Match Photo with cropped photos - what are SU Match Photo's basic assumptions?


One one hand I have precise architectural drawings of rooms that I modeled in 3D with SU. On the other, I have antique pictures of these rooms and would like to populate the 3D model with the original furniture as shown in the photos. I use Photo Match to orient the 3D-model with the corresponding photo. Of course, I have no idea whether these photos have been cropped or not. Am encountering several problems, in particular with scaling. Am reading many posts to the effect “Do not crop photos. Match Photo currently requires that the point you aimed the camera at is located in the center of the image (also called the center of projection). Although it may seem possible to use a cropped image, typically vertical lines will not align well across a cropped image and the results will be unsatisfactory.” A contributor recommends an “un-cropping” of the photos so that the horizon passing through the L, R vanishing points is in the vertical centre of the image, Is this procedure correct?
More generally, what are SU Match Photo’s requirements to deliver good results. Is there a way to un-crop the photos to avoid the above mentioned problems?


If you can locate the optical center point of the image it would perhaps be possible to add empty space into the file in Photoshop so that the point is returned to the middle point. Another problem you might encounter is that many old professional architectural photographs use perspective corrections (tilt or shift) done either in the camera or in the darkroom that also would make them invalid for use with PhotoMatch.



Thank you very much for your reply. I do not know that these photos were taken with specialized cameras, as some of them were published at the end of the 19th /begin of the 20th century in non-architectural magazines, which leads me to think that they may have been cropped either for lay-outing purposes or to mask unwanted details (please see attached reduced-size photo where I hilited in red two lateral crop indicators

How can I locate the optical center point of the image, and is it all possible? Having dealt with photogrammetry in the past, it is my understanding that SU Photo Match is extremely powerful in its simplicity and bases of simplifying assumptions. I thought that the combination of the 3D model of the room and the photograph would help me here. Apparently I am reaching the limits of SU Photo Match…


If, as it seems, you have access to the full frame, I don’t think cropping is the problem. The room seems rectangular enough, but maybe the lower part has relatively few axis direction indicators (unless we assume that the worlbenches are in a straight enough row. Basically this seems quite doable.