Samsung GS22+ owner here. I’ve watched in realtime as a blurry blob moon photo I just took was replaced with a clean recognizable “moon” in my gallery, a few seconds after taking it.
It wouldn’t surprise me at all if this has gotten more advanced in recent years. Identify a few key stars as reference, and then paint in a pretty Milky Way from pre-defined images, while pretending to collect a timelapse.
Samsung themselves admitted to their moon pictures being faked some time ago, which makes me consider the idea that these photos could also be faked in a manner you’re describing (especially with all the effort Google has been putting into their AI).
It’s called post processing. Generally it takes multiple frames and combines them or gets other camera data.
In theory, if this were true, you should be able to get a light, place it where the moon is and point it at it, and it should replace it with a moon. But it doesn’t.
I had to dig through my phone to find these photos when i got home from work, but they were taken literally seconds apart with my phone on full optical + digital zoom. The detail on the second is absolutely absurd, while the first is what I typically see on my screen when I’ve tried to replicate since.
To be clear, the assumption is that the algorithms the phone is using to determine you were trying to take a photo of the moon are “smart” enough to identify it as a photo of a night sky focused on the moon, rather than a light bulb. I’m not sure how you’d set up a light of the correct brightness at infinite focal length to test this though.
ETA: I’ve never seen this post processing happen so starkly with anything other than a photo of the moon, so it sticks out pretty hard. And I take a lot of photos at work of things that are tough to capture clearly.
Some phones just insert stock photos
😦 Is that true?
Samsung GS22+ owner here. I’ve watched in realtime as a blurry blob moon photo I just took was replaced with a clean recognizable “moon” in my gallery, a few seconds after taking it.
It wouldn’t surprise me at all if this has gotten more advanced in recent years. Identify a few key stars as reference, and then paint in a pretty Milky Way from pre-defined images, while pretending to collect a timelapse.
That’s awful. I want my pictures to be real.
Samsung themselves admitted to their moon pictures being faked some time ago, which makes me consider the idea that these photos could also be faked in a manner you’re describing (especially with all the effort Google has been putting into their AI).
Well of course their moon pictures are faked, the moon is fake
If the moon was made of cheese, which cheese variety would it be?
Wensleydale!
The stinkiest roquefort you’ve ever smelled. That’s why they put it so far away, because of how stinky it is
It’s called post processing. Generally it takes multiple frames and combines them or gets other camera data.
In theory, if this were true, you should be able to get a light, place it where the moon is and point it at it, and it should replace it with a moon. But it doesn’t.
I had to dig through my phone to find these photos when i got home from work, but they were taken literally seconds apart with my phone on full optical + digital zoom. The detail on the second is absolutely absurd, while the first is what I typically see on my screen when I’ve tried to replicate since.
I reckon it’s just processing a raw capture. The only difference is exposure, which yeah, phones could easily post process.
Im familiar with post processing.
To be clear, the assumption is that the algorithms the phone is using to determine you were trying to take a photo of the moon are “smart” enough to identify it as a photo of a night sky focused on the moon, rather than a light bulb. I’m not sure how you’d set up a light of the correct brightness at infinite focal length to test this though.
ETA: I’ve never seen this post processing happen so starkly with anything other than a photo of the moon, so it sticks out pretty hard. And I take a lot of photos at work of things that are tough to capture clearly.
No.
k, gr8
This has been debunked
What about dmention7’s comment?