Miniature Rendering

Hi,

I was wondering if anyone had some insight on how to make models look miniature through the use of DOF in Blender? The Lego movie pulled it off well.

Is it just a matter of modeling to real world units and setting up with a camera in Blender? Or is it more complicated than that?

Any help or articles would be appreciated :slight_smile:

The miniature effect on large objects is achieved by faking a shallow depth of field like you would get with a macro lens photographing miniatures.

Even real world footage can be made to look miniature by faking a shallow DOF (Google “tilt shift photography”).


Thanks! Yeah, I work in compositing and understand the tilt shift photography masking effect. However I’m wondering if there is way to get a more authentic look throughout?

How would I go about achieving a physically accurate camera that I could point at a model. So pointing a blender camera at my 4 inch models and getting realistic results in rendering of what it’d look like if I was actually shooting a 4 inch model.

Since the strength of the 3D dof isn’t actually tied to anything, you could just massively overcrank it and get an unrealisticly shallow lens.

Yep this. It’ll probably take some experimentation to get the DOF right.

If you track your camera to an empty - you can also set your camera to focus on that point. That way you have infinite control over where your camera is pointing and what it’s focusing on - just by moving an empty set of axes around.

Another trick I have seen that may give your scene a more authentic miniature look is to over saturate your colours slightly in compositor. Miniatures tend to be painted a little more garishly than your typical real scene. I think this trick has been used in the photo above looking at how vibrant the colours are.

Wow! Thanks for all the info and help. I’ll experiment with that!

I used Blender for some of the VFX in this “Miniature Rendering” of some footage, including moving shots:

Using Syntheyes, I created a virtual camera and virtual geometry. You can also do this in Blender; I just prefer Syntheyes.

I brought this camera / geometry scene into Blender to render the Z-Depth for the shot. Then I took it the depth into After Effects and created the DOF blur with the Frischluft lens filter. But, again, you can do this in Blender if you prefer.

Some of my favorite bits from the spot are at 0:18 and 0:47

Basically there are two simple techniques based on Camera Aperture Type:

  1. F-stop:
  • Camera is 1:1
  • Control/manipulate DOF: by scene scaling
    Example: diorama > small scale = shallow DOF
  1. Radius:
  • Scene Scale is 1:1
  • Control/Manipulate DOF: by using lens dimension (camera scaling)
    Example: 0 = pinhole … higher radius = shallow DOF = bigger ‘camera’ (room)

I hope it’s simple enough… Scale scene or scale lens
:smiley: Basics of any photography lesson i have heard of…
Still, this part is missing from manual:
https://www.blender.org/manual/render/camera/introduction.html#depth-of-field

Slightly relevant question: What is the best approach to Anamorphic bokeh (preferably in post)? A squeezed bokeh image alone doesn’t seem to do the trick. I tried squeezing prior to applying and then stretching back. It kind of works on a 1.33:1 ratio at least if bokeh is kept small, but fails miserably on the full 2:1 deal. I think I might have better luck if I actually rendered in expanded format first, but I didn’t. So basically, what’s a good way of doing this, in post?

Create an image to use as Bokeh form:

Use as Bokeh in Bokeh Blur (Compositing Filter Node).
Example:

Or? At least it looked something alike from video posted :confused:…
Otherwise just make a specific thread :wink:
Enjoy

I love this community, so helpful!

Benu, great video, congrats! I’ll look into the geometry approach.

Burnin, thanks for explaining! Geez, that was so hard to find online haha Thanks for explaining the bokeh effect as well.

So. Aside from scaling and using F-stop, what are some other miniature characteristics? Or does it boil down to texturing after that point. Are there common lens focal lengths or anything useful like that?

Exaggerated colors, some element of time lapse (speeding up the movement a bit, at any rate) – the overall impression I’ve tried to give in these sorts of things is a feeling that what we’re looking at is in fact a miniature. Or, in other words, that what we’re looking isn’t real, but some sort of small model that didn’t quite get it right, but got impressively close. Color and speed are two aspects that can contribute a lot to this look.

If you’re working with live footage, then wider shots tend to work better, in my opinion. Close-ups, especially with longer lenses, tend to lose the effect.