Explorer · Maker · AI

Joined June 2021
180 Photos and videos
Pinned Tweet
Followed up on the same idea to create this clip. Started with a single frame from Krea 2, used GPT Image 2 to create the 3x3 contact sheet preserving the same aesthetic and dropped both into Seedance 2. Some flaws in the final clip but I think the aesthetic holds pretty well from the original image to the seedance gen. Krea 2 → GPT Image 2 → Seedance 2 was all done within @krea_ai seedance prompt in the next post 👇
I've been chasing this for ages, Expand a frame from my favourite aesthetic models while keeping the exact visual language, colour palette, composition style, mood, and worldbuilding is much harder. I think I'm finally getting somewhere with GPT Image 2 prompt in next post
35
54
626
74,622
// The Burning
Jun 5
Here’s a system prompt you can use inside a ChatGPT or Claude project. The main idea is simple: You feed it a basic idea, a rough image prompt, a scene description, or even an uploaded image, and it enhance it and return with 10 cinematic prompts, each exploring a different composition, camera angle. For example, I used: “a gladiator riding a horse on a mountainside” And these are the results! It's a great way to explore different visual languages, discover interesting compositions. System prompt below. 👇
1
1
18
898
V retweeted
Jun 10
// no more sacrifices midjourney v8.1 → seedance 2.0 via @invideoOfficial with image references only. image and video prompts in next posts 👇
15
5
92
4,687
V retweeted
May 17
"Life Sometimes" \\ It began as one Krea 2 image. Then: - GPT Image 2 for the storyboard - NB 2 to isolate and extract frames and edit some - Kling o3 for video using start frame/end frame All done inside @krea_ai x @runwayml @suno for the music
13
8
109
8,917
V retweeted
Jun 11
🏆 Moodboard Challenge — Results! That's a wrap on the moodboard challenge. 35 entries in one week. Thank you to everyone who posted and made the first big challenge feel special. Now, the winners: 🥇 Winner: @ibexdream 🥈 Runner-up: @VictorInFocus 🌟 Community pick: @Quentin21118961 Congrats! 🎉
Jun 3
Krea Moodboard Challenge is live 🎨 Make a custom moodboard, generate with it, and submit your best results. 200k credits in rewards. Entry link below 👇
2
2
17
1,799
Jun 11
Midjourney V8.2 testing soon 💥
We've made V8.1 the new default model for all users on Midjourney. V8 will now be deprecated in 2 weeks. V8.2 will start testing extremely soon.
10
497
Jun 11
some alternate Midjourney v8.1 gens from the saem world, prompts in alt
Jun 10
// no more sacrifices midjourney v8.1 → seedance 2.0 via @invideoOfficial with image references only. image and video prompts in next posts 👇
3
3
52
2,820
Jun 10
// no more sacrifices midjourney v8.1 → seedance 2.0 via @invideoOfficial with image references only. image and video prompts in next posts 👇
15
5
92
4,687
Jun 10
Midjourney v8.1 prompt in alt
2
11
553
Jun 10
Seedance 2.0 prompts, x3 15 sec prompts 1/ Woman's monologue 0:00 – 0:02 CAMERA Wide shot from above. Pyramid towering. PERFORMANCE The woman stands at the top platform. A man stands opposite her. SOUND Wind through stone. Distant jungle. 0:02 – 0:04 CAMERA Medium shot on both. PERFORMANCE She looks at him. speaks in a Yucatecan Spanish accent, soft and deliberate, low emotion. “You built this to reach the sky.” 0:04 – 0:06 CAMERA Closer. PERFORMANCE “To be seen.” 0:06 – 0:08 CAMERA Cut to the sun overhead. PERFORMANCE (V.O. feel, but spoken) “To be chosen.” 0:08 – 0:10 CAMERA Back to her. PERFORMANCE “But it was never looking at you.” 0:10 – 0:12 CAMERA Hold. PERFORMANCE “It was waiting.” 0:12 – 0:15 CAMERA Wide shot. Sunlight shifts slightly. PERFORMANCE “For the moment.” _______ 2/ Man's monologue 0:00 – 0:02 CAMERA Medium close-up on him. Harsh sun on one side of his face. Wind moves through his hair and clothing. PERFORMANCE He looks at her, breathing through anger. He wants to stay controlled, but he is tired of performing restraint. SOUND Wind through stone. Distant jungle. Low breath. 0:02 – 0:04 CAMERA Medium close-up on her. PERFORMANCE She watches him without softening. Her face is calm, almost ceremonial. She is waiting for him to reveal what he truly wants. SOUND Silence between them. Cloth moving in the wind. 0:04 – 0:07 CAMERA Medium close-up on him. PERFORMANCE He speaks slowly, with a natural Yucatecan Spanish accent. Not theatrical. Not robotic. Wounded, frustrated, and deliberate. “You still think the sky chose you.” He lets the words sit. His jaw tightens after speaking, not before. 0:07 – 0:09 CAMERA Close-up on her face. PERFORMANCE A tiny change in her eyes. She is offended, but she will not show it fully. Her chin lifts by the smallest amount. SOUND Wind rises slightly. 0:09 – 0:11 CAMERA Close-up on him. PERFORMANCE His voice drops. Less anger now, more exhaustion. “I was here too.” He swallows. His eyes flick down, then back to her. 0:11 – 0:13 CAMERA Close-up on her mouth, then her eyes. PERFORMANCE She almost answers. She chooses not to. Her breath slows. Her focus sharpens. SOUND A quiet inhale. 0:13 – 0:15 CAMERA Extreme close-up on him. PERFORMANCE His expression changes from hurt to decision. The last of his restraint leaves him. Low, final, heavy. “I am done kneeling.” 0:15 – 0:16 CAMERA Extreme close-up on her eyes. PERFORMANCE She does not blink. Her gaze hardens. She accepts the challenge. 0:16 – 0:17 CAMERA Extreme close-up on his hand. PERFORMANCE His fingers open and close once. No weapon. He loosens his wrist, preparing for hand-to-hand combat. 0:17 – 0:18 CAMERA Extreme close-up on her hand. PERFORMANCE Her fingers flex slowly. Her palm turns slightly outward. Calm, precise, dangerous. 0:18 – 0:19 CAMERA Extreme close-up on his shoulder and neck. PERFORMANCE His breathing steadies. His posture shifts forward by a fraction. 0:19 – 0:20 CAMERA Extreme close-up on her feet. PERFORMANCE One foot slides back across the stone. She sets her stance. 0:20 – 0:21 CAMERA Extreme close-up on his eyes. PERFORMANCE He moves first. CUT before impact. 3/ Fight Scene Use the pyramid-top image as the strict start frame and environment reference. 15-second, no-dialogue, immersive high-speed martial arts duel. Two fighters: man in dark robes on the left, woman in pale layered robes on the right. Live-action realism. Brutal close-quarters hand-to-hand duel on the exact exposed pyramid-top platform. No weapons, props, debris, extra characters, or invented architecture. Use only the stone slab, steps, ledges, open drop, sky, haze, and distant pyramid silhouette. The Sequence: “Pyramid Deadlock” 0 to 2s: Instant Ignition Action: Start from the image composition. The man explodes forward with a low aggressive entry. The woman steps off-line as he closes. Camera: Rush forward from the start frame, diving low across the stone, locked to the attack. Audio: Wind breaks into footsteps, breath, snapping robes, first body collision. Visual: Dust scuffs underfoot. Robes whip in the wind. 2 to 5s: First Clash Blitz Action: Forearm strike, hard block, wrist turn, palm counter, low thigh kick, sidestep, grip fight. The man shoulder-checks. The woman absorbs, pivots, redirects him across stone. Camera: Tight side-tracking, whip pans following strikes, grips, kicks, and direction changes. SFX: Heavy impacts, feet scraping stone, cloth snaps, breath bursts. 5 to 7s: Body Lock Reversal Action: Chest-to-chest arm lock. Both push with full-body tension. She twists his wrist down. He breaks grip, ducks under an elbow, cuts behind her shoulder line. Camera: Partial 360° orbit, then fast whip pan on reversal. Visual: Wind pulls fabric. Distant pyramid remains soft and hazy. 7 to 10s: Close-Quarters Surge Action: She attacks with elbow, palm strike, low kick. He catches the arm, steps inside, shoves her back with shoulder and hip pressure. She recovers near the step and counters with a low sweep. Camera: Low stone-level tracking, fast lateral movement, sudden stop as sweep passes close. SFX: Feet grind stone, weight shifts, controlled breathing. 10 to 13s: Final Exchange Action: Dense burst: palm strike, block, elbow, knee, low kick, wrist control, shoulder bind, breakaway, immediate re-entry. No pause. Suffocating but readable. Camera: Aggressive close tracking, whip pans, impact shake. Camera feels inside the duel. Visual: Robes whip. Dust lifts. Kicks stay grounded and close-range. 13 to 15s: The Finish Action: He launches the final palm strike. She slips outside it, traps his wrist, and delivers a huge thunderous final blow to the sternum. He hits the stone hard. She remains standing over him, calm and in control. Camera: Rush into the final clash, snap with her pivot, then jolt on impact. Hold close and low on her dominant frame above him. Audio: Sharp breath, feet scraping stone, one heavy body impact. Wind returns. Silence holds.
6
682
// sunburst
7
1
34
675
V retweeted

Followed up on the same idea to create this clip. Started with a single frame from Krea 2, used GPT Image 2 to create the 3x3 contact sheet preserving the same aesthetic and dropped both into Seedance 2. Some flaws in the final clip but I think the aesthetic holds pretty well from the original image to the seedance gen. Krea 2 → GPT Image 2 → Seedance 2 was all done within @krea_ai seedance prompt in the next post 👇
3
2
12
651
V retweeted
Been playing with Krea 2. Trained a LoRA on a few cool-toned Wes Anderson shots, then threw it into a blender with random moodboards and prompts for "a scene from a surreal and impossible fantasy movies" No real objective. Just exploring the edges of the aesthetic. Some random favourites and link to moodboard.. krea.ai/moodboards?share=576…
6
4
88
7,400
Aftermath midjourney v8.1 prompts in the alt
2
27
810
If your shot burst only gives you 2 or 3 usable frames, try this... Create a few different sheets, then only extract the ones you want to use. This version of the shot burst prompt numbers each frame, which makes the extraction step easier. Also, try adding multiple images as references and increase the shot numbers, this one was created from 4 references and is a 4x4 shot burst. Visual consistency is still maintained fairly well. Prompts in the next posts 👇
Followed up on the same idea to create this clip. Started with a single frame from Krea 2, used GPT Image 2 to create the 3x3 contact sheet preserving the same aesthetic and dropped both into Seedance 2. Some flaws in the final clip but I think the aesthetic holds pretty well from the original image to the seedance gen. Krea 2 → GPT Image 2 → Seedance 2 was all done within @krea_ai seedance prompt in the next post 👇
16
25
297
20,271
Prompt 1 (numbered shot burst) "Use the attached image as the canonical visual anchor and generate a 3x3/4x4/5x5 storyboard shot burst in 21:9. Preserve the exact aesthetic, colour palette, lighting, mood, subject design, wardrobe, environment, production design, and cinematic style from the source image. Create distinct frames from the same scene, using varied cinematic coverage including but not limited to: 1. extreme wide 2. centred master 3. medium shot 4. side profile 5. rear angle 6. close-up 7. extreme close-up 8. detail shot 9. environmental mood shot Keep the same composition style, negative space, camera character, atmosphere, lens feeling, and visual restraint. No text, no labels, no redesign, no extra characters. Number each frame. The result should feel like a cinematographer’s contact sheet from the same film scene."
1
1
22
1,595
Prompt 2 (isolate and extract and individual frame) "Isolate frames [FRAME NUMBERS] from the contact sheet and recreate each selected frame as a separate high-resolution [ASPECT RATIO] image. Preserve the original composition, framing, lighting, colour palette, subject placement, atmosphere, lens feel, and cinematic style exactly. Do not redesign, reinterpret, crop awkwardly, add new objects, or introduce new details. Each extracted frame should feel like a polished standalone film still at the highest possible resolution."
1
12
1,290
Followed up on the same idea to create this clip. Started with a single frame from Krea 2, used GPT Image 2 to create the 3x3 contact sheet preserving the same aesthetic and dropped both into Seedance 2. Some flaws in the final clip but I think the aesthetic holds pretty well from the original image to the seedance gen. Krea 2 → GPT Image 2 → Seedance 2 was all done within @krea_ai seedance prompt in the next post 👇
I've been chasing this for ages, Expand a frame from my favourite aesthetic models while keeping the exact visual language, colour palette, composition style, mood, and worldbuilding is much harder. I think I'm finally getting somewhere with GPT Image 2 prompt in next post
35
54
626
74,622
prompt with krea 2 original image as start frame and shot burst as reference use [shot burst image] for camera shot reference. 15-second, no-dialogue, fully immersive high-speed samurai duel short film. Two samurai from the attached image: foreground_samurai on the left and mist_samurai emerging from the fog. Live-action realistic style. Brutal close-quarters katana duel in a misty mountain field. No slow buildup. No prolonged approach. The duel ignites immediately. High-speed battlefield kenjutsu mixed with iaijutsu and close-range jujutsu: explosive sword draws, rapid dashes, hard blade clashes, parries, counter-cuts, evasive pivots, blade locks, shoulder collisions, grip fighting, off-balancing shoves, and desperate close-range reversals. Each samurai carries exactly one katana. Each katana has one clean, continuous blade. Keep every sword physically coherent, readable, and solid at all times. No duplicate blades, no split blades, no forked swords, no ghost swords, no extra weapon geometry, no glowing trails, no afterimages. No anime effects, no glowing swords, no fantasy armour, no wirework, no blood spray. The Sequence: “Mistfield Deadlock” 0 to 2s: Instant Ignition Action: Start from the attached image composition. foreground_samurai instantly draws his katana and explodes forward through the tall grass. mist_samurai reacts immediately, surging out of the fog with his blade rising. Camera: The camera rushes forward from the wide frame, diving low through the grass, locked tightly to the attack. Audio: Wind and silence are shattered by a sharp blade draw, pounding footsteps, whipping grass, and a violent metallic clash. Visual: Fog tears open around their bodies. Tall grass bends and parts as they collide almost immediately. 2 to 5s: First Clash Blitz Action: Rapid katana exchange: diagonal slash, hard parry, wrist turn, counter-cut, sidestep, blade bind. foreground_samurai drives in with a shoulder check. mist_samurai absorbs it, pivots, and redirects him through the grass. Camera: Tight side-tracking with sudden whip pans following the fighters’ body movement and sword direction. Sharp push-pull movement as the fighters close and separate. SFX: Heavy steel impacts, robe fabric snapping, damp grass tearing underfoot, short breath bursts. 5 to 7s: Blade Lock and Reversal Action: Their swords lock close to their faces. Both push with full body tension. mist_samurai twists the blade down. foreground_samurai breaks the lock, ducks under a fast horizontal cut, and spins behind him. Camera: Partial 360° orbit around the fighters during the blade lock, then a fast whip pan as the reversal happens. Visual: Fog swirls around them. Mountains stay soft and distant behind the chaos. 7 to 10s: Close-Quarters Surge Action: mist_samurai turns hard and attacks with a rapid overhead strike. foreground_samurai catches it, slides the blade away, steps inside, and shoves him back with the hilt and shoulder. mist_samurai nearly falls, recovers instantly, and counters with a low rising cut. Camera: Low grass-level tracking, fast lateral movement, sudden stop as the low blade passes close to camera. SFX: Blade slicing through grass, boots skidding in wet earth, deep controlled breathing, sharp steel scrape. 10 to 13s: Final Exchange Action: Both samurai commit fully. Dense burst of attacks: slash, parry, counter-slash, body turn, blade bind, breakaway, immediate re-entry. No pause, no wasted motion. The fight becomes suffocating but still readable. Camera: Aggressive close tracking with rapid directional shifts, whip pans, and responsive shake from each clash. The camera feels embedded inside the duel. Visual: Robes whip violently. Mist fragments around movement. Each sword remains a single clean blade shape throughout. 13 to 15s: The Deadlock Action: Both samurai launch simultaneous finishing strikes. Their blades freeze inches from each other’s necks, locked in perfect lethal stillness. Neither wins. Neither retreats. Camera: The camera rushes into the final clash and stops sharply between them, holding close on tense faces, single clean blades, and heavy breathing. Audio: One final metallic ring hangs in the air. Wind returns. Grass sways. Environment: Misty mountain field at dawn, tall dark grass, distant mountain silhouettes, blue-grey desaturated palette, soft overcast light, heavy atmospheric fog, cold windswept mood. Rules: Preserve the exact mood and visual identity of the attached image. No modern objects, no extra characters, no subtitles, no text, no magic, no glowing weapons, no duplicate swords, no malformed blades. *final clip is two clips stitched together hence the 19 secs
1
27
2,705
I've been chasing this for ages, Expand a frame from my favourite aesthetic models while keeping the exact visual language, colour palette, composition style, mood, and worldbuilding is much harder. I think I'm finally getting somewhere with GPT Image 2 prompt in next post
28
15
237
57,651
I don’t really get the artifact anymore, I’ve noticed that you’re more likely to get it when GPT is creating the image from scratch and not using a reference or if you create iterations in the same chat thread, so I always start a new chat for each image generation
2
7
1,687

How do you avoid those fake artifacts? Or stupid texture gpt gives.
1
1,942