Open the image in photoshop, go to filter \ distort \ lens correction and use the vertical perspective to straighten the vertical lines.
To create a camera match in Max, we are going to take some measurements with photoshop using the vanishing point tool.
Go to filters> vanishing Point. Create two planes for the house on the left (front and side), then you can start measuring.
Note: We will need to use a virtual viewport in Max and to do that 3ds max needs to run in openGL . You can change the driver in the start menu> autodesk> 3ds Max> change graphics mode, or inside 3ds Max, Customize> preferences, and in the viewports Tab click in “choose driver”. Restart Max.
------------------------------------ In 3ds Max, press “8” to open the environment window. Click in the environment map slot, choose bitmap and browse for lens corrected image. With the perspective viewport selected press Alt+B to open the Viewport Background dialog box, activate “use environment background” checkbox . activate “display background” , Click OK.
Press F10 to open the “Render Setup” dialog box, and change the width and height to match the image file properties.
Creating the camera:
-With the measures taken in photoshop, we can create a box in max that will help us with the camera match process. Create a box with: length= 490cm width= 900cm height= 625 cm.
Using snap to vertex, create camera point helpers on the visible corners of the box : create>helpers>camera point. This six camera points will give us the camera angle and lens we will work with. To create that camera, go to the utilities panel and click in “camera match”, now assign a position in the image background corresponding to the selected camera point. When done, click “create camera”. Apply a camera correction modifier to the camera if needed, to make small perspective corrections.
Blocking the scene:
On the top viewport, create a ground plane and another box for the structure on the right (length=1000cm \ 3 segs; width= 270cm \ 2 segs; height 195cm \ 1 segs).
Apply an Edit Poly modifier to the last box you have created, select the top> center polygons and extrude them.
Select the top > center edges, and move them up to shape the roof.
Select the roof polygons and extrude them using local normals.
Create a few more extrusions to create the roof, there is no need to create too much detail. At this point, you may want to zoom the camera view to make some adjustments so your modeling matches the photo. To do that you need to use the virtual viewport (you can only access the virtual viewport by using OpenGL, doesn’t work with direct 3D or Nitrous). To use the virtual viewport, select the camera view and go to views> viewport configuration, in the regions Tab you can activate the virtual viewport and change zoom settings.
Adjust the vertices to shape the left building.
add some detail to the scene by adding more objects.
Create a box for the chimney (adjust by moving vertices) Repeat the process to add the detail and depth to some objects in the scene.
For the debris on the right of our scene, I used planes and shaped them by moving vertices.
The terrain it’s not flat, make some adjustments to the ground plane and left building height to follow the terrain angle.
Create a sphere (the use of hemisphere is not mandatory), for the remaining environment and apply a normal modifier.
The material is quite simple. With standard material, turn self illumination to 100%, apply a “camera map per pixel” map to the diffuse slot. Choose the camera we created with camera match and the image we used as background. Apply this material to all objects in the scene.
Select the camera, click with the right mouse button and clone (choose copy, not instance). Select the new camera, activate “auto key”, move the time slider to the 100th frame and move the camera to the final position. There are limitations to the camera movement due to the texture distortion. Make some test renders to check the distortion.
Render the animation!
If you render by layers you can have more control over the final shot. Don’t save your render to a movie file, save it as an image sequence. It’s better to edit and can save you lots of time if something goes wrong with the render. Keep the original aspect ratio of the photo when rendering.