dario1122 @criptopifia

As a builder in the AI video space, I was terrified about how much markup API providers charge for simply routing the models. Higgsfield API adds a 40% markup even with its maximum advertised discount. Fal adds 100%. Runway adds 414% (Seedance 2 480p) I might be hated by the industry, but : 
 Today, we’re launching Sume API, the first image & video model provider with 0% markup. For example with Seedance 2.5 1080p • 50% cheaper than fal • 50% cheaper than higgsfield api (29% even with biggest discount) • 28.02% cheaper than kie ai The routing infrastructure layer is never the moat. We believe that APIs should be commoditized, and the moats will be built at the application layer. Check the pricing comparisons below, and see how the broken industry is saved.

Most AI films don't look fake because the model is bad. They look fake because nobody directed the light. You can generate 4K footage, perfect faces, and elaborate camera moves. But if everything is evenly bright, the scene still feels artificial. Traditional filmmakers have spent decades learning to control light. AI filmmakers need those same skills. Here are 6 lighting rules that changed how I approach AI scenes: → Start with one key light. Place it 30–60° to one side and slightly above eye level. Direction gives a face shape. → Shoot the shadow side. Position the camera toward the darker side of the face. You get depth without asking for "more cinematic." → Motivate the light. A window, lamp, monitor, or streetlight should explain where the glow comes from. Random light sources make a scene feel staged. → Light the environment first. Give the room depth, then light your character to match it. Otherwise, they look pasted onto the background. → Use negative fill. Sometimes the fix is removing reflected light, not adding more. Keep the shadows that give a face dimension. → Control softness. A big, close source creates gradual shadows. A small, hard source creates sharp ones. Choose based on the story. Now compare these AI video prompts: Weak: "Ultra-realistic cinematic lighting, dramatic mood, 4K." Better: "Warm practical lamp frame-left motivates a large soft key 45° off axis, slightly above eye level. Camera on the shadow side. Cool window glow shapes the background. Gentle falloff; negative fill deepens the near cheek." One is a list of adjectives. The other is a lighting setup a cinematographer could actually build. That difference matters whether you're generating a single image, an AI short film, or a full commercial. I built this lighting playbook to make those principles easier to apply, scene by scene. Better AI films won't come from prompts that sound cinematic. They'll come from people who know how to direct a shot. . .

@epee_boy No m lo creo. K se comore una casa y en 2 años lo vuelve a cobrar. Salud

@Madridpulse0 K gran error ni venderlo .... x lo k dieran

@trunksRM7 Follarian a pelo?

@TraderMystico T da para pagar la plataforma y poco mas no?

The problem: How do you avoid the stiff camera work? Blender is notoriously stiff and unnatural. We solved this by having Claude screenshot the blender sequences as storyboards and then re-writing our Seedance prompts on Dreamina, attaching storyboards like this.

keyboard_arrow_left Previous keyboard_arrow_right Next

We lost thousands because we couldn't control the Seedance outputs. We'd write a prompt and Seedance would do whatever it wanted. Wrong camera, wrong blocking, ships changing size between shots. So we'd regenerate. Then regenerate again. That's how you hit $2,000+ a day in hit or miss gens. Our director, @TheoDudley made all these stunning assets but they were impossible to show the scale and blocking just by reference images alone.

keyboard_arrow_left Previous keyboard_arrow_right Next

@pablo_talifan Que entren con seguro al pais. Como yo cuando viajo que me saco uno privado x si pasa algo. Como vamos a estwr trabajando para pagarle la sanidad a colombianos y negros. No me jodas es un puto chiste

@tirajnews Quedo genial. Ya no se nota

Stop wasting money on AI videos 💸 Most of the time you don’t even need a prompt to generate one. Just copy this: “Replace the person in {video} with the person in {character sheet}.” The motion, the camera and the audio come from the video. The face comes from the sheet. But if your character sheet is bad, you’ll just waste tokens. At $0.318/sec every failed 8-sec try burns ~$2.50, and the face starts drifting by clip 3. Meet Leo Carter - my character: → 8 reels posted → 56.4K followers → Every single reel over 892K views, top clip at 7.5M → The same skinny shirtless kid in grey pants and grey sneakers The character: dark messy curly hair, thin mustache, extremely skinny build with long arms, dark grey pants with white drawstring, worn grey sneakers. A quiet gym kid who looks like he shouldn’t be there until he starts lifting. He doesn’t talk. He doesn’t flex for the camera. He just walks in, grabs the bar, and the whole gym watches. Watch 0:03, the guy next to him stops mid-set That’s the whole format. Same face, same body, new gym every time. So I fed 2 of his photos into this prompt: Create a photorealistic character continuity sheet from the reference photos. Include a character profile, four full-body views (front, 3/4, side, back), three facial close-ups, eight expressions, six poses, costume details, a colour/material palette and a “Do Not Change” checklist. Preserve the same face, eye alignment, hair, body proportions, clothing and accessories throughout. Use the layout reference only for design. White portrait page, bold black headings, clean grid, readable labels, light-grey studio backgrounds and soft natural lighting. No invented personal details, social-media interface or watermarks. Attach 2-3 pics of your character. Look at my result (pic 1), done in about 5 minutes: → Profile + 5-angle turnaround: front, 3/4, side, back 3/4, back → 3 face close-ups + eyes, nose, mouth, skin texture → 10 expressions, 8 poses including lifting and flexing → Costume details + colour/material palette → A “DO NOT CHANGE” block the model can’t argue with What to steal from Leo: → One silhouette people recognize in 1 frame → The contrast IS the hook skinny kid, heavy weight, shocked gym bros → Build the sheet once, then every video starts from it → 8 videos. 7.5M on the best clip. 56K followers. The sheet did the work. The internet is full of fake faces. The money goes to the ones that never change

keyboard_arrow_left Previous keyboard_arrow_right Next

AI_GOAT @0xAIGOAT

2 weeks ago

x.com/i/article/2108…

10 32 297 894K 1K

@jerequaglino A veces hay k empezar a pegar puñetazos

@DrKrkcolu @rand_longevity Why q10?

@rand_longevity Nad. Omega 3. Zinc.

African in France says that Chinese businesses were looted in France because they have become too rich & wear nice clothes. They hate & resent anyone that works hard to build themselves a life. They feel entitled to destroy that life & steal from those that work hard because they hate & resent them. This is Multiculturalism. After Whites, East Asians are the 2nd most targeted group in the West. They hate the 2 highest IQ groups, that’s not a coincidence. It’s envy x.com/SleeplessBKKNe…

@yamatsullu 3

@eris_28 No. Sin dos rojos. Y respondem a los mismos

@RMadridUniido Perdio velocidad con las operaciones

@JordiWild Lleva a alvise jordi

@m_goes_distance Nos jugamos demasiado a al carta er100. Si sale mal...no veo mas opcion