[fwwidall]

For the past few days I've been playing around with creating 3D images from 2D images using depth maps created in Google Colab and then processed in Stereophoto Maker.

I'm certainly no expert at this, but I'm quite impressed with the results so far. I'm sure there's plenty more to learn about both pieces of software.

You need a pair of red/cyan glasses to enjoy these images.

Hope you like them

[IMG][/IMG]

[Bob 256]

Great work.

I'd like to see what these look like on a 3D monitor without the anaglyph process (which leaves something to be desired in viewing them), but from their appearances here, they should be outstanding for simulated 3D. If you haven't already done so, be sure to convert these to an mpo file (with Stereophoto Maker) and look at them on a 3D display (many of which will take the mpo format).

Depth mapping was a fairly recent addition to Stereophotomaker and you can do something similar with Photoshop by using a map with the "displace" filter. You can create one image using the depth map for displacement (horizontal only) and then invert the map for the opposite eye. You have to tweak the parameters for the best effect (start with a 2% horizontal displacement for each). Photoshop doesn't do any anaglyph conversions, however, so it would be back to Stereophoto Maker for that. Of course, the key is getting a good depth map to begin with and it appears you did that with your photos.

[fwwidall]

Thanks Bob,

I have played in the past with manually creating a depth map and then using it to displace images in Photoshop. It works ok, but can be tedious to do, and the results 'iffy'.

Running this '2dtodepth' program inside Google Colaboratory does all the work for you and seems to give excellent results. You can open the output in SPM then process it as desired.

I'll attach a sample of the output.

[swanlefitte]

I played around with anaglyphs a while back trying to figure out what they work well for. That first one would be cool in a comic strip peeling off the page. The last one really gives a sense of the space you would get being there. I bet the sports ones would work really well if you could separate each person first and render them separately.
I have 2 traditional 2 picture anaglyphs in this thread while you have the glasses out. post 192 and 203
Stupid Camera Tricks (and Stupid Lens Tricks) - Page 13 - PentaxForums.com

[fwwidall]

Nice lens collection

I've played around with the two camera approach myself. I hooked up two *istDS's and fired them with a wireless trigger. Sometimes it worked



Processing a single image makes for interesting possibilities - how about a 3D snowflake. No easy way to create that with two cameras.



It really brings out the details of these incredible structures

[Bob 256]

Thanks Bob,

I have played in the past with manually creating a depth map and then using it to displace images in Photoshop. It works ok, but can be tedious to do, and the results 'iffy'.

Running this '2dtodepth' program inside Google Colaboratory does all the work for you and seems to give excellent results. You can open the output in SPM then process it as desired.

I'll attach a sample of the output.

I didn't mean to say the photoshop method is better, just that it's there as you know (and now I know, you know). Your method is much better and takes the pain out of getting a depth map, so one more ata boy (or ata girl) to you for doing it that way and sharing. I haven't tried the Google Colaboratory yet, but I sure will now. Ordinarily though, I shoot in 3D when I want 3D results, and would do the 3D conversion for those shots that I only have 2D info for (and there are plenty of those cases). Your post is really welcomed in that respect. Thanks for the info.