The procedure for handling photographs from unknown cameras is:
• Review the photographs to ensure they are suited to PhotoModeler use. See the Inverse Camera section above for photograph guidelines. Also review the overview steps below in Inverse Camera Review Steps.
• If they are print or physical, scan the photographs making sure they are as square to the scanner as possible. Scan them at sufficient resolution to see your details (usually a monochrome image file size over 1MB is a good starting point).
• For digital images, you will import them directly.
• Start a new PhotoModeler project (usually Manually Marked Project).
• Load the photo(s).
• Create a new Camera using the "An unknown camera / solved by Control Points”
• Note that imported photographs assigned to a camera with the "Inverse Camera" flag will automatically have their "solve for focal length" property set (see Photo Properties).
• Each photo (if more than one imported) will be assigned its own Camera (with an inverse camera flag set).
• Since the Inverse Camera parameters are solved on a per photo basis, any photo with an Inverse Camera flag set will ensure the camera is unique and only assigned to it.
• Mark your Control Points or constrained Edges, Lines and Surfaces.
• Process the project (Inverse Camera processing occurs only during the Orientation stage of processing).
• Review the status of the Inverse Camera processing in the Processing Finished Dialogs.
• Review the results to ensure they are accurate and consistent. Use the Point Table to check the largest residuals (should be as small as possible and usually under 5 pixels). Use the control projections on the photo to see how well the result matches. Use the Post-Processing Report to check for high parameter correlations.
• Proceed with the remainder of the project as normal, do the required marking, referencing and then processing to get the result.
The Inverse Camera process is carried out automatically by Processing if required. All the user needs to do is make sure sufficient control points are marked and that the appropriate Inverse Camera flags are set for the photograph in Photo Properties.
If the processing fails very quickly when doing Inverse Camera the control points are immediately suspect. Check you control points:
• to make sure the inputted x,y,z locations are correct (even if the surveyor that gave you the points said they are fine make sure yourself!)
• to make sure the correct control points are marked on the correct points on the photographs.
• to make sure the starting parameters for the Camera or Cameras are close to reality. Sometimes when they are very far off (i.e. giving a first approximation of 35mm focal length when the photograph was really taken by a 150mm lens) the Inverse Camera part of processing can fail.
Note: If you want to reprocess just one Photograph for Inverse Camera purposes (perhaps you have moved or changed the control points on the photo and you want to see how it affects the camera solution), you need to reorient just the one photograph. See the Force Unoriented button in Photo Properties or a way to do this.