One photo in · Same subject, new viewpoint · Nothing to install
Move the camera, not the subject. Upload one photo and ask for the shot from a lower, higher or three-quarter position.
A real angle change, made with the model behind this page. Drag the divider: the same man, the same olive jacket, the same grey backdrop — but the camera has dropped low and moved round to his right, and the perspective and shadows moved with it. Every output is AI-generated, so your own result varies with the photo you upload — read this as a guide to what the tool does, not a fixed result you will get back.
Drag to compare before and after
It re-photographs a picture you already have from a position the camera never occupied. You ask for a lower viewpoint, a higher one, or a swing around to three-quarters, and the subject is rebuilt with the perspective, foreshortening and shadow direction that the new position implies. This is not rotation and it is not a tilt.
Skewing an image stretches the pixels that already exist; changing the viewpoint means working out what the subject looks like from somewhere else and drawing the parts the first frame never recorded. That is why the search for this is so specific — people have tried the transform tools and found they do not do the thing.
The re-shoot runs on the Nano Banana image model, in the browser, with free credits to start. Drop in the original frame — the flow opens in image-to-image — choose the format, and the credit estimate updates before you commit; a failed run is refunded automatically (unless it broke the content policy). Two rules decide whether the result holds.
Keep the move modest: thirty to forty-five degrees around, or a step up or down, stays believable, while asking for the back of a subject photographed from the front is asking the model to invent everything. And say what must not move — most disappointing runs are the ones where the camera angle changed and the model helpfully zoomed in as well, so the two frames no longer read as the same shot.
Creative engine
Upload a photo, say where the camera should stand, pick the format, and the credit estimate updates before you generate.
Rebuilds perspective and foreshortening for a new camera position instead of skewing the existing pixels.
Moves the shadows with the viewpoint, so the frame reads as a photograph rather than a warp.
Works in the browser with free credits — no 3D scene, no second reference image, no plugin. Changing the viewpoint leaves the subject alone. When the stance is the problem instead, the pose changer restages the person, and the selfie generator settles both at once.
From one frame to a second position, with the cost shown before you commit.
A sharp, well-lit photo with the subject fully in shot. Everything the new viewpoint reveals has to be inferred from what this frame contains.
Low and to the right, above looking down, forty-five degrees around to the left — and add the words that lock the framing, such as the same distance and the same crop.
Generate, then keep whichever viewpoint frames the subject best. A failed run is refunded (unless it broke the content policy), so a second position costs nothing but the wait.
Listing photos, portfolio crops, a group shot taken from the wrong side of the room.
Turn a single good frame into two or three viewpoints so a page, a carousel or a storyboard has variety without a second shoot.
Straight-on and eye-level is rarely the most flattering position. Drop the camera slightly, swing it a few degrees, and see the version you would have shot.
See an object from a low hero angle or from above without rebuilding the set, the light and the tabletop.
Common questions about moving the camera on a photo you already have with Kavel.
Thirty to forty-five degrees around the subject, or one clear step up or down, is the range that stays convincing. Beyond that the model is inventing surfaces the original frame never saw — ask for the back of someone photographed from the front and you get a plausible person who is no longer the same person.
Because a new viewpoint implies a new framing unless you forbid it, and the model tends to move closer. This is the single most common disappointment with this edit. Add the constraint explicitly — same distance, same crop, the subject filling the same share of the frame, do not zoom — and the two frames stay comparable.
Completely. A rotation or a perspective skew moves the pixels you already have, so the subject stretches and nothing new is revealed. Changing the viewpoint redraws the subject as it would appear from another position, including surfaces the first frame never recorded. That is the reason the transform tools in an editor cannot do this.
A plain or shallow-depth background moves cleanly. A detailed environment is harder, because every object in it has its own geometry to re-solve, and errors compound with distance from the subject. If the background matters as much as the subject, keep the move small.
Yes, and objects are often easier than faces — a bottle or a box has simple geometry and no identity to preserve. Reflective and transparent materials are the exception, since the reflections themselves have to be re-solved for the new position and that is where a wrong result is most visible.
From a modest move, yes. The further the camera swings, the more of the face is being inferred rather than copied, and identity drifts with it. If the likeness matters, keep the rotation small and check the result before using it anywhere the person will be recognised.
The estimate tracks the format and resolution you set, so the cost is visible first. Failed runs are refunded automatically (unless it broke the content policy), and new accounts start with free credits — enough to try a second viewpoint before spending anything.
Your upload is used to produce your image and retention is set out in the privacy policy — read it first if the frame is under embargo, since this tool is most useful precisely on work that has not shipped yet. When in doubt, run the test on a stand-in shot before the real one.
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.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
One-time top-ups — buy extra credits any time you run low.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
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.
Powered by
Use one balance across every supported image and video model — eleven of them today, listed below — and check the credit cost before each request.
Video models
Image models
Model credit guide
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. About 8 credits per 1K image, 12 at 2K, 18 at 4K. |
| Image | Nano Banana Pro | Consistent run times across generations. About 12 credits per 1K or 2K image, 21 at 4K. |
| Image | Nano Banana 2 Lite | Faster, simpler variant. A flat 8 credits per image at every resolution. Start here to test. |
| Image | GPT Image 2 | The lowest-cost image model here. About 3 credits per 1K image, 6 at 2K, 12 at 4K. |
| Image | Seedream 5.0 Lite | Flat pricing — about 8 credits per image whether you generate or edit. |
Other ways to re-stage a photo you already have, and the model behind the edit.
Say where the camera should have stood and get that frame instead. The subject does not move, so nothing else in the shot has to be reshot either.
A real edit made with this tool — drag your own photo in on the left, the prompt is already written, then press Generate.
