MiniMax Hailuo 2.3 Standard on Kanvora
Animate one owned starting frame into a fixed six-second 768p clip for 24 credits through fal. Allow the existing 3-8 minute wait window and review motion only after the still is approved.
- Resolution
- 768p
- Credits
- 24 credits
- Input
- 1 starting frame + prompt
- Duration
- 6 seconds
- Expected wait
- 3-8 minutes
- Provider path
- fal image-to-video
- Registry verified
- Aug 20, 2026
Start with a complete direction, not a blank box.
This is a direct handoff to Kanvora's real creator, not an embedded generator. The model, operation, prompt, and recommended review settings are already attached. Upload the required private reference after opening the creator.
Open the generatorLock this starting frame for the full six seconds. The walnut drafting table, blackened-steel desk lamp, cream tracing paper, brass parallel rule, mechanical pencil, half-unrolled glass-pavilion blueprint, thin porcelain espresso on the cork coaster, linen pinboard, dusk Polaroid, charcoal merino throw with KANVORA embroidery, unsoaked spilled-ink pool, smoked-glass eucalyptus cylinder, and the rain-wet Tokyo street outside must keep their identity, silhouette, and placement. No people enter. No object morphs, melts, swaps, or is added. Materials that must stay readable: coarse walnut grain and micro-scratches on the brass fittings; paper tooth on the tracing sheet; a thin coffee-oil sheen on the porcelain rim; wool nap on the throw; rain on the floor-to-ceiling glass. Light that must stay: a steady 2700K pool from the lamp across the paper, never flickering; cool cyan-green city bounce through the glass; distant neon as soft bokeh. Mixed humid volume light in the room, not a new key. Atmosphere: rain, thin espresso steam, and wet-asphalt mist only. No fog bank that hides the desk. Lens: 35mm look, f/2.0 falloff, locked horizon, focus on the cup and the lamp pool. Layered motion across six seconds: 0-2s: dense rain streaks continue down the glass; two rivulets merge at the right pane and catch cyan bounce; wet asphalt reflections outside shimmer in place. 2-4s: a thin ribbon of espresso steam rises through the 2700K pool, threads past the lamp, and thins. Neon bokeh outside warms a half-stop then settles. Tiny water beads on the eucalyptus catch one specular. 4-6s: steam is almost gone. Rain continues at the same density. Camera eases forward about two percent total over the whole six seconds, no tilt, no pan, no whip, no rack the other way. The desk lamp stays a steady warm pool. Blueprint lines stay sharp. The ink puddle does not spread. The throw does not slip. Natural photoreal motion only. No on-screen text, watermark, or logo bug.
Official Kanvora examples
A site-owned starting frame and playable six-second AI-generated clip. The fixed stage keeps its size before and after the video loads.
Lock this starting frame for the full six seconds. The walnut drafting table, blackened-steel desk lamp, cream tracing paper, brass parallel rule, mechanical pencil, half-unrolled glass-pavilion blueprint, thin porcelain espresso on the cork coaster, linen pinboard, dusk Polaroid, charcoal merino throw with KANVORA embroidery, unsoaked spilled-ink pool, smoked-glass eucalyptus cylinder, and the rain-wet Tokyo street outside must keep their identity, silhouette, and placement. No people enter. No object morphs, melts, swaps, or is added. Materials that must stay readable: coarse walnut grain and micro-scratches on the brass fittings; paper tooth on the tracing sheet; a thin coffee-oil sheen on the porcelain rim; wool nap on the throw; rain on the floor-to-ceiling glass. Light that must stay: a steady 2700K pool from the lamp across the paper, never flickering; cool cyan-green city bounce through the glass; distant neon as soft bokeh. Mixed humid volume light in the room, not a new key. Atmosphere: rain, thin espresso steam, and wet-asphalt mist only. No fog bank that hides the desk. Lens: 35mm look, f/2.0 falloff, locked horizon, focus on the cup and the lamp pool. Layered motion across six seconds: 0-2s: dense rain streaks continue down the glass; two rivulets merge at the right pane and catch cyan bounce; wet asphalt reflections outside shimmer in place. 2-4s: a thin ribbon of espresso steam rises through the 2700K pool, threads past the lamp, and thins. Neon bokeh outside warms a half-stop then settles. Tiny water beads on the eucalyptus catch one specular. 4-6s: steam is almost gone. Rain continues at the same density. Camera eases forward about two percent total over the whole six seconds, no tilt, no pan, no whip, no rack the other way. The desk lamp stays a steady warm pool. Blueprint lines stay sharp. The ink puddle does not spread. The throw does not slip. Natural photoreal motion only. No on-screen text, watermark, or logo bug.
AI-generated · Kanvora officialA wide cinematic product still of a midnight architecture studio overlooking a rain-wet Tokyo side street. Center: a walnut drafting table with a large sheet of cream tracing paper, a brass parallel rule, a mechanical pencil, and a half-unrolled blueprint of a slender glass pavilion; a single espresso in a thin porcelain cup sits on a cork coaster, steam catching the light. Left: a blackened-steel desk lamp throwing a warm 2700K pool across the paper, specular highlights on the brass fittings and on a small pool of spilled ink that has not yet soaked in. Right: floor-to-ceiling glass, cool cyan-green city bounce, rain streaks, bokeh from distant neon and wet asphalt reflections. Mid-ground: a linen pinboard with a single Polaroid of the same pavilion at dusk, and a folded charcoal merino throw over the chair back, the word KANVORA embroidered in small graphite caps, slightly out of focus but readable. Foreground: a shallow-focus sprig of eucalyptus in a smoked-glass cylinder, tiny water beads on the leaves. 35mm look, f/2.0 falloff, rich wood grain, paper tooth, metal micro-scratches, no extra people, no watermark, no logo bug. Photoreal, quiet luxury, generous negative space in the upper third of the frame.
- Resolution
- 768p
- Credits
- 24 credits
- Duration
- 6 seconds
- Expected wait
- 3-8 minutes
From model page to stored proof
One public decision path, followed by the real authenticated workflow.
- 1
Open the prepared motion direction
The CTA selects the legal image-to-video operation, Hailuo model ID, and six-second motion prompt in the real creator.
- 2
Upload one starting frame
Add one JPEG, PNG, or WebP you are allowed to use. Kanvora does not accept multiple references or a region selection on this path.
- 3
Confirm the 24-credit attempt
Hailuo returns a fixed six-second 768p clip. The product uses a 3-8 minute wait expectation and a 900-second registry timeout.
- 4
Review the complete clip
Inspect camera movement, subject drift, shape changes, and the final frame before downloading the stored video.
Compare this live operation
Choose by the live Kanvora path, not by a model-family claim from somewhere else.
| Model | Resolution | Credits | Provider path | Use it when |
|---|---|---|---|---|
| MiniMax Hailuo 2.3 Standard | 768p | 24 credits | fal image-to-video | One starting frame plus a prompt, fixed six-second 768p video, 24 credits. |
| Seedance 2.5 Image to Video | 480p or 720p | 74-1,156 credits | fal Seedance video | fal · flagship starting-frame video · native audio · 4-30 seconds |
| Seedance 2.5 Multimodal Reference | 480p or 720p | 74-1,156 credits | fal Seedance video | fal · image, video, and audio reference · native audio · 4-30 seconds |
| Kling AI Avatar Standard | Source-framed portrait video | 10-1,405 credits | fal Kling AI Avatar | fal · Standard cost-effective talking portrait · audio-timed |
| Kling AI Avatar Pro | Source-framed portrait video | 20-2,875 credits | fal Kling AI Avatar | fal · Pro higher-quality talking portrait · audio-timed |
| Veo 3.1 Starting Frame | 720p or 1080p or 4K | 134-400 credits | Google direct Veo video | Google direct · Standard · one private starting frame · native audio |
| Veo 3.1 Reference Images | 720p or 1080p or 4K | 267-400 credits | Google direct Veo video | Google direct · Standard · up to 3 subject images · fixed 8 seconds |
| Veo 3.1 Video Transition | 720p or 1080p or 4K | 134-400 credits | Google direct Veo video | Google direct · Standard · first and last frames · native audio |
Frequently asked questions
Current Kanvora behavior, stated without extending the model's capability boundary.
How many credits does Hailuo 2.3 use on Kanvora?
MiniMax Hailuo 2.3 Standard uses 24 credits per six-second 768p image-to-video attempt on Kanvora.
How long should I expect to wait?
Use Kanvora’s existing 3-8 minute expectation for this video path. The registry timeout is 900 seconds so the longer provider job can settle without encouraging a duplicate submission.
Can Hailuo create a video from text alone here?
No. Kanvora requires one starting frame plus a motion prompt. This page does not expose text-to-video, multiple reference images, or region controls.
Does the starting frame guarantee stable motion?
No. Review the full clip for drift, morphing, lighting changes, and unintended subjects even when the starting frame is approved.
Take this model into the real creator.
The prepared prompt and legal model parameters go with you. Sign-in remains at the generation boundary.