain a scene, not just a beautiful frame

Two character references, one location image and a 30-second fight. Johnson Sheng’s official hands-on makes continuity the interesting question: can the action stay readable when the camera, bodies and props all move?

Read article

Conceptual illustration of a floating silver window with a botanical scene continuing beyond its frame, warm yellow backlight and separate CONTINUUM typography

Tools / Analysis2 Oct 2026

Runway Continuum: what happens when the frame becomes an interface

Runway’s Continuum demonstration turns generated images into responsive interfaces. It is a research preview, with access and reliability still open questions.

Read article

Two matching cinematic frames pass through a floating satin-silver decoder gate on a blue Gemini gradient

Tools / Analysis1 Oct 2026

Gemini 4 Argon: same frames, faster decoder

Google’s new model has an unusually concrete connection to filmmaking: improving the software that plays video. The decoder example, the mixed comparison with Claude and OpenAI, and who can actually use Argon now.

Read article

A floating pearl-white conversation bubble holds three cinematic storyboard frames, with the centre frame lifted forward, against large white EDITS lettering and an Instagram gradient

Tools / Analysis1 Oct 2026

Edits assistant brings account analysis into video planning

Instagram’s assistant connects account insights with concepts and scripts. Its launch leaves a useful distinction: an explanation of the numbers is not yet proof that a recommendation works.

Read article

OpenAI’s yellow Alfred dot with black spectacles and bow tie floats beside a silver workspace panel, in front of warm gold dots lettering

Tools / Analysis30 Sep 2026

OpenAI’s dots keep working after you leave

Always-on agents, a more affordable model for complex work, and a shared home for people and AI. The important DevDay announcements, the access limits, and the comparison with Claude that actually matters.

Read article

Silver physical reset device on a dark stand, in front of large warm copper RESET lettering

Field note29 Sep 2026

Codex’s reset button is real. Here’s the story behind it.

A silver panel, an OpenAI logo and a handle you pull. Vaibhav Sisinty’s new clip brings Tibo’s unusual desktop device back into view. The original posts explain where the story began.

Read article

A doorway sketched on curled paper becomes a real illuminated door, with a small figure approaching the light

Tools / Analysis29 Sep 2026

Claude Sonnet 5.5: from a tiny emoji to a 3D print

A printable model, a forest written in code and a working digital sketchbook. The early Sonnet 5.5 experiments worth watching, with their creators credited and the results put in context.

Read article

A silver miniature farm and three levitating stones with cobalt typography illustrate a gravity-reversal scene

Field note28 Sep 2026

Seedance 2.5: a farm falls upward. The creator shared the prompt.

A tiny moon, an office outbreak and a ten-second spear duel. Five September posts show what creators are asking Seedance 2.5 to do—and three include the actual prompts.

Read article

A silver miniature city with deliberately duplicated Empire State towers; one highlighted in terracotta

Field note27 Sep 2026

2.5 million views. Two Empire State Buildings.

A viral AI reconstruction of 1993 New York shows how period atmosphere can conceal errors in the city itself.

Read article

A silver stage with one chair lifted independently while another chair and a table stay in place

Tools / Analysis27 Sep 2026

Autodesk Flow Studio: what remains editable after AI makes the image?

Autodesk combines 3D scene control with AI rendering. Its documented demonstration offers a concrete way to examine what stays editable.

Read article

A human hand chooses an unopened-letter storyboard from alternative images

Field note27 Sep 2026

After the demo, someone still has to direct.

Editable Blender scenes, an interactive lens experiment and video tools inside the timeline: what September’s published examples actually show.

Read article

Three manuscript leaves show a local revision while the repeated final story frame stays unchanged

Tools / Analysis27 Sep 2026

Claude Opus 5.5: what changes beyond the first answer

Anthropic’s cost claim, API changes and Ryan Sael’s interactive lens project. What the release documents—and what a selected demo cannot prove.

Read article

One creative brief branches toward an editable room, modular blocks and repeatable caption cards

Tools / Analysis27 Sep 2026

GPT‑6: what separates Astra, Sol and Luna

Three model tiers, different API rates and a published Blender project. A guide to the distinctions behind the GPT‑6 name.

Read article

A teal scarf remains consistent across three consecutive views of the same woman and street

Tools / Analysis27 Sep 2026

AI video’s next move is inside the edit.

Runway’s editor plugins, Aleph’s frame-guided changes and export options show how generation is moving into existing editing workflows.

Read article

Tools / Analysis12 Sep 2026

Runway in the edit: what the Adobe plugins change

Runway’s Adobe plugins bring generation into the timeline. The useful question is what happens to the cut.

Read article

Tools / Analysis12 Sep 2026

Astra: from a rendered image to an editable scene

Thomas Ricouard’s architectural case study shows Astra working with geometry, materials, lighting and cameras in Blender.

Read article

Tools / Analysis12 Sep 2026

GWM Worlds 2: when the viewer controls the camera

GWM Worlds 2 explores responsive generated worlds. The creative promise comes with a different continuity problem.

Read article

Release / Tools10 Aug 2026

Seedance 2.5: what thirty seconds change

Because the real threshold is not duration. It is whether performance, geography and intention survive the whole scene.

Read article

Field note30 Jul 2026

A generator makes a shot. An agent tries to remember the film.

I saved this because continuity—not prettier single shots—is becoming the real production problem.

Read article

Tools29 Jul 2026

Over 200 models in InVideo: how do you know which one is working?

Useful when speed matters; risky when routing hides the model, cost and repeatability.

Read article

Open call26 Jul 2026

AI Film Festival Japan asks filmmakers to disclose the process—not hide it.

Process disclosure makes authorship visible instead of treating AI as a magic trick.

Read article

Open call25 Jul 2026

We Are Human puts AI cinema against a human-rights question.

A human question gives the technology something more demanding than spectacle.

Read article

05 / About

The film comes first. The technology follows.

I’m Dmytro Uhnivenko, an AI director, creative director and AI artist based in Kyiv, Ukraine, working worldwide.

I create cinematic brand films, commercials and visual worlds from concept to final frame—combining story development, art direction, graphic design, motion, 3D, AI production and post.

About, services & collaboration

06 / ContactKyiv, Ukraine · Worldwide

Have an idea that refuses to be ordinary?

Let’s give it a world

For brands, agencies, production studios and creative teams: AI films, cinematic campaigns and visual worlds. Project briefs and creative roles welcome.

dmytro.uhnivenko@gmail.com