Runway
Use Runway Gen4 Turbo on Martini for fast iteration on 3D-world-derived shots — captured stills from the navigable world feed into Gen4 image-to-video nodes that ship 5- or 10-second clips in under a minute. The 3D world is canvas-internal reference, not an exportable .obj/.fbx/.glb/USD file. Gen4 Turbo is the speed pick when the brief lands at 4 PM and a sequence ships at 9: per-clip generation completes faster than Sora 2 or Kling 3, which makes it the right tool for fast multi-shot iteration before committing the bigger render budget on the hero clips.
Source the world from Nano Banana 2 or FLUX.2 → World Labs / Image-to-3D-World node, OR from a text prompt routed through the Marble 3D node. Generation runs ~5 minutes. The output is a navigable canvas-internal scene preview — orbit, pan, screenshot. Cannot be exported as .obj/.fbx/.glb/USD from Martini. Captured stills are the bridge between the world and Gen4 Turbo.
Inside the navigable world, capture 4-5 matched-angle stills: front, three-quarter left, three-quarter right, back/over-shoulder, plus optional tight close. Each lands as an image node. Gen4 Turbo's speed advantage means you can iterate on more shot variations per session — capture more angles than you might for slower models, then pick the strongest sequence in post.
Runway Gen4 Turbo is image-to-video — there is no text-to-video; the source still is mandatory. Drop a Gen4 Turbo node per shot on the canvas. Wire each captured still as the starting frame. Each clip is 5s or 10s; the 5s clip is the iteration sweet spot for Turbo speed. Each clip inherits the world; only the camera-move prompt changes.
Gen4 Turbo's motion is anchored by the source still. Write the prompt as a motion description that flows from the existing composition: "slow camera push forward through the room toward the fireplace, soft afternoon light unchanged, no character motion." Avoid prompts that conflict with the source ("character runs into frame from the left" when there is no character). Keep motion subordinate to what's already there.
Generate 2-3 motion variations per shot at Gen4 Turbo speed (~30-60s per 5s clip). Compare on the canvas. Promote the strongest variant per shot to the final sequence. For hero shots that need maximum quality, switch the chosen variant to Sora 2 or Kling 3 — Gen4 Turbo is the iteration tool; the bigger models are the finishing tools when budget allows.
Drop the chosen Gen4 Turbo outputs (or the upgraded Sora 2 / Kling 3 hero shots) into Martini's sequence builder in story order. Layer audio. Export as native sequence to Premiere, DaVinci Resolve, or Final Cut. The locked 3D world made the multi-shot read as one set across all the iteration variants and the final hero shots; the NLE export is the final delivery.
Iteration variant 1. Gen4 Turbo speed lets you generate this in ~30-60s — fast feedback on the camera move.
[Captured still: wide establishing of foggy alley] + Gen4 Turbo prompt: slow camera push forward through the alley, neon signs flicker in mid-distance, rain particles in air, atmospheric mood, 5 seconds, 16:9.
Iteration variant 2 — same starting frame, different motion. Compare to variant 1; pick the stronger.
[Same captured still] + Gen4 Turbo prompt: gentle dolly forward with subtle parallax through the alley, signs unchanged, rain particles, 5 seconds, 16:9.
Medium shot orbit. Gen4 Turbo produces clean orbital motion when the source still has clear depth structure.
[Captured still: three-quarter angle on the ramen shop] + Gen4 Turbo prompt: gentle orbit clockwise around the lantern outside the shop, soft lantern light unchanged, no character, 5 seconds, 16:9.
Detail close-up. Static + slow zoom is Gen4 Turbo's clean territory; complex parallax is where Sora 2 wins.
[Captured still: tight close on neon sign] + Gen4 Turbo prompt: static camera, slow zoom in on the neon sign, sign flickers subtly, atmospheric, 5 seconds, 16:9.
Gen4 Turbo is image-to-video only — there is no text-to-video. The captured still from the world is mandatory.
Use 5-second clips for iteration speed. Switch to 10s only when the camera move needs more time to develop.
Generate 2-3 motion variants per shot; pick the strongest in post. Gen4 Turbo speed makes this affordable.
For hero shots that need maximum quality, switch the locked variant to Sora 2 or Kling 3. Gen4 Turbo is iteration; bigger models are finishing.
Avoid motion prompts that conflict with the source still ("character enters from left" when no character exists). Keep motion subordinate to the existing composition.
The 3D world is canvas-internal — Gen4 Turbo uses captured stills, not the navigable world directly. Export from Martini = NLE-ready video, not a 3D file.
Gen4 Turbo returns 5- or 10-second 1080p clips per node, with generation time under a minute per 5s clip. Speed advantage makes it the iteration tool for multi-shot 3D-world workflows; Sora 2 and Kling 3 are the finishing tools. The 3D world remains canvas-internal (not exportable as .obj/.fbx/.glb/USD); Gen4 Turbo outputs are exportable video deliverables. Iterate fast on Gen4 Turbo, promote hero shots to Sora 2 / Kling 3 when budget allows, then NLE-export the final locked sequence.
Connect Runway Gen4 Turbo with other AI models on Martini's infinite canvas. No GPU required — start free.
Get Started FreeOpenAI
Lock down a location once with a navigable 3D world on Martini, capture matched-angle stills, and feed each as a starting frame into Sora 2 for a five-shot sequence that reads as one set instead of five different rooms. Sora 2 has deep understanding of 3D space, motion, and scene continuity — captured stills from a navigable world translate to consistent camera moves with parallax, occlusion, and depth honored. The 3D world is canvas-internal reference, not an exportable .obj/.fbx/.glb/USD file; Sora 2 reads the captured stills and ships exportable video clips that all share the same spatial anchor.
View guideKling
Drive Kling 3.0 multi-shot sequences on Martini using captured stills from a navigable 3D world — Kling supports 2-6 scenes per video with explicit per-scene descriptions, which makes it the strongest single-pass multi-shot pick when paired with locked location backplates. The 3D world is canvas-internal reference, not an exportable .obj/.fbx/.glb/USD file. Kling reads the captured stills as starting frames and renders multi-shot videos that all share the locked location, with 3D Spacetime Joint Attention handling parallax and occlusion across cuts.
View guide