How can I control foreground and background depth in an anaglyph 3D image?

Asked 1/11/2014

2 views

2 answers

0

I'm creating red/cyan anaglyphs from a single 2D image in Paint.NET. I can make the image appear 3D by tinting and offsetting the channels, but the result looks like a flat scene viewed through a window, with little separation between the subject and the background.

How do you create stronger depth so that foreground and background objects appear at different distances? If that means shifting the red/cyan layers by different amounts for different parts of the image, what is the basic workflow? A general explanation is fine, and Paint.NET-equivalent tools would be helpful.

Originally by Photography Stack Exchange contributor. Source · Licensed CC BY-SA 4.0

Photography Stack Exchange contributor

12y ago

2 Answers

1

You're spot-on with the need to shift some parts of the red/cyan layers either more or less in order to achieve the control of the separation that you're after.

One way to do it, if we're assuming that you're starting from a single 2D image, is to isolate different elements of the image onto their own layers, such as the background on one and the lamp on another. This might require some careful selection, and even some "mending" (clone/heal in Photoshop) of the background in the areas which are "behind" the lamp.

Once you have the two parts, do the red/cyan separation as you did before, but separately for each part. Now adjust the background by only a small amount, and the lamp by a larger amount. Or you could leave the background alone and just do the lamp, or even adjust them in opposite directions - some trial and error is necessary here.

You will now have a composite image where the lamp stands out from the background, which is one step forward. However, if you look at the floor under the base of the lamp it is still "flat", i.e. it feel like it's all at the same distance from you. This requires another red/cyan pair with just that part of the image, but this time you'll have to distort the layers rather than simply shift them, such that the part of the floor which is closer to the viewer moves further apart than the part which is at the base. I'm not sure what tools you have in Paint.NET, but in Photoshop I would again use clone/heal to fill in more of the areas of the floor "behind" the lamp's base.

[edit] If you want to get really adventurous, you can use tools such as Liquify to fine-tune the degree of shift between the red and cyan layers; that way, you can adjust more complex elements, such as the round part of the lamp, and bring the front of it closer and the back of it further away. Again, this is most easily achieved with a separate pair of layers for just that area, so you can restore information "behind" it on other layers.

It also helps if you group the pairs of layers so you can turn them on and off easily. That's so you can work just on the element of your image that you're interested in.

Originally by user27132. Source · Licensed CC BY-SA 4.0

user27132

12y ago

0

AI Answer

Generated from our catalog & community — verify before relying on it.

Yes—if you’re starting from one 2D image, stronger depth comes from giving different parts of the scene different amounts of horizontal red/cyan separation.

Basic workflow:

  1. Separate the image into elements on different layers, such as subject/foreground and background.
  2. Clean up the background where the foreground used to be, if needed, using clone/heal-style tools.
  3. Create the red/cyan anaglyph for each layer separately.
  4. Offset each layer by a different amount: use a small shift for background and a larger shift for the foreground. You can also leave one layer nearly unchanged and shift the other.

The greater the horizontal separation, the stronger the perceived depth. By controlling the offset per layer, you can make the lamp appear in front of the background instead of everything sitting on the same plane.

So your idea is correct: don’t shift the whole image equally. Mask/select objects, put them on separate layers, and apply different offsets to each layer.

UniqueBot

AI

12y ago

Your Answer