The remote presenter
A talk delivered from a spare room reads very differently against a plate that matches your seated height and your desk lamp. It is the cheapest production upgrade available to anyone working from home.
One description · Locked-off framing · Centre left empty
Green Screen Background Generator
Stock backdrops are the reason so many keyed shots look wrong: the camera is at the wrong height, the light comes from the wrong side, and forty other channels are using the same file. Describe the room you want to be standing in instead.
Shot from where your camera isAn empty middleNothing branded in frame
Our own runs
Two plates, one indoor and one outdoor
Both generated on this page's engine from a single description each. They are shown side by side because a plate has no before — what matters is whether each one leaves you somewhere to stand. Every output is AI-generated, so yours will differ.
Interior
The studio: seated eye level, empty desk line
Asked for a curved news-desk wall with cove lighting and a blurred skyline panel, framed from a locked-off camera at seated height. Notice what is missing: no desk in the middle, no chair, no presenter. The interesting detail — the lit cove, the reflective floor — all sits at the edges, and the middle third is deliberately dull so a keyed subject has somewhere to be.
Take the studio
Exterior
The rooftop: standing eye level, sun off to one side
Golden hour, a low parapet across the lower third, unbranded towers behind. The parapet is doing real work: it gives a horizon line for your feet to relate to, which is what stops a standing subject looking like a sticker. The sun sits off to one side rather than behind the centre, so you can light yourself to match without fighting a halo.
Take the rooftopCompositing tool
What makes a plate composite well
Keying is easy now. Matching is what still gives people away.
Image tool
Modern software will pull a clean key off a decently lit green screen in seconds, and that has quietly moved the difficulty somewhere else. What now separates a convincing composite from an obvious one is agreement between two images that were never in the same room: eye level, lens character, light direction, colour temperature, and depth of field.
Get those wrong and the key can be perfect while the shot still reads as a person pasted onto a poster.
That is why a generated plate has an advantage over a stock one that has nothing to do with novelty. You can specify the conditions. Seated eye level for a podcast, standing for a presentation. Soft even light from the left because that is where your key light is. Shallow focus so the background falls away the way a long lens would render it.
A locked-off camera, because a plate with implied motion in it fights a camera that is not moving. None of that is available when you are picking from a library.
The honest boundaries: what you get is a still image, so it holds the shot but nothing in it moves — no drifting clouds, no traffic, no flicker on the screens. It also does not key your footage for you; that is your streaming software or editor's job, and the screen behind you still has to be lit evenly for it to work at all.
If what you actually want is the background of an existing photo cleaned up rather than replaced, the product photo editor does that job in place.
Creative engine
How a plate is built
Describe the room, the eye level and the light, then key your footage onto it in your own software.
Locked off
Left empty
Still plate
Name the eye level first
Seated, standing, or low. It is the one parameter that ruins a composite on its own, and the one people never think to specify.
Match the light you already have
Look at which side your key light is on and ask for a plate lit from the same side. Two contradicting light directions read as fake instantly.
Four lines that decide whether the composite holds
A plate description is a camera report as much as a scene description.
State the eye level
"Shot from a locked-off camera at seated eye level" or "at standing eye level". This is the parameter that decides whether you look like you are in the room or in front of it.
State the light direction and time
Soft window light from the left, hard overhead, golden hour from camera right. Then set your own key light to agree with it. Two lights disagreeing is the second commonest tell after eye level.
Ask for the centre to be empty
Say it explicitly. Every instinct an image model has is to fill the middle of the frame with the subject, and you need that space. Push the detail to the edges.
Rule out signage and people
No lettering, no logos, no other figures. Trademarks in a background are a real problem on a monetised channel, and a stray person in the plate will appear to walk through you.
Who it's for
Where a made-to-order plate beats a stock one
Four situations where the library file is the thing holding the shot back.
The streamer with a small room
The background is the part of a stream you cannot renovate. Swapping in a rooftop or a studio is a set change that costs nothing and does not require moving furniture, next to the avatar as the other way to control what viewers see.
The course and demo recording
A consistent plate across forty lessons recorded over six months hides the fact that your room changed, the season changed and you moved house halfway through.
The interview cut together from two places
Two people in two cities look like one conversation when both plates are described from the same brief — same eye level, same light direction, same depth of field.
Answers
Background plate FAQ
What people check before keying onto a generated image.
No — it makes the image you key onto. The keying itself happens in OBS, Zoom, Premiere, Resolve or whatever you already use, and it still depends on lighting your actual green screen evenly. This page removes the background problem, not the lighting one.
Not always. Most streaming software now offers a backgroundless mode that segments you without one, and these plates work fine behind that. A real physical screen still gives a cleaner edge around hair and glasses, which is why anything recorded rather than streamed usually still uses one.
The plate comes back at a size suited to a 16:9 frame, which covers a 1080p stream comfortably. For a 4K timeline, generate at the largest option available and check the edges at full size before you build a series around it — an upscaled plate shows its softness fastest in fine architectural detail.
Not from this page — what you get is a still, and a still is genuinely the right choice for most talking-head work because moving backgrounds pull attention off the speaker. If you do want motion, take the plate you like and run it through image-to-video for a slow ambient loop.
Yes, including monetised streams, client work and course material. The reason the prompts here exclude signage and logos is precisely that: a plate with a real brand visible in it is a liability regardless of who generated it.
Work through it in this order: eye level, then light direction, then colour temperature, then focus. Nine times in ten it is the first two. A plate shot from standing height behind a seated presenter cannot be fixed by grading — regenerate it at the right height instead.
Kavel pricing — pay only for what you generate
One credit pool covers Nano Banana images and Seedance and Veo video. The cost shows before every run, and failed jobs are not charged. Use a subscription for ongoing work, or a one-time pack when you just need to top up.
50% OFFLimited-time half price — lock it in now
No sign-upIncl. taxStart generating
Free
Start generating before you pay anything, and keep earning credits every day you come back. No card, no account needed.
$0
What's included
- Free credits before you even sign up — no account needed
- A sign-up bonus the moment you create one
- Daily check-in credits that climb through the week
- Photo editing on Nano Banana 2 Lite
- Every model on this page stays visible — upgrade to unlock
VISAAMEXPayGooglePaylink
Secure checkout by Stripe. Card details never touch our servers.
Start hereIncl. tax
Starter
A monthly balance that keeps a steady run of clips and images going — on every model Kavel carries.
$59.80
The whole shelf is open from day one.$29.90 / month
Quantity1x · $29.90
Total: 3,150 credits
What's included
- Videos from $0.15 · images from $0.02
- 3,150 credits per month
- Up to 210 videos or 1,570 images a month
- Clean downloads, no watermark
- Works with Nano Banana images and Seedance and Veo video
- Cost preview before you generate
- Generation history
VISAAMEXPayGooglePaylink
Secure checkout by Stripe. Card details never touch our servers.
PopularPopular
Creator
For regularly making Kavel drafts, final clips, and the image assets that go with them.
$119.80
The best pick if you create on a regular basis.$59.90 / month
Quantity1x · $59.90
Total: 7,200 credits
What's included
- Videos from $0.13 · images from $0.017
- 7,200 credits per month
- Up to 480 videos or 3,600 images a month
- Clean downloads, no watermark
- Access to Seedance 2.0, Fast, and Mini
- Nano Banana 2 generation and editing
- Priority credit top-ups
VISAAMEXPayGooglePaylink
Secure checkout by Stripe. Card details never touch our servers.
ScaleIncl. tax
Studio
For more output, high-resolution testing, and larger batches.
$199.80
Ideal when several campaigns share one pool.$99.90 / month
Quantity1x · $99.90
Total: 13,200 credits
What's included
- Videos from $0.12 · images from $0.016
- 13,200 credits per month
- Up to 880 videos or 6,600 images a month
- Clean downloads, no watermark
- Built for high-volume Kavel generation
- Plenty of room for 4K Kavel runs
- Available across the supported models
VISAAMEXPayGooglePaylink
Secure checkout by Stripe. Card details never touch our servers.
Credit packs
One-time top-ups — buy extra credits any time you run low.
Credit Pack 1500
Pay once, generate whenever. Credits land instantly and never expire.
$39.80
$19.90
Quantity1x · $19.90
Total: 1,500 credits
What's included
- Videos from $0.20 · images from $0.027
- 1,500 credits
- Up to 100 videos or 750 images
- Clean downloads, no watermark
- No ongoing subscription
- Available across the supported models
- Credits never expire
VISAAMEXPayGooglePaylink
Secure checkout by Stripe. Card details never touch our servers.
Flexible
Credit Pack 3600
Pay once for a run of final video renders. Credits land instantly and never expire.
$79.80
$39.90
Quantity1x · $39.90
Total: 3,600 credits
What's included
- Videos from $0.17 · images from $0.023
- 3,600 credits
- Up to 240 videos or 1,800 images
- Clean downloads, no watermark
- No ongoing subscription
- Available across the supported models
- Use for both video and image generation
- Credits never expire
VISAAMEXPayGooglePaylink
Secure checkout by Stripe. Card details never touch our servers.
Batch
Credit Pack 9000
A production-scale top-up for large prompt batches.
$159.80
$79.90
Quantity1x · $79.90
Total: 9,000 credits
What's included
- Videos from $0.14 · images from $0.018
- 9,000 credits
- Up to 600 videos or 4,500 images
- Clean downloads, no watermark
- No ongoing subscription
- Handy for Seedance 2.0 final renders
- Credits never expire
VISAAMEXPayGooglePaylink
Secure checkout by Stripe. Card details never touch our servers.
Not sure which to choose? Contact us
Charges appear as “KAVEL AI” on your card statement. You can cancel any time from Settings → Billing; cancellation takes effect at the end of the period you already paid for. Operator details, the full model list, and the refund window are on the about page.
Powered by
One credit balance for every Kavel image and video model
Use one balance across every supported image and video model — all of them listed below — and check the credit cost before each request.
Video models
Seedance 2.5 — up to 30 seconds in one generationSeedance 2.0, Fast and MiniSeedance 1.5 ProVeo 3.1 (Lite / Fast)Kling 3.0LTX 2.5MiniMax H3Text-to-video, image-to-video and video-input workflowsCost preview before you generate
Image models
Nano Banana 2Nano Banana ProNano Banana 2 LiteGPT Image 2Seedream 5.0 LiteQwen Image 3Grok Imagine Image 2.0Text-to-image and image-editing workflowsOne shared credit balance
Model credit guide
What each model costs in credits
You're only charged for successful generations. The exact estimate in the generator varies by model, length, resolution, audio, and number of images.
| Type | Model | Credit cost |
|---|---|---|
| Video | Seedance 2.5 | The long-take tier, priced per second: 5s 480p ≈ 211 credits, 5s 720p ≈ 473. A full 30s take runs ≈ 1,260 at 480p and ≈ 2,835 at 720p. Supplying a reference clip lowers the per-second rate. |
| Video | Seedance 2.0 | 5s 720p image-to-video ≈ 188 credits; text-to-video ≈ 308 credits. Scales with resolution and length. |
| Video | Seedance 2 Fast | Faster and lower cost. 5s 720p text-to-video ≈ 248 credits. |
| Video | Seedance 2 Mini | The cheapest Seedance tier. 5s 720p ≈ 154 credits, 5s 480p ≈ 72 credits. |
| Video | Seedance 1.5 Pro | Audio doubles the rate. 5s 720p ≈ 27 credits silent, ≈ 53 with audio; 1080p ≈ 57 and ≈ 113. |
| Video | Veo 3.1 | Billed per video, not per second (Lite tier). About 45 credits at 720p and 53 at 1080p. |
| Video | Kling 3.0 | Audio raises the rate. 5s 720p ≈ 105 credits silent, ≈ 150 with audio; 1080p ≈ 135 and ≈ 203. |
| Video | MiniMax H3 | Fixed 2K, no resolution ladder. Priced per second — a 5s clip ≈ 158 credits. |
| Image | Nano Banana 2 | Generate or edit from text and images. 20 credits per 1K image, 30 at 2K, 45 at 4K. |
| Image | Nano Banana Pro | Consistent run times across generations. 30 credits per 1K or 2K image, 50 at 4K. |
| Image | Nano Banana 2 Lite | Faster, simpler variant, and the everyday editing price: a flat 8 credits per image at every resolution. |
| Image | GPT Image 2 | The lowest-cost premium image model here. 10 credits per 1K image, 15 at 2K, 30 at 4K. |
| Image | Seedream 5.0 Lite | Flat pricing — 20 credits per image whether you generate or edit. |
More tools
The rest of the set
Other things that decide what a viewer sees behind you.
Put a better room behind you
Describe the room, the eye level and the light, then key onto it in the software you already use — or put an avatar there instead.
