Image to video

Image to video turns a still, a product shot, or a reference frame into a short clip. Creators use it when they already have a key visual and need motion for TikTok or Shorts, or when a faceless storyboard starts from locked frames. This matrix only includes the eight tools with verified pages on this directory as of 2026-10-16. Photo avatars and URL-to-ad image uploads are called out as different jobs so they are not silently upgraded into generic I2V.

How to read this table

• Web is observed when the tool page documents image-to-video, reference frames, or multi-image video in the creator or API product.
• API quotes developer URLs and named modes. Kling explicitly lists image-to-video and multi-image-to-video.
• MCP lists official servers. Absence of an official Kling MCP stays not observed.
• Price is a starter band from the tool page. Per-second or per-unit arithmetic belongs on model pages when printed there.
• Verified is 2026-10-16. Re-open host schemas before you assume aspect ratios or durations.

Selection tips

• Locked storyboard frames for faceless: Kling, Seedance, Higgsfield, or Runway, then faceless.
• Product stills inside an ad template: prefer Creatify for the ad job; this matrix keeps classic I2V not_observed there so URL-to-ad is not confused with a general I2V playground.
• Talking photo presenters: HeyGen photo avatars are a lip-sync presenter path — see lip sync and AI avatar, not this row as generic I2V.
• CapCut's Seedance spend can animate inside an editor; a dedicated CapCut I2V SKU name was not observed on the pricing extract, so the web cell stays conservative.

Related workflows and models

• Workflows: faceless, ugc-ads.
• Models: kling and seedance set i2v: true on their model pages; runway-gen keeps a conservative i2v: false filter where the Gen-4.5 price card is silent — read both the tool and model pages.

What creators usually get wrong

Image to video is not a photo avatar and it is not a product-URL ad. HeyGen photo avatars start from a still and produce a talking presenter; that job is filed under lip sync and AI avatar, which is why this matrix leaves HeyGen I2V as not observed. Creatify can ingest product images for URL-to-ad; that is still an ad pipeline, so classic playground I2V stays not observed on that row until a tool page isolates the mode.

Kling's developer list is the clearest API wording among the eight tools for image-to-video and multi-image-to-video. Dreamina and Higgsfield document references and frames on the creator side. Runway's tool capabilities include i2v while the Gen-4.5 model page stays conservative where the price card is silent — read both pages before you quote a mode.

CapCut can spend Seedance credits inside an editor. Without a dedicated CapCut I2V SKU on the pricing extract, the web cell stays not observed rather than upgraded from a related Seedance sentence. Finish captions somewhere else; I2V outputs still need CapCut or another caption bench for Shorts and TikTok type.

FAQ

Why is Runway web observed while the Gen-4.5 model filter is cautious?

The Runway tool page lists i2v among capabilities. The model page for Gen-4.5 refuses to assert I2V from a price card that does not split modes. Both statements can be true. Confirm the mode in the app or API you bill.

Is a HeyGen photo avatar image-to-video?

It starts from a photo, but the product is a talking presenter. This matrix leaves HeyGen I2V not_observed and points avatar buyers to the lip-sync and AI avatar pages.

Does CapCut Pro include image-to-video?

Seedance generation is on paid CapCut plans. A dedicated image-to-video SKU label was not observed on the CapCut pricing extract used for the tool page.

Where are fal and Replicate?

Omited until this directory verifies tool pages for them. Do not invent API cells from memory.

How should I finish captions?

Use captions / CapCut. I2V tools on this matrix are not caption benches.