Google's Veo 3 is the model most people mean when they say AI video "got good." It is worth understanding what changed, because it explains what you should expect from any app in this category — including this one.
What Veo 3 changed
Earlier video models failed in recognisable ways: faces drifted between frames, hands morphed, motion violated physics, and lighting shifted incoherently from second to second. Veo 3 made visible progress on all four:
- Temporal consistency — a subject stays the same subject across the clip.
- Physics-aware motion — hair, cloth, water and dust behave roughly the way they should.
- Prompt fidelity over time — the video still matches the description at second 8, not just second 1.
- Cinematic composition — depth of field, colour and lighting behave more like a camera and less like a filter.
The practical result is footage that can sit next to real production material without immediately announcing itself. We wrote more about the implications in Veo 3 and the new bar for cinematic quality.
Where Rigel fits — and what we don't claim
Rigel is not Google's Veo app, and we don't claim to resell Veo 3. Rigel is a mobile product built on top of frontier multimodal video models of this generation, with a pipeline our team re-benchmarks as new models ship. What you are buying is the workflow — photo in, finished vertical clip out, on a phone — running on models of this class rather than on a model we trained ourselves.
How to judge a Veo-class result
When you compare outputs from any app, look at the things that used to break:
- Face stability. Pause at 1s, 4s and 8s. Is it the same person?
- Hands and edges. Fingers and the boundary between subject and background are where weak models fall apart.
- Motion weight. Does hair settle, does cloth have inertia, does the camera move like a camera?
- Light continuity. One consistent light source, or a slideshow of different ones?
- Prompt obedience at the end. Most models drift late in the clip.
Getting frontier-quality video out of a photo
The workflow in Rigel is the one described on the photo to video page: upload a still, write one specific sentence about motion, framing and light, generate, post. Two things move the needle far more than anything else — a clean source image, and a prompt that specifies camera movement and lighting rather than mood adjectives.
For competing frontier work in the same generation, see the Seedance 2.0 page, which covers the multi-reference, audio-aware side of the field.
Pricing, plainly
We won't quote Google's pricing for Veo, because it changes and we don't control it. What we can state is our own: Rigel is free to download on iOS 16.6+ and Android and includes starter credits, with credit packs and a subscription available in-app. Each generation shows its credit cost before you confirm it, so nothing is spent by surprise.
The fastest test is your own photo: download Rigel, spend the starter credits on an image you know well, and judge the five points above for yourself.