Sora 2 Image-to-Video
Animate a photo with OpenAI Sora 2 image-to-video: 4–20s clips, synchronized audio, 720p portrait or landscape. 200–1,000 credits per clip, no waitlist.
Sora 2 Image-to-Video turns a single reference image into a coherent clip while preserving the subject's identity, lighting and composition. It supports durations from 4 to 20 seconds, generates synchronized audio with the picture, and outputs 720p in portrait or landscape orientation. On Conut.ai pricing is fixed per clip rather than per second: 200 credits for 4 seconds, up to 1,000 credits for 20 seconds.
- Duration: 4s – 20s
- Resolution: 720p
- Price: 200–1,000 credits/clip
- Audio: Synchronized
How to use Sora 2 Image-to-Video
The image fixes the look; the prompt directs the motion. Pick a source that can carry the intended camera move and keep the described action small.
- Upload a sharp, well-lit source image A clear subject with space around it gives the model room to move. Sora 2 preserves the reference closely, so source problems carry into the clip.
- Describe motion and sound, not appearance Say what moves, what stays still, one camera move and the sounds that belong to the scene. The image already fixes the subject's look.
- Choose the orientation before uploading Portrait and landscape are both 720p. Crop the source to the destination ratio so the model does not have to invent edges.
- Start at 4 seconds The cheapest tier tests whether the photo animates cleanly. Extend the winning prompt to 8 or 12 seconds only after identity holds.
- Watch identity through the final second Faces and hands drift first. If the subject changes, shorten the duration or use a cleaner source rather than adding prompt detail.
Settings reference
Sora 2 Image-to-Video starts from one photo, keeps its subject and composition, and adds motion plus synchronized audio.
| Setting | Value | Notes |
|---|---|---|
| Start image | 1 image, required | Upload a PNG or JPEG. Sora 2 preserves identity, lighting and composition from the source. |
| Prompt | Required | Describe what moves, the camera behavior, the mood and any sound cues. |
| Duration | 4s / 8s / 12s / 16s / 20s | Five fixed tiers; there is no continuous duration slider. |
| Orientation | Portrait 720×1280 or landscape 1280×720 | Both output 720p; the source image is fitted to the chosen frame. |
| Price | 200–1,000 credits per clip | Billed per clip, not per second: 200 for 4s up to 1,000 for 20s. |
| Audio | Synchronized with the picture | Ambience and effects line up with the action; there is no audio upload. |
| Last frame | Not supported | The clip evolves from the start image without a defined ending frame. |
| Seed & negative prompt | Not exposed | Reuse the same prompt for a similar result, but expect variation between takes. |
What Sora 2 image-to-video is good at
Sora 2 is built for believable motion and camera behavior. When you give it a reference image, it keeps the subject recognisable and the overall composition close to the source while adding movement, camera dynamics and environmental effects. The result is less of a photo animation and more of a short shot that happens to start from your image.
Synchronized audio is the other differentiator. The model generates sound with the video, so footsteps, wind, traffic and other ambience line up with what is happening on screen. That makes it a strong fit for cinematic social clips, trailers and concept shots where silence would feel unfinished.
- Preserves identity, lighting and composition from the reference image
- Natural motion and camera dynamics
- Synchronized audio generated with the clip
- Portrait or landscape 720p output
- Durations from 4 to 20 seconds
Sora 2 duration and pricing
Pricing is per clip, so the cost rises in fixed steps rather than by resolution or per second. The 4-second tier is the cheapest way to test a prompt; the 8- and 12-second tiers are the sweet spot for social clips; 16 and 20 seconds are for shots that need a longer beat.
- No per-second billing; the price is fixed by duration
- Only 720p output is available on this endpoint
- Portrait (720×1280) and landscape (1280×720) sizes
- No negative prompt or seed controls exposed
| Duration | Credits | Best for |
|---|---|---|
| 4s | 200 | Prompt and motion tests |
| 8s | 400 | Social clips |
| 12s | 600 | Short scenes |
| 16s | 800 | Longer cinematic beats |
| 20s | 1,000 | Trailers and hero shots |
Sora 2 vs WAN 2.6 vs Seedance 2.5
Sora 2 sits between the cheaper WAN models and the high-resolution Seedance tiers. It offers the longest clip among the 720p options and generates synchronized audio, but it does not match WAN 2.6 on resolution or Seedance 2.5 on 4K.
- Choose Sora 2 for its motion style, synchronized audio and long 720p clips.
- Choose WAN 2.6 for 1080p at a lower per-second cost.
- Choose WAN 2.5 when you need to upload a specific voice track.
- Choose Seedance 2.5 when 4K or a 30-second take is required.
| Model | Resolution | Duration | Price | Audio | Best for |
|---|---|---|---|---|---|
| Sora 2 I2V | 720p | 4–20s | 200–1,000 /clip | Synchronized | OpenAI motion and audio |
| WAN 2.6 I2V | 720p / 1080p | 5–15s | 30–50 /s | Native sound | Cheaper 1080p |
| WAN 2.5 I2V | 480p–1080p | 5–10s | 20–40 /s | A/V sync + upload | Talking shots |
| Seedance 2.5 I2V | 480p–4K | 5–30s | 60–600 /s | Native audio | 4K and 30-second clips |
Prompting Sora 2 for motion and sound
Write the prompt like a shot description: subject, action, camera, light and mood. Sora 2 responds well to cinematic language and to a clear statement of what stays still. Because the reference image already fixes the look, the prompt should focus on what changes over the clip.
For audio, describe the sounds that match the visible action: rain on a window, footsteps on gravel, a crowd in the background. Keep the audio description short and avoid asking for a scripted voice performance; this endpoint generates synchronized ambience and effects, not a dialogue tool.
- State what moves and what stays still
- Name one camera move: slow zoom in, tracking shot, orbit, tilt up
- Describe the light and mood rather than repeating the image contents
- Add one or two sound cues tied to visible action
- Draft at 4 seconds, then extend the winning prompt to 8 or 12
Limitations and when to use another model
This endpoint outputs 720p only. If you need 1080p or 4K, WAN 2.6 or Seedance 2.5 are the better choices. Pricing is fixed per duration, so a long clip can cost more than a per-second model at the same length. The endpoint also does not expose a negative prompt or seed, which limits reproducibility.
Sora 2 is strongest when the shot benefits from its motion and audio style. If you need precise dialogue lip sync, WAN 2.5 is the direct alternative. If you need cheap silent drafts, WAN 2.2 Spicy costs far less per second.
- 720p maximum resolution
- Fixed per-clip pricing; compare against per-second models at your duration
- No negative prompt or seed exposed
- Audio is synchronized ambience, not scripted dialogue
- For lip sync with a chosen voice, use WAN 2.5
Prompt examples for Sora 2 Image-to-Video
- Cinematic street scene: The subject walks slowly toward the camera through a rain-soaked street at night, neon reflections on the pavement, slow dolly-in, shallow depth of field. Distant traffic, rain on metal, low synth hum. One action, one camera move, two sound cues.
- Product hero shot: The product sits on a dark reflective surface while the camera orbits slowly; a single spotlight sweeps across the label, dust particles drifting in the beam. Soft mechanical click, deep room tone, luxury commercial lighting. Keep the product static and let the camera and light create the motion.
- Portrait in motion: The subject turns their head toward a window, hair moving slightly, curtains drifting in the breeze, warm late-afternoon light. Slow push-in, cinematic portrait, quiet room tone and distant birds. Small human motion is the safest way to preserve identity.
- Landscape reveal: Morning fog rolls over a mountain ridge as the first sunlight breaks through the trees, camera slowly pushes forward, volumetric light rays, cold green and amber palette. Wind in the pines, distant birds. Landscapes tolerate large camera moves without a subject to warp.
- Sci-fi corridor: The subject stands in a dark sci-fi corridor as overhead lights activate one by one, leading toward a sealed door. Camera tracks forward past the subject, steam venting from pipes, cool blue and amber lighting. Electrical hum, distant alarms. Use environmental motion to add energy without moving the subject much.
- Trailer moment: The character stands on a rooftop as a storm gathers, coat whipping in the wind, lightning illuminating the skyline behind them. Camera slowly tilts up from the feet to the face, dramatic orchestral-scale atmosphere. Thunder, wind, rain. A 16- or 20-second tier gives the build-up room to breathe.
Pro tips for Sora 2 Image-to-Video
Image conditioning makes this endpoint predictable. Spend your effort on the photo and the one camera move.
- Small motion preserves identity A head turn, a slow push-in or drifting curtains survive better than a full-body action. Choose the smallest motion that tells the story.
- There is no last frame Unlike Seedance and Kling image endpoints, Sora 2 has no end-frame control. If the shot must land on a specific composition, plan that in the prompt or edit around it.
- One action and one camera move The prompt should read like a shot brief. Competing instructions are the fastest way to get sliding feet or a warped background.
- Use 8 and 12 seconds for social The middle tiers are the practical range for short-form video. Four seconds is a test; sixteen and twenty are for trailers and hero shots.
- Move up only when pixels matter Sora 2 stops at 720p. If the final file needs 1080p, render the motion idea here and rebuild the shot on WAN 2.6 or Seedance 2.5.
Frequently asked questions
How much does Sora 2 image-to-video cost?
Pricing is fixed per clip: 200 credits for 4 seconds, 400 for 8, 600 for 12, 800 for 16 and 1,000 for 20 seconds.
How long can a Sora 2 image-to-video clip be?
Up to 20 seconds. The endpoint supports 4, 8, 12, 16 and 20 second durations.
Does Sora 2 image-to-video generate audio?
Yes. It generates synchronized audio with the picture, including ambience and effects that match the visible action. It is not a scripted dialogue or lip-sync tool.
What resolution does Sora 2 image-to-video output?
720p. You can choose portrait (720×1280) or landscape (1280×720). For 1080p or 4K, use WAN 2.6 or Seedance 2.5.
What is the difference between Sora 2 text-to-video and image-to-video?
Text-to-video generates the scene from a prompt alone. Image-to-video starts from your reference image and preserves its identity, lighting and composition while adding motion and audio.
Is a prompt required?
Yes. This endpoint requires a prompt. Describe the motion, camera behavior, mood and any sound cues you want.
Can I upload my own audio track?
No. Sora 2 image-to-video generates audio with the clip but does not expose an audio upload field. Use WAN 2.5 image-to-video if you need a specific voice track.
What images work best with Sora 2?
A sharp, well-lit image with a clear subject and composition. The model preserves the reference closely, so a clean source image gives it the best starting point.
Compare with other image-to-video models
- Sora 2 Text-to-Video
- WAN 2.6 Image-to-Video
- WAN 2.5 Image-to-Video
- Seedance 2.5 Image-to-Video
- Kling O3 Pro Image-to-Video
- Browse all AI video models
Specifications
- Provider: Wavespeed
- Model type: Image to video
- Credit pricing: 200 credits per video
- Capabilities: new, cinematic, realistic, audio-sync, multi-shot
About Conut.ai
Conut.ai is an AI image, video and audio generation platform with 50+ creative tools. It supports text-to-image, image-to-video, face swap, motion control, relighting, inpainting and video effects through models including Nano Banana Pro, Seedream, GPT Image, Sora 2, Kling, WAN and Seedance. Start free — no credit card required.