by VictorBorun » Tue Aug 23, 2022 10:33 pm
Chris Peterson wrote: ↑Wed Aug 10, 2022 1:31 pm
johnnydeep wrote: ↑Wed Aug 10, 2022 1:07 pm
SeedsofEarfth wrote: ↑Tue Aug 09, 2022 10:28 pm
One 3-D technology (aside from Facebook) that used to be quite effective in creating a real 3-d effect without was by polarising the light from the subject using two lenses that polarised at different angles so that each eye saw a slightly different image in each eye. Melded together by the brain, the depth perception was remarkable. I remember using those polarising glasses in movie theaters back in the 1950s and 1960s. As for Facebook:
How do I create a 3D photo on Facebook?
Tap "What's on your mind?" at the top of your Feed or tap "Write something"... in a group.
Tap Photo/Video.
Select your photo and tap Done.
Tap Effects.
On the Choose Effect panel, select 3D.
Add any text you'd like, then tap Post.
However, I'm not sure about the actual process that converts a 2-D image into a 3-D image.
I see no 3D option when posting a picture with the Facebook website on my PC. Perhaps it's only available on the Facebook app? Oh well, I detest Facebook anyway.
We would not want APOD images displayed like this. The image is analyzed, and based on what the AI thinks is going on, a depth map is created, and then different views are simulated, with missing background estimated from the surrounding image. An animation is then created. Sometimes it works pretty well. Sometimes the AI fails miserably and you get truly bizarre artifacts.
Starting with true 3D data (e.g. an image pair) you could create a similar animation. I think I've seen software that would do that. But it's up to an image author to decide if they want to do that, and up to the editors if they think it would make a good APOD. In scientific publications, however, the standard way of presenting stereo data is with a pair of images, viewed either directly or using a pair of lenses (like a stereoscope). Occasionally anaglyphs like this APOD are used, but not commonly, since it requires special glasses and you lose color information.
A correction: to create a 3D photo on Facebook, you supply a pair of images, one a 2d colour image, another a
depth map monochrome image.
It's only after you upload the pair that the AI starts to think what is going on and simulate different views to show for a Facebook user when they are wiggling a gadget or using their mouse to caress the pic.
As expected, the AI fails epically to simulate a non-trivial background part eclipsed in the original 2d colour image or to simulate a complex 3d map with many small objects of different depth, lake hazy globe of a galactic core.
It's only good and possibly wow-good when the depth map is simple and the eclipsed parts are trivial.
Here is my examle.
I uploaded this pair:
The original was
Webb NIRCam composite image of Jupiter from three filters – F360M (red), F212N (yellow-green), and F150W2 (cyan) – and alignment due to the planet’s rotation. Credit: NASA, ESA, CSA, Jupiter ERS Team; image processing by Judy Schmidt.
I generated a depth map of a sphere, fitted it to the Jup pic and made it 20% transparent to turn bright details into hills for an unscientific wow
[quote="Chris Peterson" post_id=324974 time=1660138309 user_id=117706]
[quote=johnnydeep post_id=324972 time=1660136850 user_id=132061]
[quote=SeedsofEarfth post_id=324963 time=1660084115]
One 3-D technology (aside from Facebook) that used to be quite effective in creating a real 3-d effect without was by polarising the light from the subject using two lenses that polarised at different angles so that each eye saw a slightly different image in each eye. Melded together by the brain, the depth perception was remarkable. I remember using those polarising glasses in movie theaters back in the 1950s and 1960s. As for Facebook:
How do I create a 3D photo on Facebook?
Tap "What's on your mind?" at the top of your Feed or tap "Write something"... in a group.
Tap Photo/Video.
Select your photo and tap Done.
Tap Effects.
On the Choose Effect panel, select 3D.
Add any text you'd like, then tap Post.
However, I'm not sure about the actual process that converts a 2-D image into a 3-D image.
[/quote]
I see no 3D option when posting a picture with the Facebook website on my PC. Perhaps it's only available on the Facebook app? Oh well, I detest Facebook anyway.
[/quote]
We would not want APOD images displayed like this. The image is analyzed, and based on what the AI thinks is going on, a depth map is created, and then different views are simulated, with missing background estimated from the surrounding image. An animation is then created. Sometimes it works pretty well. Sometimes the AI fails miserably and you get truly bizarre artifacts.
Starting with true 3D data (e.g. an image pair) you could create a similar animation. I think I've seen software that would do that. But it's up to an image author to decide if they want to do that, and up to the editors if they think it would make a good APOD. In scientific publications, however, the standard way of presenting stereo data is with a pair of images, viewed either directly or using a pair of lenses (like a stereoscope). Occasionally anaglyphs like this APOD are used, but not commonly, since it requires special glasses and you lose color information.
[/quote]
A correction: to create a 3D photo on Facebook, you supply a pair of images, one a 2d colour image, another a [i]depth map[/i] monochrome image.
It's only after you upload the pair that the AI starts to think what is going on and simulate different views to show for a Facebook user when they are wiggling a gadget or using their mouse to caress the pic.
As expected, the AI fails epically to simulate a non-trivial background part eclipsed in the original 2d colour image or to simulate a complex 3d map with many small objects of different depth, lake hazy globe of a galactic core.
It's only good and possibly wow-good when the depth map is simple and the eclipsed parts are trivial.
[url="https://www.facebook.com/photo/?fbid=5320726828048385"]Here is my examle[/url].
I uploaded this pair:[attachment=1]Jupiter Judy Schmidt 3d.jpg[/attachment][attachment=0]Jupiter Judy Schmidt 3d_depth.jpg[/attachment]
The original was
[url="https://blogs.nasa.gov/webb/2022/08/22/webbs-jupiter-images-showcase-auroras-hazes/"]Webb NIRCam composite image of Jupiter from three filters – F360M (red), F212N (yellow-green), and F150W2 (cyan) – and alignment due to the planet’s rotation. Credit: NASA, ESA, CSA, Jupiter ERS Team; image processing by Judy Schmidt.[/url]
I generated a depth map of a sphere, fitted it to the Jup pic and made it 20% transparent to turn bright details into hills for an unscientific wow