Direct AI like a cinematographer.
A cinematic shot library tagged the way a DP thinks - size, angle, camera move, lens, light. Every frame comes with a prompt written for the model you actually use.
Reference libraries were built for film crews. You direct a model.
ShotDeck and FilmGrab show you a beautiful still and stop there. A video model needs to know how the camera moves, what the lens is, where the light comes from - in words it understands.
Filter by the camera, not the film title
Push-in, orbit, handheld, drone. Shot size, lens, light, time of day, palette, mood. The filters a video model actually responds to.
A prompt per model, not one prompt for all
Seedance wants structure, Veo wants audio, Runway wants motion only, Midjourney wants parameters. Each frame ships with six formats.
Decks that become shot lists
Collect frames into decks, share a link with the client, export a printable shot list with every prompt attached.
The one filter nobody else has.
Motion is what separates a video prompt from an image prompt. Every shot in the library is tagged with its camera move.
One frame. Everything a model needs.
A man in a dark coat eating noodles at a tiny counter under pink and cyan neon signs, steam rising, rain behind him. He pauses mid-bite as someone sits next to him out of frame. Camera: slow dolly-in camera move, medium close-up, eye-level angle. Lens: anamorphic. Lighting: neon. Style: cinematic sci-fi, neon magenta cyan grade, film grain, mysterious mood. Duration 5s.
[medium close-up, eye-level angle] A man in a dark coat eating noodles at a tiny counter under pink and cyan neon signs, steam rising, rain behind him. He pauses mid-bite as someone sits next to him out of frame. Camera movement: slow dolly-in camera move. neon lighting, neon magenta cyan tones, shot on anamorphic, cinematic, realistic motion. Negative: text, warping, morphing faces.
A cinematic sci-fi shot. Medium close-up from an eye-level angle, slow dolly-in camera move, anamorphic lens. A man in a dark coat eating noodles at a tiny counter under pink and cyan neon signs, steam rising, rain behind him; he pauses mid-bite as someone sits next to him out of frame. Lighting: neon; color: neon magenta cyan; mood: mysterious. Audio: sizzling wok, rain, low synth hum. No dialogue, no subtitles.
A man in a dark coat eating noodles at a tiny counter under pink and cyan neon signs, steam rising, rain behind him. Cinematography: Camera shot: medium close-up, eye-level angle, anamorphic lens Camera motion: slow dolly-in camera move Lighting: neon, night Palette: neon magenta cyan Mood: mysterious Actions: - He pauses mid-bite as someone sits next to him out of frame Sound: sizzling wok, rain, low synth hum
Slow dolly-in camera move. He pauses mid-bite as someone sits next to him out of frame. Keep the neon lighting and neon magenta cyan palette of the reference frame. Natural motion, cinematic mysterious tone.
A man in a dark coat eating noodles at a tiny counter under pink and cyan neon signs, steam rising, rain behind him. He pauses mid-bite as someone sits next to him out of frame. The camera dollies in. Medium close-up, eye-level angle, anamorphic lens, neon lighting, night, neon magenta cyan color grade, cinematic sci-fi, film grain. Negative prompt: text, subtitles, watermark, blurry details, deformed hands, extra fingers, morphing faces.
cinematic film still, a man in a dark coat eating noodles at a tiny counter under pink and cyan neon signs, steam rising, rain behind him, medium close-up, eye-level angle, anamorphic, neon lighting, neon magenta cyan color grading, mysterious, sci-fi film, 35mm grain --ar 21:9 --style raw --v 7
Honest comparison.
ShotDeck is the best reference library for film crews. We are built for a different job.
| ShotGrammar | ShotDeck | Flim | FilmGrab | |
|---|---|---|---|---|
| Prompts for video models | 6 formats per frame | No | Generation credits | No |
| Filter by camera move | Yes | No | No | No |
| Frames you can reuse | Original, licence-clean | Reference only | Reference only | Reference only |
| Lens, light, palette tags | Yes | Yes, very deep | Partial | No |
| Library size | 130 | 1.5M+ | 1.5M+ | 100k+ |
| Price | Free / $9 per month | ~$13 per month | €0-35 per month | Free |
Questions.
Are these real film stills?
No. Every frame is original and generated for ShotGrammar, then tagged by hand. That is why you can use them as image-to-video input and in client decks without worrying about who owns the film.
Which models are the prompts written for?
Seedance, Kling, Veo 3, Runway and Midjourney, plus a neutral image prompt that works in Flux, Imagen, Sora and most others. Each follows the structure that model responds to best.
Can I use a frame as the first frame of my video?
Yes, on Pro and Studio you can download the full-resolution 2.39:1 frame and use it as image-to-video input, including in commercial work.
What does the free plan include?
The whole library, all filters, 15 prompt copies per month and one deck. No card needed.
Can I cancel anytime?
Yes. You keep your plan until the end of the paid period, then drop back to Free. Your decks stay.