I'm trying to re-render the view from a photo -

I hopped to generate some jewelery out of the rendered view. So I found some DEM-data (DEM=digital elevation model) of the region and added a camera and a plane to mirror the mountains there.

Here's the blend file of the scene I created so far

I just can't seem to recreate the view from the photo, or even something close to it. I first tried with just the plane with a displacement modifier on the UV-data and tried to move the camera in to the region, but this didn't came to fruition. After some other things I tried I can kind of recreate the view with the one side of the UV-map (of the plane) being scaled to zero - trying to mirror the view from the camera. (you also see this in the blender file, this is where the shot was approximately made)

I was wondering if someone did something similar to this and might help me.

The most straight forward thing I came up with was that I massively scaled the plane up and tried to physically move the camera in to the region, but this just didn't work. The blender cycles camera just skewed the view too much, the angles and everything seemed off...

edit: To be clear: my problem doesn't lie wit generating the mesh with displacement modifiers, subsurfing, etc. But more with how to then make the camera see more or less the same as in the photo

  • $\begingroup$ Make sure the aspect ratio and lense simulation is the same as was present in your camera. $\endgroup$
    – J Sargent
    May 23, 2015 at 16:11
  • $\begingroup$ duplicate?: blender.stackexchange.com/questions/27451/… $\endgroup$
    – user1853
    May 23, 2015 at 17:02
  • $\begingroup$ Do you mean something like using the photo as an semi-transparent reference in your camera view? $\endgroup$
    – Bithur
    Jul 24, 2015 at 8:48
  • $\begingroup$ If you mean view from camera looking somehow like this, then you can accomplish that by using Lens Distortion node (of course, camera itself is tilted slightly) $\endgroup$
    – Mr Zak
    Aug 24, 2015 at 22:07

2 Answers 2


Start with a plane. Subdivide it like 100 times. In Top ortho view unwrap it using project from view bounds.

enter image description here

to the plane add a displace modifier that uses your hieght map as texture. Make sure the texture coordinates are set to UV.

For a smoother surface add a subsurf modifier.

From there just set the camera where you want.


  • $\begingroup$ My problem isn't really with these steps - I managed to do so too - that was what I meant with "first tried with just the plane with a displacement modifier on the UV-data and tried to move the camera in to the region" My problem is with the next step - how to get the camera to render the same view as on the image. - But thanks for trying to answer anyways. $\endgroup$
    – Deadolus
    May 25, 2015 at 7:52

If you're trying to do (render a scene that perfectly mimics the real-life photo) what I think you're trying to do, then there's a lot more involved than what we'd reasonably be able to write here. But there are a few steps that can get you closer.

  • A good place to start is Blender's 2D Camera Tracking documentation. You'll want to look at the metadata for the photograph, or otherwise learn a bit about your camera. There are a lot of parameters that go into making a photo look the way it does.
  • It looks like your camera used an 'HDR' setting to get the image's colors and contrast to look so stellar. You may have some difficulty making this happen in blender, but it can be done.
  • As for the orientation and location of your camera, you're right to start by blowing the mesh up considerably (it will help make the light and shadows more realistic). Moreover, the world is really really big. It may take a lot of scaling before your geometry even starts to approach the correct scale. Whenever I'm having difficulty with scale (it's easy to get 'lost' in Blender-units) I add a roughly human-shaped object (normally just a tapered cylinder with an icosphere head) and put it in my scene. Then I scale objects until I could see them being the right height. (It's hard here because mountains are big... But it'll help you when you're orienting the camera. It'll probably go a lot closer to the mesh — that is, 'lower' to earth — than you're imagining.)
  • Before you even begin with textures and colors, though, you need to get the geometry right. I'd suggest making this image your camera's background reference image and using this to establish how to position your camera to get the right image.
  • Note that your heightmap may not be scaled properly in the z-direction: Once you get the camera in the right position, I often find that my mountains look a bit mole-hilly. sz to scale in the z-direction only; with a camera at 'eye level' this may give you a better-looking middle-distance.
  • Use clues in the photo to orient your camera. Clearly the camera is pointed at a certain set of mountains from a certain direction, so start by putting your camera in the right geographical location. Point it in the right direction. Now look at your horizon in the photograph — it's definitely angled! Orient your camera so that the xy plane is at the same angle, and your image will start to come together.

My parting suggestion is to play around with the focal length: I find a much lower angle will make the environment appear larger (with the requirement that you move the camera 'back' to maintain the contents of the frame). Play around with that slider and see the results; I think that may be why you're not getting the right 'feel', even though the geometry is all there.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .