The gift
A card, a print or a mug of somebody rendered as a small handmade object. It reads as effort in a way a filtered photo does not.
One photo in · A clay sculpture out · No sign-up
Matte plasticine, softened edges, a faint thumbprint in the cheek — the same face, remade as something you could pick up.
A real run on the model this page uses. Drag the divider and watch the sweater: the knit does not survive as knit, it becomes a smooth clay surface with a soft dent where a fold used to be. That loss of detail is the effect working, not failing. Every generation differs, so read this as the range rather than a fixed outcome.
Drag to compare before and after
It rebuilds the subject of a photograph as if someone had modelled them out of plasticine — smooth matte surfaces, rounded volumes, a slightly oversized head, hair as sculpted strands rather than individual hairs.
The appeal is oddly specific: clay is the one stylisation that makes a person look *cared for*. A cartoon flattens you, a caricature exaggerates you, a 3D render polishes you — but clay implies somebody sat down and shaped you by hand, and left their thumbprints in the process. That is why the results get used as profile pictures and gifts far more than as jokes, and why people run it on their parents and their partners rather than on themselves.
Two things decide whether a clay pass works. The first is surface: real modelling clay has no specular highlight, so a version of you with glossy skin and a catchlight in the eye reads as plastic instead. The second is subtraction. Clay cannot hold fine detail — a sculptor working at this scale loses eyelashes, pores and stray hairs, and the model has to lose them too.
Results that keep photographic skin texture under a clay palette land in the uncanny middle where the material and the detail argue with each other.
It runs on the free image editor in the browser, image-to-image, with no sign-up wall in front of a first result. Simple, evenly lit photos convert best, and there is an honest limit worth knowing before you start: glasses. Thin metal frames have no clay equivalent, so they either thicken into something chunky or vanish, and neither is wrong exactly — it is what a real sculptor would face too.
Creative engine
Upload a photo, say how stylised you want the sculpt, and check the credit estimate before generating.
A mustard jumper comes back as mustard clay, so the sculpture still looks like a portrait of a specific person on a specific day.
The setting stays where it was rather than being replaced with a studio sweep, which keeps the result a photo of a sculpture rather than a stock render.
Clay is one point on a scale of hand-made looks: the moving, thumbprinted version lives in photo to claymation, a glossy collectible object in the AI figurine generator, and the animated-film treatment in the Pixar filter.
The first step matters more than the wording of the prompt.
Flat, soft lighting converts best, because clay has no glossy highlight to inherit. Hard side light and phone flash both fight the material you are asking for.
Ask for a bigger head and simpler features for the toy end of the scale, or a subtler sculpt that stays close to the real proportions. Both are the same run, one sentence apart.
If it comes back too glossy, ask for matte. Too plastic, ask for visible fingerprints. Changing one variable at a time gets you there faster than rewriting the prompt.
Mostly people making something for somebody else.
A card, a print or a mug of somebody rendered as a small handmade object. It reads as effort in a way a filtered photo does not.
Recognisably you without being a photograph of you, which is exactly the trade a lot of people want from an avatar.
Small brands running a soft, handmade look across a whole feed, where a consistent material matters more than any single image.
What the clay pass keeps, what it throws away, and where it struggles.
They get thicker or they disappear. Thin metal has no equivalent in modelling clay, so the model either sculpts a chunky version of the frame or drops it entirely — the same choice a person working in plasticine would face. If the glasses matter, say so in the prompt and expect them to come back heavier than they really are.
Almost always the highlights. Plastic and clay differ mainly in how they catch light: plastic has a hard bright glint, clay has none. If the result looks like a toy figure, ask for a matte surface with no gloss and faint fingerprint texture, and it moves back toward clay.
Yes, and animals often convert better than people, because nobody has a fixed idea of what a specific dog's cheekbone should look like. Fur becomes sculpted clumps rather than individual hairs, which is exactly what a modelled version should do.
Ask for a subtler sculpt and normal head proportions. The default leans toward the toy end — slightly oversized head, simplified features — because that is the look people arrive wanting. Dialling it back is a sentence, and the result stays closer to the photograph underneath.
Scale and intent. This makes a still art object: smooth, studio-lit, something you could hold. The claymation treatment makes a character from an animated film — visible tool marks, a seam at the jaw, real fabric clothes, lit like a miniature set. Same material, different craft, which is why they live on separate pages.
It should keep it. The pass is meant to change the material of the subject and leave the setting alone, so a plain wall stays a plain wall. If the background comes back replaced, run it again and say to keep the original scene — the model is not deterministic and a second attempt usually holds.
Not to see a result. The tool runs on the free image editor in the browser with credits to start, so you can put a photo through and judge it before deciding anything. Saving your history and reaching the heavier models is what an account adds.
The estimate appears before you generate and updates with the resolution you pick. Failed generations are refunded automatically unless they broke the content policy, so a second attempt at a different level of stylisation is a cheap thing to try.
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. |
The same photo, rebuilt in a different material or a different craft.
Soft matte surfaces, rounded volumes and a thumbprint in the cheek — from a photo you already have, free to start.
A real edit made with this tool — drag your own photo in on the left, the prompt is already written, then press Generate.
