| Author | Message | ||
     Dave Rating: N/A |
Can someone explain the theory behind 2D conversion? I have converted avi films to 3D but I get depth yet nothing leaps out as 3D whereas I have seen pictures converted to 3D that really stand out and would like to apply this to my avi`s. I can`t seem to get anything to stand out. | ||
| clyde Rating: N/A |
Dave you can search this board ..most answers are there. To get started on your quest for "pseudo 3d" you can download the software called 3DCombine or buy the hardware VFX3D. Regards Clyde | ||
| Anonymous Rating: N/A |
The VFX3D is old technology and very poor quality. | ||
| Peter Wimmer Rating: N/A |
2D->3D conversion is as silly as black/white->color conversion... | ||
| Dave Rating: N/A |
I`ve searched the board but not found what I need. Specifically why it is that some objects in a 2D converted to 3D image really stand out? As I understand it pseudo 3D is just creating a fake second image and generating displaced viewpoints based on these two created 2D images. I`m applying this process to avi to dvd conversion via avisynth but while I get great depth nothing is looking that 3D. | ||
| clyde Rating: N/A |
Hmm Dave... You will have to try a "power google search" on this board, menwhile to kick start you... The easiest way to get Somewhat "satisfactory" 3D from 2d content is in the case of whats called the Pulfrich effect.. Download software like StereoMoviemaker, then take a handycam and stick it out of the passenger side of a car window. Start taping as the car is driving. Load two copies of the grabbed avi from your cam into the software and delay the Start frame of one of the copies of the vidoe clip. You will get "respected" and good 3d effect. This same technique applies to ANY kind of footage you get where there is camera panning. You can extract the 1st frams then the 3rd frame for instance in such a clip and you will have a reasonable stereo-pair. (Google the pulfrich 3d effect for an easy to understand explaination of this). Now what happens if you dont have panning footage? Well *YOU* will have to look at a scene and then determine what should be in front and what should be in the back. You do this by creating a "depth map" of all objects in a scene. (the user guide from 3Dcombine software will tell you how) Ofcourse there is an option for the software to start with a single depthmap created by you and then "track" the objects in successive frames of the video.. but as u would image this wont work for long, any camer cuts to a new scene will cause havok, even managing a steady camera scene is a challenge. Companies like DDD use a hybrid software-human approach to create reasonable 3d. LIKE EVERYONE ELSE WILL SAY AS WILL I, THERE IS NO SUBSTITUTE FOR REAL STEREOSCOPY FILMING< AND CONVERSIONS IN FACT MAY CAUSE IR-REPAIRABLE DAMAGE TO STEREOSCOPY ITSELF. But commercially speaking, if you do put in dedicated effort, you can create some good masterpieces on a case to case basis. I myself create regular 2d - 3d stuff using a slew of software tools and hardware. What i WONT DO is simply offset an image "inside" a monitor or "out of the monitor plane" to pass it off as 3D. Hope this helps a bit Cheers Clyde PS. @ anonymous: the algorithms at work in the VFX box shouldnt be taken lightly, unless you personally know whats actually going on inside it. Its got multiple intelligent algorithms and processes at work all burned on a chip and doing its thing in real time. -- its not true 3d but its the best you'll get IMO for automated conversion. I would Ideally pass footage thru it, get the converted stuff out and then MANUALLY , frame by frame edit cheesy looking segments.  PPS .. no im not gonna be dragged into yet another debate of 2d-3d conversions .. been there done that.. tired of typing!! :D | ||
| clyde Rating: N/A |
Actually come to think of it, I did have an idea floating in my head for Black/White to color conversion. (im sure its already the basis for the current process anyways but still...) What if you assigned a Color value to every Luma pixel in a B/W image (assuming you have 16million color shades of gray sampling).. wont you get a good color representation? Im sure thats how it's done right? --Clyde | ||
| Dave Rating: N/A |
Ah, that`s what I`m trying to do - convert a flat 2D avi to 3D. The problem is that without manual intervention I don`t seem able to make anything stand out, I`ve heard of the Dubois algorithm and though of implementing that via an avisynth script but can`t find any info on what it is or how it works. I tried overlaying one frame over another but that didn`t work so was toying with other ideas. Thanks for the heads up | ||
| Anonymous Rating: N/A |
VDUB Filter DeinterlacePALMovie can add field delay between frames to introduce a Pulfrich 3d effect in an interlaced format. That's about as good as it gets for 2D to 3D without slavishly converting one frame at a time through depth mapping etc. | ||
| Dave Rating: N/A |
Hmm, might have to try that - would that still work on non interlaced material? I don`t mind depth maps but I can`t seem to find any way of generating them automatically even though I`ve seen software that says it can. Think I`ll try the vdub suggestion as I can always import the filter into an avisynth file. Thanks | ||
| Anonymous Rating: N/A |
Yes it the vdub filter will work with non interlaced videos and Avisynth |
