I would like to see the Photomatch tool improved in the following ways.
Add a default option on the Photomatch dialog to adjust aspect ratio of view to match that of the photo being used. (Making sure the resulting view dimensions are not the result of rounding to the nearest whole number, avoiding fractional pixels. This will ensure that renders of the view will exactly match the aspect ratio of the original photo.)
Set the Active Background to use the photo automatically.(If possible, eliminate the need to use a raster reference!)
Allow the Photomatch dialog to be resized!
For fine control while picking the image point, allow the point to be picked directly in the Photomatch dialog rather than the view window. This could be achieved by use of a hotkey to toggle from view window to Photomatch dialog. This would allow much finer control of point selection!
Add a Remove Point button to allow the removal of any previously accepted point.
Add a button to save the result as a new saved view.
Add the ability to store the Active Background info (filename and path) in the saved view. This would allow the appropriate photo to automatically be set as the active background when a saved view is applied to a view window.
If a raster reference must be used (See note 2 above), detach the raster reference when the clear button is clicked.
When the Clear button is clicked ask the user if they want to save the view first.
As a side note Reality models are not replacement for photo match as not all scenes can be captures using reality meshes. Photomatch is a must and modernized.
I'm very disappointed to hear that this idea is being dismissed.
This idea was liked by at least 22 users, so there was definitely interest in improving the command.
Ron Jones over 3 years ago
One more item to add:
It would also be great if the actual original photo was used in the Photomatch dialog, rather than the screen pixels of the view being used for the photomatch.
Sometimes objects that are easily seen in the original image are indiscernible in the view window because of the reduced resolution of the view. If you can't see it in the view pixels you can't pick it for the photomatch. If the original image was used you could easily pick those small details.
Dominic Seah over 2 years ago
http://www.phocad.de/en/en.html
Robert Jones 8 months ago
I have developed a manual workflow to photomatch using edges and vanishing points because the method in the MicroStation's photomatch tool requires known matching points in the model AND in the photo that might not be available in both.
I.e. While the Bentley method works well enough if you have an accurate model, and a clear photo, the tool tends to get in the way when you have less accurate data, or an obscured photo, and you just end up trying to eyeball things.
My method relies on use 2 or more lines per vanishing point on the photo that relate to a corresponding direction (edge) in the model. I only need 2 matching points on the photo and model to scale the image.
From this I can calculate the position of the camera, with either;
2 vanishing points and the 3D angle of the corresponding edges on the model, or
3 vanishing points and the corresponding edges.
Because I am manipulating lines on the photo to calculate the vanishing points, I can easily align these even when the corners of my target are obscured by tress or foreground.
The vanishing point does not need to be on the photo, and slight barrelling of the photo can be allowed for by averaging the line by eye.
I also prefer not to use the raster manager for the image backplate, and instead prefer the view background - I wish this was saved as part of the saved view setting, rather than a file preference.
This does however require careful view-port aspect ratio management, so having a lock on the "output" aspect ratio would be a great help generally for all kind of rendering.
However because the photo aspect ratio may not be my output ratio, I have also developed methods for adding the original photo as a physical, camera facing, backplate & foreground masks within the model.
Obviously for this I need to set up textures, but the advantage is that I can use tools directly in MicroStation not as a photoshop post-process.
These clipped out billboard shapes remain in the correct perspective/position for my camera shot, because they are not tied to the view, but instead to the cameras position relative to the model. This means I can change my Aspect ratio, Camera Zoom and Rotation (but not position) to frame the shot as I wish and everything stays in perspective.