Ask Slashdot: Tips On 2D To Stereo 3D Conversion?

An anonymous reader writes "I'm interested in converting 2D video to Stereoscopic 3D video — the Red/Cyan Anaglyph type in particular (to ensure compatibility with cardboard Anaglyph glasses). Here's my questions: Which software(s) or algorithms can currently do this, and do it well? Also, are there any 3D TVs on the market that have a high quality 2D-to-3D realtime conversion function in them? And finally, if I were to try and roll my own 2D-to-3D conversion algorithm, where should I start? Which books, websites, blogs or papers should I look at?" I'd never even thought about this as a possibility; now I see there are some tutorials available; if you've done it, though, what sort of results did you get? And any tips for those using Linux?

Comments

Here's a tip

[Hatta]

Don't do it.

Re:Here's a tip

[spidercoz]

Exactly. People with the proper equipment and money fail at this regularly.

Re:Here's a tip

[vlm]

Exactly. People with the proper equipment and money fail at this regularly.

I think it is important to arrive at the correct mindset. This has never stopped people from snapping pix at weddings and sporting events and tourist traps, even if their pix look like garbage compared to a pro photo on a postcard or whatever.

If you want to do it for fun, heck yes go for it. Go Go Go. You don't need help just try it.

If you think you'll turn out something that means anything to anyone else in the world, you'll probably be disappointed. Insert stereotype of goans when someone wants to show you old fashioned slides of their vacation. Although that old tech is getting kind of retro cool now.

Re:Here's a tip

[gfxguy]

You're in luck... Peter Jackson is pushing 48fps over 24 in the cinemas, stating enough digital projectors are capable. He's shooting the Hobbit at 48fps, and shooting it in 3D from the get-go. I'm more interested in the content of the movie, but I'm expecting it'll be one of the best, if not THE best, attempts at 3D so far (Jackson Explains "Hobbit" 48FPS Shooting).

He's trying to encourage future film productions to step up to 48, too.

Re:Here's a tip

[Jappus]

Don't do it.

Well, there is a way to do it, a very elegant way even. One that can be, for all purposes and intents, as good as you can get with the raw material; even to the point where the average human will not be able to tell the difference.

The thing is: That solution has a big catch. How big? Well, to put it mildly, you will most likely win the Turing Award in the process of doing so and will at some point end up with a Nobel Prize in your hand, too. As you can imagine, the solution is: Artificial Intelligence; and if you want to really do it, only strong artificial intelligence will do.

The fact is, as others have quite succinctly pointed out, that the issue is in determining what is "in front" and what is "in the background" on top of how far away everything is. This is, quite simply, impossible to do right if you approach it as a purely algorithmic picture-to-picture problem. There is just not enough information inside the frames/movie to do it well enough even at the best of times.

So, what do you do? Easy, you import external information. Things like: "This is a tree; That is a human. A tree is bigger than a human. Both take up the same space in the picture. Assumption: The human is closer than the tree. Proof: The tree casts a shadow on the human and the only light source is behind the tree. Angles point to a distance of 20 meters between human and tree. Etc. pp."

This line of reasoning imports lots of information from the outside; essential things like "What does a tree/human look like?", and "What are their relations to each other size-wise?". But if you grant that this information can be derived and used by an AI, the result can be a very precise derivation of the distances between objects.

It is exactly the same line of reasoning the human brain uses for large distances (where the parallax of your eyes is too small, focus is unimportant and difference between eye positions negligible), or when you have lost vision in one eye (or just plainly covered it). Even though your brain suddenly has only half the information, it is capable of giving you a good feeling for distance and depth.

Of course, it doesn't always work, as far too many optical illusions like the Ames-Room show, but it works significantly better than a "pure" picture-to-picture approach and is the sole reason why almost everyone here feels that 2D-3D conversions are so horrible:

Their brain tells them, that what they see just can't be correct, even if their eyes have actually seen it.

But of course, just using 2 cameras is much simpler. So good luck with (strong) AI. I would be surprised if you solved this issue all by yourself. :)

This sounds familliar

[localman57]

I'm interested in converting 2D video to Stereoscopic 3D video

George Lucas, is that you?

Special FX

[damaki]

A friend of mine used to work for a French special effects company and he had to work on this. He told me that this is basically a world of pain and it produces great piles of smocking shit. It just sucks, even when done properly by highly trained people. Can you imagine making 3D out of a 2D tree? Make every background 3D or properly cut out the character to get the desired effect?
It sucks, it's mostly manual, get over it.

this can be done easily with ffmpeg and imagemagic

[nitrofurano]

this can be done easily with ffmpeg and imagemagick - you need two video sources, and from a ffmpeg script, extracting a picture sequence from both videos, one sequence from the left camera, and another from the right - with a bash script using imagemagick you will separate the colour channels from each frame: red from one camera, and green/blue from another - and having the separation done, you will join with imagemagick again the red channel picture frame from one and green/blue from another, into a new picture sequence, and when you have this sequence ready, you convert it into video again with ffmpeg - try googling for ffmpeg and imagemagick instruction arguments when coding this bash script

Creating "3d"

[Maximum Prophet]

There was a recent NOVA episode about aerial photo reconnaissance during WWII. To make stereoscopic images, they'd fly the plane straight and level over the target. If they could take multiple pictures with 60% overlap, they could use two adjacent images to make one stereoscopic image that was good enough to tell a ship from a decoy.

Any motion picture where the camera pans side to side gives an opportunity to create a "3d" image. If an object moves across a still camera, you can also derive 3d information. (Also if it spins)

An interesting exercise would be to process a film, and make stereoscopic only what what can be done properly, and leave the rest flat. A scene would start out flat, then people and things would begin to jump out at you.