• Welcome to the Internet Infidels Discussion Board.

Is this real? Or AI? Discuss...

unapologetic

55+ years without a god
Joined
Dec 12, 2023
Messages
2,536
Location
Penna., Fascist States of Amerika
Basic Beliefs
Hasa Diga Eebowai, antitheist,
Sometimes I need a second opinion. And we could train ourselves to better spot fake shit.
Watch the Ornithopter Flap Its Wings


I want to believe this, because it would be neat. But I find a few things suspicious, like why haven't I seen this before.
 
Last edited:
I occasionally see similar acts on AGT videos on YT where it's done live.
 
Google AI indicates it's unlikely.

real-world inventors from Leonardo da Vinci to modern drone engineers have built working models, though scaling them up for heavy human transport remains extremely difficult.

My brother had a friend who, as an inventor, made a good deal of money designing speaker systems. He fooled around with the ornithopter concept for a while. My brother assisted his experimenting with wing cross sections by holding model wings out the window of a traveling car. I think nothing came of the project.
 
Sometimes I need a second opinion. And we could train ourselves to better spot fake shit.
Watch the Ornithopter Flap Its Wings


I want to believe this, because it would be neat. But I find a few things suspicious, like why haven't I seen this before.

The video description says
"Watch this incredible aircraft lift gracefully into the sky in Microsoft Flight Simulator 2020, featuring cinematic camera work and immersive flight simulation."

I think the best tool to check whether something seems to be real or fake is:

Though it seems to cost money.

BTW here is a video from 9 months ago where people try to guess whether videos are real or AI and they seem to get it wrong about half the time (false positive and false negative) which you would expect for chance when they have no idea how to tell the difference.


On the other hand this is from 2023
 
I only got 4 right.
The TV still had picture on the broken pieces.
'Prompting' was spelled wrong on his cell phone. (I shouldn't talk about other people's spelling)
 
Is this real? Looks surreal and easily generated.
Mayyas Official - HORRA (Official Music Video)

I think the people are real because it would be too much work to have them all controlled by AI (or motion captured) and it is from a year ago where generative AI wasn't as good.
 
Sometimes I need a second opinion. And we could train ourselves to better spot fake shit.
Watch the Ornithopter Flap Its Wings


I want to believe this, because it would be neat. But I find a few things suspicious, like why haven't I seen this before.


The roads on both sides in the background are too strange to be real. About 45 degrees up?
 
I think the people are real because it would be too much work to have them all controlled by AI (or motion captured) and it is from a year ago where generative AI wasn't as good.
The video in the origiinal post says that it is not available, but your one is fine. I agree that it would be real, as nothing there that actual performers don't do. I have only one question - which Goa'uld is that? ;)
 
The roads on both sides in the background are too strange to be real. About 45 degrees up?
I believe that it possible to make a machine that can flap its wings fast, but not that that would provide actual lift as shown, and how did it change direction? Of course the answer is that as someone said, it is in a flight simulator.
 
BTW about that dancing video, which was from a year ago: (copilot said this)
Twelve months ago, generative‑AI video models could only produce 4–10 second coherent scenes.
Today (mid‑2026), the top models can generate 20–25 second scenes reliably, and some can produce up to ~120 seconds (2 minutes) of continuous, coherent footage in a single pass. Segment‑chaining can extend this to 5+ minutes, though coherence drops.
 
BTW about that dancing video, which was from a year ago: (copilot said this)
Twelve months ago, generative‑AI video models could only produce 4–10 second coherent scenes.
Today (mid‑2026), the top models can generate 20–25 second scenes reliably, and some can produce up to ~120 seconds (2 minutes) of continuous, coherent footage in a single pass. Segment‑chaining can extend this to 5+ minutes, though coherence drops.

Define continuous. In a mathematical sense, switching cameras (with different locations and perspectives would be discontinuous). That is, just because the time goes from 0 to 100 seconds (or whatever) does not necessarily mean continuous. Define coherent and coherence.

Let's skip all that for a moment and get to the meat of the problem:
Watch the video from about 20 seconds to about 24 seconds.

Within this timeframe the camera swaps to an alleged other camera or camera view on the right hand side. Compare the functional image content from camera 1 to camera 2:
(1) The yellow sign (on the left) and its content and proximity to the flight machine before and after;
(2) The back legs of the flight machine and whether or not they touch the ground and whether or not that is feasible.
(3) The number of legs on each side and how close they are together.

Sometimes I need a second opinion. And we could train ourselves to better spot fake shit.

Take a close look at the timeframe I have documented and compare the 3 features both before and after camera swap.
 
Last edited:
BTW about that dancing video, which was from a year ago: (copilot said this)
Twelve months ago, generative‑AI video models could only produce 4–10 second coherent scenes.
Today (mid‑2026), the top models can generate 20–25 second scenes reliably, and some can produce up to ~120 seconds (2 minutes) of continuous, coherent footage in a single pass. Segment‑chaining can extend this to 5+ minutes, though coherence drops.
Define continuous.
A single AI output video and it would involve one camera angle
Define coherent and coherence.
Parts of the video stay consistent - e.g. the characters and scene keep on looking the same
 
BTW about that dancing video, which was from a year ago: (copilot said this)
Twelve months ago, generative‑AI video models could only produce 4–10 second coherent scenes.
Today (mid‑2026), the top models can generate 20–25 second scenes reliably, and some can produce up to ~120 seconds (2 minutes) of continuous, coherent footage in a single pass. Segment‑chaining can extend this to 5+ minutes, though coherence drops.
Define continuous.
A single AI output video.

A single video can be multiple videos strung together. With AI VIDEO CREATION tools you can do that by specifying the initial frame as the last frame from the last video. That will give you a second video but you can use traditional tools to merge them.

Define coherent and coherence.
Parts of the video stay consistent - e.g. the characters and scene keep on looking the same

In the video in question, the alleged camera changes several times. After the second time, there are changes I listed. Did you look? That is probably one of the points where it was stitched together. On the other hand, the AI itself could have gotten confused if instructed to switch camera angle in its prompt. Either way, did you look at the three features at ~20 vs ~24 seconds?
 
Define continuous.
A single AI output video.
A single video can be multiple videos strung together. With AI VIDEO CREATION tools you can do that by specifying the initial frame as the last frame from the last video. That will give you a second video but you can use traditional tools to merge them.
The join might not be seamless if the velocities of the objects don't match up perfectly but I guess that could count as continuous.
Define coherent and coherence.
Parts of the video stay consistent - e.g. the characters and scene keep on looking the same
In the video in question, the alleged camera changes several times. After the second time, there are changes I listed. Did you look? That is probably one of the points where it was stitched together. On the other hand, the AI itself could have gotten confused if instructed to switch camera angle in its prompt. Either way, did you look at the three features at ~20 vs ~24 seconds?
I think none of the video is related to AI. It is just from a computer game called Microsoft Flight Simulator 2020 and the aircraft model would have 100% been made by humans. There are hotkeys that allow the view of things to be changed. The YouTube description says "Pure MSFS gameplay - no AI".
 
Last edited:
Back
Top Bottom