Veo 3.1 Video Generator: Text, Image and Reference

About Veo 3.1

Veo 3.1 is Google's video model, and it is the one most people mean when they ask for footage that looks shot rather than rendered. Here you can run it three ways — from a written prompt, from one or two still frames, or from reference images — and see the exact credit cost before a single frame is generated.

Most video generators fall apart on the two things viewers notice first: how light behaves and how weight moves. Veo 3.1 holds both better than the open-weight models on this site, which is why it is the default recommendation for anything client-facing — product hero clips, real-estate walkthroughs, ad cutdowns that will be watched at full screen.

The workspace exposes two quality tiers. Fast renders at 300 credits per run and is the right choice while you are still hunting for the prompt. Quality renders at 3,000 credits and is worth paying for once the composition is locked, because it adds detail to exactly the frames you are going to keep.

Everything runs against a live model, so a run that never returns a video refunds your credits automatically. You can iterate without paying for failures, which is the part that usually makes people stop experimenting.

Reference to video accepts up to three images in a single run, one more than image to video, which stops at two. Each reference should show the same character or product from a different angle; the model uses them together to hold that subject steady as the shot moves. The catch is that this mode runs on Fast only and pins the frame at 16:9: choosing references switches the model back to Fast and the ratio to widescreen, so there is no Quality pass and no vertical version of the same shot. For vertical output, animate a single still in image to video instead.

The prompt field stops at 5,000 characters, which is more room than most shots need. Spend it on ordered events rather than adjectives: what enters frame, what the camera does next, what the subject does last. Once a render exists, a 16:9 result can be pulled again as an HD file from the result panel; that option is offered only for widescreen output, so a vertical clip finished at Fast stays at the resolution it was rendered. The HD pull re-fetches the same task, so it does not start a second generation.

How Veo 3.1 works

1. 1

   Pick a mode

   Text to video starts from a prompt alone. Image to video animates one still, or interpolates between a start and an end frame. Reference to video holds a character or product consistent across the shot.

2. 2

   Write the motion, not the picture

   Describe what moves, how far it moves, and what the camera does. Leave colour, lighting and styling out of the prompt — the model already has the frames.

3. 3

   Choose Fast or Quality, then render

   Fast costs 300 credits and is built for iteration. Switch to Quality at 3,000 credits for the final take. The cost is shown before the run starts, never after.

What Veo 3.1 is best for

• Product and brand spots

  A still product shot becomes a slow push-in with a believable highlight sweep — the kind of clip that normally needs a studio and a slider.

• UGC and short-form ads

  Vertical 9:16 output with natural motion holds up on Reels, TikTok and Shorts without the uncanny drift that makes viewers scroll past.

• Real estate and interiors

  Animate a daylight photograph into a slow camera move. Light and shadow stay consistent frame to frame, which is where cheaper models give themselves away.

• Storyboards you can show a client

  Turn the key frame of each scene into a moving beat. Cheaper than a previz artist and far more convincing than a static animatic.

Veo 3.1 at a glance

Model

Google Veo 3.1

Credit cost

300 per Fast run, 3,000 per Quality run

Modes

Text to video, image to video, reference to video

Aspect ratios

16:9 landscape, 9:16 vertical, auto

Inputs

Prompt, one or two key frames, or reference images

Output

MP4 video you can download and reuse

Veo 3.1 FAQ

How much does a Veo 3.1 video cost here?

Fast runs cost 300 credits and Quality runs cost 3,000 credits. New accounts start with 300 bonus credits and can claim 50 more on every day they log in, expiring after 30 days. The Starter plan at $4.90 per month includes 300 credits.

What is the difference between Fast and Quality?

Both run Veo 3.1. Fast is tuned for turnaround and costs a tenth as much, so it is the right tier for prompt iteration. Quality spends the extra credits on frame-level detail and is the one you keep for publishing.

Can I animate one of my own photos?

Yes. Image to video accepts a single still, or a start and an end frame when you want the shot to land on a specific pose. Portraits work best when the prompt describes a small movement — a blink, a breath, a turn of the head — rather than a full action.

What happens if a render fails?

Credits are returned automatically when the provider does not return a video. You are charged for output, not for attempts.

Do I need a subscription to use Veo 3.1?

No. You can buy credits without a plan, and the 300-credit Starter tier costs $4.90 a month or $24 a year. Plans exist to make repeat rendering cheaper, not to unlock the model.

Where should I start if I have never prompted a video model before?

Start with image to video and a still you already like. Write one sentence about what moves, one about the camera, and generate at Fast. The image-to-video prompt guide walks through the full structure.

Can I combine reference images with a Quality render?

No. Choosing reference to video switches the model back to Fast and sets the framing to 16:9, and that path accepts up to three reference images. If the final shot has to be a Quality render, or has to be vertical, rebuild it in image to video with a single still: that mode keeps the model selector and the 9:16 ratio, and the still already carries the pose you approved.