3DP Ocean

Docs

Create models

The create flow is intentionally short. You say what you are making, you choose one photo, the platform does the heavy lifting, and you confirm one picture before the expensive half runs. This page is the reference for each step plus the small choices that change the result.

The flow

Every generation walks through the same chips at the top of the screen, and you can go back to any previous step without losing state. From a single photo the platform builds one clean reference picture, shows it to you, and only then generates the model. You don't choose engines or modes, it's one streamlined path.

Step 1: What are you making?

Three choices, and they cost the same: 7 credits, whichever you pick. They change only what we tell the AI, so that a hand comes back as a hand and a whole figure comes back head to toe:

  • A whole figure or model. The full subject, head to toe. If the photo cuts part of it off, we draw the rest back in.
  • A body part. A hand, a head, a foot. Crop the photo to just the part you want; that crop is the instruction.
  • An object. A shoe, a helmet, a mug, a toy. We zoom out to take in the whole thing, but we never invent parts of it you didn't photograph.

Step 2: Choose a photo

Drag a JPG / PNG / WEBP image (up to 8 MB) onto the big drop zone, or click to browse. The image is uploaded to private storage immediately and never displayed to anyone else. No photo? You can describe what you want in words instead.

What makes a good photo

  • One clear subject. The AI generates whatever the photo's about. Crop or pick a photo where the subject takes up most of the frame.
  • Plain background helps. The platform tries to isolate the subject; a busy or cluttered background means more work for the AI and more chance of artifacts.
  • Even lighting. Harsh shadows confuse the depth estimator. Diffuse indoor lighting or open shade outdoors works best.
  • Show what you want made. For a whole figure, the whole body in frame. For a body part, crop tight to that part. For an object, every side that should end up on the model.
  • Higher resolution is fine. The pipeline downsamples to about 2048 px on the long edge, so anything above that just uploads slower without improving the output.

Step 3: Enhance

This step runs automatically. The platform sends your image through our image-enhancement pipeline and gives back a clean, studio-lit version of the same subject on a plain background.

That cleaned-up picture is what the 3D step works from. Nothing is drawn on top of it and nothing is re-posed. The model is built from your own subject, so what you approve on the next screen is exactly what goes in.

Step 4: Confirm the picture

Before the expensive half runs, the platform stops and shows you that reference picture. This is the image the 3D generator will work from, so it is the right moment to spot a swapped subject, a bad crop, or missing detail.

Three buttons: Confirm to continue, Try again to re-run the clean-up on the same photo for 1 credit, or Cancel and keep the rest of your credits.

What it costs

There's only one quality, and every model comes fully coloured, with full PBR textures (colour, normal, metalness, roughness) baked in at no extra charge. What you pay depends only on how much of the work you want us to do:

  • Automatic: 7 credits. Hand us a photo (or just describe it) and we handle every step: preparing the image, drawing the reference view, and building the model. One flat price for every kind. This is what most people use.
  • Step by step: from 5 credits. If you already have a clean image, the 3D model on its own is 5. Add 1 credit if you want us to prepare your photo, and 1 for each extra view we build. You see each price before you spend it, and you skip whatever you don't need.

See Billing for the full breakdown.

Step 5: Generate

The progress strip walks through pending → enhancing → generating → ready. Typical wall-clock time is 1–3 minutes. The viewer flips to the live 3D mesh as soon as it's ready, where you can spin it and toggle between the shaded and solid views.

Failed generations are rare but they happen, usually because the photo had no clear subject or the AI flagged a content rule. When a generation fails the credits are not charged.

Public or private

On the result screen you name the model and choose its visibility. Both cost nothing:

  • Public. Shown in Community models, where other people can find it, vote on it and download it.
  • Private. Only you can see and download it. You can flip it either way at any time from your library.

Where the model goes

Every successful generation lands in your dashboard library, your collection of models you can rename, re-file, re-download, delete, or share in Community models. Nothing is shared until you say so, and sharing is permanent: once a model is in Community models it stays there.