Hayden Lee (@haydenleewrites) 's Twitter Profile
Hayden Lee

@haydenleewrites

Filmmaker+VFX+Actor+Existentialist. I write about creative discipline, philosophy and personal development.

ID: 1703960076749529088

linkhttps://gritdrivenmindset.com calendar_today19-09-2023 02:32:01

2,2K Tweet

106 Takipçi

189 Takip Edilen

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

The feeling's more important than logic. Stats: — Time: ~29 hours Image to video generations: 55 — Flux Kontext and image gens: ~500 — Mental state: mild existential angst I couldn't nail the motion of the tornados but good enough for now #AIart #aivideo #cyberpunk #cyberstate

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

When a cyborg goes wall climbing without a safety harness... Continuity of action for the climbing shots was done with Flux Kontext. AI Film Production Stats: - Time: ~70 hours - Image to video generations: 168 - Image generations: ~1,900 - Mental state: numb Tools: - Video:

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

Using AI to recreate recurring childhood dreams... On my 7th week of learning AI filmmaking, it took ~8 hours to make this 30-second sequence: •200+ image iterations •50+ video iterations I'm finding that storytelling with AI is still manual, just different. Hailuo gave the

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

"This meeting with oneself is, at first, the meeting with one’s own shadow. The shadow is a tight passage, a narrow door, whose painful constriction no one is spared who goes down to the deep well. But one must learn to know oneself in order to know who one is. For what comes

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

“She wasn’t dancing. She was broadcasting.” “I felt something… like a memory that wasn’t mine.” “The lights on her skin moved like a prayer.” #Cyberstate #NeuroSyncDancer #cyberpunk #CORESEC #City17 #SciFi

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

I used to make Backrooms videos using Maya and rendered with Arnold. It'd take anywhere between 3-6 minutes a frame, depending on resolution and sampling rate. A 720P five second clip would take about 8-10 hours on a single 3090 GPU. I generated this 5-second AI clip with

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

"Awakening of Consciousness" [fragment] SERA: Valkyrie Unit 17 CYBERSTATE Tools: Kling AI Hailuo AI (MiniMax) FLUX AI Thanks: Stevie Mac Kling AI [CORESEC INTERNAL LOG: N-GRFT.017] retrieved from CORESEC blacksite ██-██ > neural graft assimilation rate: 98.7% >

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

"Protocol S.E.R.A." [identity imprint protocol] • SERA: Valkyrie Unit 17 • CYBERSTATE Tools: Kling AI Hailuo AI (MiniMax) FLUX AI Thanks: Stevie Mac Kling AI [CORESEC INTERNAL LOG: N-SERA.73B] > retrieved from CORESEC subnet node ██-17 > identity imprint: STABILIZED >

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

AI generated fluid dynamics are becoming really amazing. Was blown away by the realism and detail in splashes, bubbles and even water droplets smearing down the lens. Its only a matter of time before AI replaces even high level FX software like Houdini. It also seems to

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

AI filmmaking is 90% mistakes. You iterate repeatedly, applying your direction, aesthetics, continuity, and timing— until it transcends "AI slop" and turns into cinema Final video: x.com/HaydenLeeWrite…

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

AI filmmaking process funnel: - 1100+ image generations -> 50 selects (4% of total generated images) - avg. 6-8 image to video gen iterations per shot - final edit runtime: 1 min 45sec Final result: x.com/HaydenLeeWrite…

AI filmmaking process funnel:  
- 1100+ image generations -> 50 selects (4% of total generated images)
- avg. 6-8 image to video gen iterations per shot
- final edit runtime: 1 min 45sec

Final result:
x.com/HaydenLeeWrite…
Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

Happy to announce that I've just partnered with Higgsfield AI 🧩's CPP. Their newest model includes a painter workflow, which gives creators the ability to direct the action with hand-drawn annotations. This is a major step towards greater control and consistency in generative

Dustin Hollywood (@dustinhollywood) 's Twitter Profile Photo

I gotta say something because this one account regularly attacks AI creators and daily tries to do the same to me, unfortunately that shit doesn’t work on me, I’ll go all day, I’ve got way more energy then them 🤣 but it doesn’t mean we can’t fight back :) Is this a 3D model? is the

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

Coinciding with the release of Kling 2.5, I'm posting a preview of my latest Cyberstate video which I've been working on for over a month now. Most of the shots were made using their newest model. Congratulations to the Kling AI team! Tools used: - Kling AI 2.5, 2.1 +

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

Totally off-brand Sora 2 text to video test. Will be back with my usual Cyberpunk stuff soon! Prompt: "1980s analog VHS recorded real life commercial of teenage mutant ninja turtles sitting around a table at Chuck-E Cheeses eating pizza. the chuck e cheese mascot mouse stands

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

CYBERSTATE: GYNOID Track: And One - Sometimes (unofficial music video) Production time: four days spread out over two weeks Tools: Kling AI Midjourney CORESEC // Retrieval Order [URBAN ASSET – OPERATION RED FAN] • CYBERSTATE Objective: Immediate containment and termination

Hayden Lee (@haydenleewrites) 's Twitter Profile Photo

SOCIAL HARMONY Tools: Kling AI Midjourney CORESEC // Public Assurance Broadcast [HARMONY INITIATIVE — TRANSMISSION 07] • CYBERSTATE Objective: Public rollout & normalization of Neural Harmony™ implant — Phase: UPTAKE / STABILIZATION. [CORESEC BROADCAST LOG] • uplink

Orcton (@orctonai) 's Twitter Profile Photo

"Talon Mist" A One Shot AI Short Cinematic Action Film. Video model is KlingAI upscaled with Topaz. Rest of the tools used listed below.