Image and video generation, in one place

One sentence in.
Finished visuals out.

Describe the shot you want. Cavro handles references, composition, and rendering — product photos, brand visuals, and social assets in seconds.

  • Reference-guided runs
  • Flexible aspect ratios, multiple results per run
  • Image and video in one workspace
  • Pay per credit or subscribe

Models

Integrated models, one place

Cavro picks a sensible default — or you pick the model yourself.


Showcase

MadeintheCreateworkspace.

Four runs, four things Create actually does — a reference, a sentence, a ratio, a still. Each one carries the settings it was made with.

Type
Image
Model
Nano Banana 2
Ratio
4:3
Count
1
References
1

Prompt sent to Create

Move the same candle jar onto a linen tablecloth in warm morning light. Keep the jar, the label and the proportions exactly as in the reference. Soft natural shadows, editorial product photography.

The same candle jar moved onto a linen tablecloth in warm morning light
Reference photo of an amber glass candle jar on a plain grey backdrop
Before · your referenceAfter · Cavro
⇔
Drag to compare · ← → keys

Every pair below is a real run: the reference on the left was attached as a reference image, and the result on the right is what Create returned for the prompt above. Switch the pair to see a different model and ratio.

  • Reference upload accepts up to 8 images.
  • Non-square results keep the ratio they were run at.

A real reference and a real result, side by side. Attach up to 8 references and describe the change you want — switch between three of these runs.


User feedback

What people make with Cavro

How people ship product shots, campaign visuals, and short videos — without leaving one workspace.

★★★★★5.0

I describe the shot, attach two references, and the first or second result is already usable. I stopped booking a photographer for small product drops.

Maya R.E-commerce founder
★★★★★5.0

Reference-guided runs are the part that sold me. The same bottle keeps its exact shape and label across every scene — that’s what every other tool got wrong for us.

Daniel K.Brand designer
★★★★★5.0

I shoot a still and a short rotation video from the same conversation. No export, no second tool, no re-describing everything from scratch.

Priya S.Social content lead
★★★★★5.0

Cost used to be the scary part of AI tools. Here I see the estimate before I hit Generate, and a failed run costs me nothing — so I actually experiment.

Tomás L.Freelance art director
★★★★★5.0

Batch variants changed how I work: four options in one pass, side by side, keep the one that fits. It feels like having a small studio on call.

Hannah W.Marketing manager
★★★★★5.0

I’m not a designer. I typed one sentence and had something I could put on a landing page the same afternoon. Ten minutes to learn the whole thing.

Kenji M.Indie founder

Pricing

Two ways to pay — pick what fits

Credits refresh every period and you can cancel anytime — or pay per use with one-off credits that never expire.

Pay as you go

For occasional, on-demand use

  • Buy only what you need
  • Top-up credits never expire
  • All integrated models
See pricing & plans

All prices and plan details are on the pricing page

Failed runs are not chargedSee the cost before you runSecure checkout, cancel anytime

FAQ

Questions, answered

Cavro is an AI generation workspace for images and video. You describe what you need, attach references if you have them, and get assets you can download or keep iterating on.

Open the workspace, type a description, and press Generate. You can attach up to 8 reference images and set the aspect ratio and count before running.

You can explore the workspace for free. Generations are paid for by subscription or per credit, and the estimated cost is always shown before you run one.

Different models, counts, and video durations cost different amounts. Credits are deducted only for successful generations. Top-up credits never expire; subscription credits refresh each period.

Yes. Switch the type to Video and pick a model: MiniMax H3 offers 5, 10, or 15 second durations, while the Seedance models offer 5 or 10 seconds.

Assets you generate are yours to use, including commercially. You are responsible for making sure your inputs and prompts do not infringe on anyone else’s rights.

Failed runs are not charged. You see the reason in the conversation and can retry from there.

The Model control lists the image models currently integrated, plus MiniMax H3 and the Seedance models for video. That list is always the live one — this page does not hard-code a model catalogue.

Start with one sentence.

Bring a reference, pick a model, and iterate in one conversation.