Keep one complete subject inside the frame
Show the whole object, including feet, handles, stems, ears, tips, and any base it rests on. Leave visible background around the outer edge. When an object touches the frame, the generator cannot tell whether the shape ends there or continues beyond the photograph.
A larger image is useful only when it contains clearer evidence. Upscaling a blurred or tightly cropped photograph adds pixels without revealing missing geometry. Start with the cleanest original image instead of an enlarged copy.
Preserve real detail during export
Use the original camera image when possible. Repeated screenshots and messaging-app copies can soften edges or add block artifacts that look like surface detail. Export with a normal color profile and moderate compression. A clean 1200-pixel image can be more useful than a noisy or artificially sharpened 4000-pixel copy.
Do not add a fake transparent background by deleting parts of the subject. If you remove a background, inspect the mask at full size and make sure it has not clipped fur, fingers, spokes, or thin edges. A natural contrasting background is often safer than a rough cutout.
Protect the silhouette and the gaps inside it
The silhouette tells the model where the subject ends. Choose a background with a different brightness or color from the object. Preserve the open space inside handles, between legs, around spokes, and beside other narrow parts. If two parts overlap in the photograph, the result may join them into one surface.
Imagine filling the subject with solid black. If its outline and important openings are still obvious, the image has a readable silhouette.
Use lighting that explains the shape
Even lighting with moderate shadows helps reveal curvature. A completely flat image can hide depth, while hard shadows may be mistaken for grooves or separate geometry. Avoid blown white highlights and crushed black areas because both remove surface information.
Reflections are especially difficult. A bright stripe across polished metal can look like an edge; a reflection in glass can look like another object. Diffuse the light or change the camera angle when possible.
Know which materials are difficult
| Source | Common problem | Better approach |
|---|---|---|
| Transparent glass | The background is visible through the object. | Use a matte proxy, edge lighting, or a different source. |
| Polished metal | Reflections resemble seams and depth changes. | Use diffuse light and a plain surrounding area. |
| Fur or leaves | Many small edges can become noisy surface detail. | Use a clean silhouette and expect to simplify the mesh. |
| Black object | Dark regions lose visible form. | Raise soft fill light without clipping highlights. |
| White object | Bright regions merge with the background. | Use a mid-tone contrasting background. |
Choose a useful camera angle
A three-quarter view often reveals width, depth, and the relationship between major parts in one photograph. A perfectly straight front view may hide depth; a steep overhead view may hide the sides. Match the angle to the feature that defines the object.
Perspective also matters. A very close wide-angle photograph can enlarge the front and shrink the back. Move farther away and zoom or crop modestly when you need more natural proportions.
Keep pose and support objects intentional
Flexible subjects should hold one clear pose. A dangling cord, moving tail, open fabric fold, or hand supporting the object may be reconstructed as solid geometry. When a temporary stand is unavoidable, separate it visually from the subject and plan to remove its geometry later.
For symmetrical objects, center the camera and keep the horizon level unless an angled view reveals essential depth. For asymmetrical objects, choose the side that exposes the most distinctive structure instead of assuming the front label is the most useful view.
Use a back view only when it matches
Pixal3D accepts one optional back image. It should show the same physical subject, pose, configuration, and attachments. Keep the camera near the same height and distance. If the front shows an open lid and the back shows a closed lid, the model must reconcile two incompatible shapes.
A matching back view helps describe hidden surfaces, but it does not guarantee exact dimensions. Continue to inspect the result and use a measuring or CAD workflow when fit matters.
Avoid several competing subjects
Photographs containing a hand, stand, ruler, neighboring object, or busy scene can produce merged geometry. Crop only enough to remove distractions while keeping the complete intended subject. If scale is important, apply the real dimensions later in a slicer or editor rather than leaving a reference object touching the subject.
Run a final source check
- The intended object is complete and is the only main subject.
- The outside edge and internal openings are visible.
- Highlights and shadows retain surface detail.
- The file opens normally as JPG, PNG, or WebP and is no larger than 12 MB.
- You have permission to process the photograph and its subject.
Open the file once at full size before uploading it. Zoom around the subject edge, confirm that no part is missing, and make sure the intended object is sharper than the background. This one-minute check prevents many failures that a later seed change cannot solve.
When those conditions are met, open the Photo2STL converter. If a clean source still gives an unexpected model, compare its symptoms with the troubleshooting guide.