Start with the evidence you can collect
Imagine two projects. You have a drawing of a character that has never existed physically. You also have a small stone ornament on your desk that you want to document. The drawing provides an appearance to interpret; the ornament can be photographed from multiple sides. Choosing the method starts here, before comparing file formats or software prices.
For a concept figurine, a plausible back may be enough. For documenting a distinctive carving, inventing the back would defeat the purpose. Write down which details must be preserved and whether you can actually observe them.
What one-image generation can and cannot tell you
An image-to-3D generator proposes volume from a reference. It must resolve ambiguity about depth and unseen surfaces. The useful result is a new mesh inspired by visible evidence, not an automatic measurement of every side. A frontal image of a hand holding a torch cannot prove the exact gap behind the wrist.
This is why the Prometheus community example on Reforgers is labelled an interpretation. Rotate it, compare the silhouette with the reference and look for areas where the model makes a choice the image cannot verify. Better visual coherence is valuable, but it should not be confused with historical or dimensional fidelity.
What changes with photogrammetry
Photogrammetry uses observations from different camera positions to reconstruct shared features. Agisoft’s explanation describes that multi-view principle. The job is not merely collecting many photos: the views must provide useful common evidence.
For capture, plan overlapping views around the object and avoid leaving important surfaces unseen. Agisoft’s capture guidance highlights overlap and difficult reflective or featureless surfaces. Do not apply powders or coatings to valuable objects simply to improve a scan; choose an appropriate, non-damaging capture method instead.
A practical decision before you start
Choose generation if you have a drawing, an illustration or a single reference and want to develop a visual idea. Choose a measured capture workflow if the object exists, you can access it, and preserving observed geometry matters more than inventing a pleasing result.
If you cannot photograph the back, mark that uncertainty rather than filling it and calling the result exact. If dimensions matter, introduce a suitable known scale and validate it. Cultural Heritage Imaging’s scale-bar guidance explains why calibrated references matter for measurement; a reconstruction is not automatically a certified survey.
The same export does not mean the same reliability
Either workflow may ultimately give you an STL or another mesh format. The extension does not reveal how well the shape matches reality. Before printing, inspect dimensions, disconnected parts, thin details and supports. Keep the reference images and a note of the method with the file.
A useful handoff says “generated from one image” or describes the capture method, instead of simply saying “3D scan”. This article is a method-selection guide, not a side-by-side accuracy benchmark. The cover is a schematic of multi-view capture, not a camera plan validated for every object.
Frequently asked questions
Can a single photograph give me an exact 360-degree scan?+
Not by itself. It does not record all hidden surfaces. A generation can infer them, but that is different from observing them.
Does photogrammetry guarantee printing success?+
No. Capture quality, scale and mesh condition still need checks, followed by slicing and printer-specific preparation.