Gallery pages now spin objects the catalog never walked around. A registrar weighing image to 3d against a studio visit is deciding what the wall label may still claim.

Registrars face a narrower question than a software roundup. The exhibition page wants a form the visitor can turn. The label on the plinth already states a height, a date, and sometimes an edition. Those lines were not measured inside a browser. Mixing the two records is how a study model starts to impersonate the work.

A Small Gallery Has One Catalog Still

Most small shows already own a photograph. It was made for the checklist, the press mailer, or the page that lists the bronze before the doors open. The camera stood in front. The light was even enough to read the silhouette. Nobody walked to the back, because the photograph was never asked to be a survey.

The web page changed the ask. A visitor who can drag the object expects a side and an underside. The gallery can answer that in two ways. Send the bronze out, or bring a scanner in, and pay for a capture that actually circles the piece. Or start from the still the gallery already published and accept an estimated shell. The first path records. The second path suggests. They should not share a caption.

The pressure is practical. The work may already be on the plinth, insured for the run of the show, and awkward to deinstall for an afternoon in a studio. A reshoot is not a filter. It is a loan, a handler, and a gap on the wall. That cost is why a catalog still looks tempting. It is also why the temptation needs a rule before anyone treats the result as the object.

What The Catalog Photograph Already Settled

A competent still settles the front. It shows the pose, the break of a patina, the point where an arm leaves the torso, the relation of the base to the figure. It does not settle the back of the head, the inside of a fold that faced away, or the thickness of a wrist. Those are not minor texture problems. They are missing observations.

An opening-night wide shot is a worse source than that still. It includes the plinth, the wall color, a neighboring frame, sometimes a hand. Those shapes will try to become volume. The file to start from is the object photograph made for the checklist, cropped so the bronze is the only subject. If the gallery never made that file, the honest next step is a new still of the object alone, not a mesh built from the party picture.

Any later file has to confess which of those observations it inherited. A commissioned scan inherits the walk around the object, if the studio actually did the walk and wrote down the units. A mesh from one photograph inherits the front and estimates the rest. The estimate can be visually calm. Calm is not a measurement. A registrar who lets a calm underside into the collection record has changed what the gallery claims to know.

There is a second inheritance that is easier to miss. Printed or incised marks, a signature, a foundry stamp, a date on the base. On the photograph they are either legible or they are not. On a generated surface they can become ornament: strokes in the right place, wrong shapes. If a mark is part of why the sheet lists the bronze, the mark has to be read from the object or from a sharp still, not from a surface that only looks inscribed.

Where A Single Photo Mesh Belongs

Image To 3D is the fit when the gallery wants a study model beside the catalog still, and the label stays the label. One clear photograph can produce a mesh you rotate and inspect. The product is explicit that this is a starting point, not a measured scan. Unseen areas are estimated. For a web exhibition, that honesty is the feature. The page can say the form was developed from the installation photograph. It cannot say the bronze was digitized to the millimeter.

The model list on the workspace includes Tripo3D v2.5 for a more finished surface. Its standard texture preset is set at 300,000 faces. The HD PBR preset is set at 700,000, with PBR switched on. A face count is not a measurement. It is a density setting. Raising it can keep a patina from looking muddy. It does not recover the back of a head the camera never saw, and it does not convert a catalog height into a verified dimension. Use the denser preset when the front must stay legible. Do not cite the number in the wall text.

Turn The Object Before The Label Moves

Before the study model is linked from the exhibition page, someone turns it. Not a glance at the angle that matches the photograph. A turn to the back, then to the underside, then to whatever is thin: fingers, a staff, the edge of a drape, the lip of the base. The failure is specific. An arm melts into the hip. A gap opens under the foot. Two fingers fuse. A stamp that was sharp in the still becomes unreadable scratches. Any one of those holds the file.

The hold is editorial, not technical theater. The exhibition page can wait. The label cannot be rewritten to match a guess. If the shell is wrong, the gallery publishes the still alone. That is a smaller page and a truer one. Fixing a fused arm by hand is real modeling time. Spending it on opening week, after the caption has already called the file a scan, is how a correction becomes a notice.

Where A Commissioned Scan Earns The Fee

A studio scan earns its fee when the gallery needs the record, not the tease. Edition size is not the point. The point is a claim. If the page, the checklist, or a borrower's condition report will state a dimension taken from the file, the file has to come from a capture that circled the bronze and kept its units. One photograph cannot supply that capture. Sending the work, or scanning it in the room under the registrar's eye, can.

The fee also buys a negative result the generator will not give you. A studio can say the underside was occluded by the mount and was not recorded. An estimate will usually fill the same underside with something plausible. Plausible is the wrong answer when the mount hid a repair, a seam, or an old bracket. The commissioned path is allowed to have a hole in the data. The exhibition page should prefer a hole to a fiction.

Loan terms often decide this before taste does. A borrowed bronze may not leave the building, and the lender's courier may be the only person allowed to move it. A capture inside the gallery, scheduled with that courier, is still a commissioned record. A mesh made overnight from the press still is not a workaround the lender agreed to. If the condition report will travel back with the work, the file quoted there has to be one the lender could recognize as a record of their object.

The Wall Label Cannot Inherit A Guess

Keep two sentences in the checklist and do not merge them. The label carries the title, the date, the medium, and any dimension the registrar measured or copied from a prior trusted sheet. The web model carries a viewing aid. If a dimension is missing, the missing line stays missing. It is not filled from the bounding box of a mesh. A visitor can enjoy the spin without being told the spin is the document.

This split also decides the vendor. A scanning studio is the vendor when the bronze can leave the plinth, or when the gallery can host the capture without breaking the loan terms. Image To 3D is the path when the still is all the gallery will have until the deinstall, and the page is willing to caption an estimate. Taking the cheaper path and then writing the expensive caption is the actual mistake. The money saved does not repair the sentence.

A Plain Split For The Exhibition Page

The split is easier to apply if it sits in one place the front desk and the web editor both see. It does not need a scoring system. It needs the same questions every time a small bronze, a ceramic, or a study maquette is about to spin online.

Read the table as a gate, not as a ranking. A mesh can be the right public file and still be forbidden as a source of millimeters. A scan can be the right record and still be the wrong file for a light web page until someone exports a compact version and looks at it on the actual screen. GLB is a convenient package for mesh and materials on that page, and it still has to be opened on the device the gallery expects visitors to use. A package that stalls or shifts color has not finished the job.

An image to 3d model belongs in the row where the still is the source and the caption admits the estimate. It does not belong in the row where a borrower's condition report will quote a new height. Those rows can share a visual style on the site. They cannot share a verb. One was modeled from a picture. The other was recorded from the work.

Where This Mesh Still Needs A Human Check

Image To 3D does not certify a printable stand-in. Closed surfaces, enough thickness, and a slicer check still sit between any mesh and material. Scale on the study model is not the scale on the label. Pixal3D shows in the model list with a note that the provider is not configured yet, so a show cannot plan on it. Until a person has turned the back, the catalog still is the record.

Publish The Study Model Beside The Label

Put a turnable study next to the catalog still only when the wall label stays untouched. Image To 3D is the file for that limited job. The photograph stays the evidence of the front. The mesh stays a reading of that photograph.

A commissioned scan remains the right spend when the object can be circled and the institution intends to store a dimension. Deinstalling a bronze to avoid that sentence is wasted handling. Publishing an estimated back as if it were the circle is the larger waste, because the page will keep saying it after the show closes.

For the ordinary online exhibition, publish the study model when the back has been looked at and the caption names it as an estimate. If the back fails, or the label would have to change, keep the still. The work is already in the room. The page does not need to invent the side the room can still show in person.