-
Security incident: ISF was recently accessed by intruders. Please change your password, and change it anywhere else you used it. Read more
Ed Applying luma curve to photos and footage
- Thread starter nt1
- Start date
-
- Tags
- 9/11 9/11 evidence
Thermal
August Member
JayUtah
Penultimate Amazing
None of this is even remotely related to image analysis techniques designed to detect manipulation or to ascertain the likely authenticity of an image. You seem to be randomly manipulating sliders—or as it seems, asking an AI to apply adjustments for you. The adjustments you've made to the luminous intensity of the images are simply absurd. The halo you're observing is 100% the product of an absurd amplification of midrange tones. That would be appropriate for a crude edge-detection algorithm, but it is in no way revealing something hidden in the picture.
Most armchair detectives on the internet have absolutely no idea what actual image analysis looks or feels like, and have no understanding whatsoever of the statistical tools and models that we actually use. They tend to just fiddle with the tools presented to them in toy image manipulation programs and imagine all kinds of things that they have suddenly revealed.
I cannot stress this enough: nothing that you're doing has the slightest to do with how images are actually analyzed for authenticity.
Norman Alexander
Penultimate Amazing
Interestingly, another poster here uses the same chat bot to "prove" the absurd claim that the Turks invented everything. So why should we take these chat bot claims as gospel?
I am standing right here birdy. I have much more than just these little pictures i can come uo with all day. Hope they pay ya'll overtime.
Attachments
JayUtah
Penultimate Amazing
Was there meant to be a rebuttal?
Norman Alexander
Penultimate Amazing
JayUtah
Penultimate Amazing
I do not recognize ChatGPT as an expert in my licensed professional field.
Are you able to explain in your own words the rationale behind your severe manipulation of the luminance channel? Are you able to describe mathematically how that would result in evidence of digital manipulation? Are you able to show examples of how this particular technique has previously been used to detect manipulated images?
Chatty? He is wearing an inspector gadget coat. And you are late to the party because i started using chatty for fun when it became apparent that this is a round robin of hilarious teammates.
Am i able to mathmatically describe altering photos? Are you able to visually do a math problem with a flashlight?
Jk. But if you are being genuine and not s
Just showing up for the second shift. I will answer tour question.
When it gets this hilarious i barely make it past the first couple words.
Its on you guys. We could score. Ill go back ill tally up effort. I guarantee i will show more work and effort than the team that is here.
Thermal
August Member
This should be entertaining, if slightly painful, to watch.
JayUtah
Penultimate Amazing
Yes it is
Quoting just
its on you guys
What context bud?
You using that out of context?
JayUtah
Penultimate Amazing
No, you don't get to simply assert that your method is valid.
Why did you apply the particular curve that you did?
What do you mean by "bends?"
No. More importantly, without proper statistical controls, contrast contraction and stretch-out will create artifacts. These are not things that are present in the image but rather artifacts created from the behavior of the algorithm at the extrema in the data.
Ahhh.
Its on you guys....
Let me spell that out for the guy in the british judge wig.
Your team has been nothing but a swarm of little bees. Ill be happy to walk through this message for message.
And not just three words out of context like you.
"Its on you guys." If you want to calm your little horses. Not purposely going round robin.
Im ready to have a fun tough discussion.
But you are either purposely doing this or your excited like kids or dogs to see a new person.
Your tactics are hilarious.
Its on you guys to calm down and act like you have some class and seniority and speak to someone as if you have some manors.
Go ready my welcome message.
Im going go read it now for fun.
I would love to have a tough conversation about tough issues. That doesnt mean i want a guy in a british judge wig acting like he's never spoke to another man before.
Edited by Agatha:
Edited to remove incivility. Please refresh your understanding of the Membership Agreement to which you agreed on joining the forum, paying particular attention to rules 0, 6, 11 and 12.
Last edited by a moderator:
Oystein
Penultimate Amazing
- Joined
- Dec 9, 2009
- Messages
- 18,941
I know little about digital image manipulation, even though I do play around a bit with simple programs such as IrfanView and MS Paint. Yes, I know, massive legacy.
When I present photos on the internet - from vacations, portraits of pretty persons or cute cats, say - there are a few steps I used to do typically or routinely on raw photos downloaded from my real digital cameras (as opposed to smartphones, which do tons of manipulation before even saving the first version):
BUT what it does in reality is to look for "edges" (small regions with drop in brightness between pixels) and then increase local contrast. You can use this function in IrfanView with a standard setting, or with a customizing dialogue where you define how finely or coarsely edges are to be detected, over how many pixels of area, and to increase contrast by how much.
In short: That function finds edges that are really there, and manipulates them to make them even more edgy.
This is typically not done to alter the content of the picture, just make it look better.
So, if a nifty forensic analyst detects signs that edges in my cat photos have been manipulated (sharpened by algorithm), is that evidence of malfeasance? Does it suggest the cat has been added to an originally cat-free image by impostors?
All video of the 9/11 events that we find on YouTube have necessarily gone through a series of conversions, and in each conversions, "manipulations" take place:
In 2001, many video cameras worked in TV-mode, with interlacing and what have you. Such videos would often be stored on a tape reel, and someone would make (lossy!) copies of the tape real to a copy tape, to give to the FBI or NIST for analysis.
These would then be converted (lossily!) to digital files using whatever codec and encoding format was vogue at the time.
In the best case, someone got hold of that 3rd-generation incarnation of the video and upload it to YouTube. But that is yet another lossy conversion, for YouTube does its own routine manipulations, though I can't tell you which, and what options there are or were available. Surely, compression is involved. Surely, adjustment to some standard image format is involved. Surely, there is a change to some different encoding format and/or codec.
Generally, however, we don't know if a video we find on YT has been uploaded from the very 1st generation of digital file.
It could be that the digital copy of the original analogue tape has been enhanced, compressed, adjusted, transformed to a different aspect ratio or image size. It could be that the uploader had downloaded it from an earlier YT video and re-uploaded, with now two (or more) sets of manipulations associated with the YT upload.
Do we know that routines such as "sharpening" have not been done on a given video at some point of its evolution?
So, someone playing around with a tool that forensicists may also use finds evidence of "artefacts".
Is that reason for concern, or fully expected and a nothingburger?
My hunch is it's the latter.
The burden of evidence to show that the particular artefact found is indeed indicative of malfeasance falls squarely on that someone. They'd have to show that...
...on one hand, this artefact often appears where an actual manipulation (that is well-defined) has taken place, even in YT videos of unknown pefigree
...on the other hand, the same artefact appears essentially never in YT videos, no matter their pedigree, that have not undergone that same sort of manipulation.
Our "someone", nt1, so far does not seem to have started to show proof of either hand of that task.
When I present photos on the internet - from vacations, portraits of pretty persons or cute cats, say - there are a few steps I used to do typically or routinely on raw photos downloaded from my real digital cameras (as opposed to smartphones, which do tons of manipulation before even saving the first version):
- optionally: crop
- optionally: retouch some detail; such as perhaps stamp out something ugly or private
- resize to optimize size for the context in which I am going to post the photo
- adjust color balance, gamma, color saturation so the photo looks most natura, pleasing, interesting, dramatic, or whatever character I want it to have
- final step: sharpen image.
BUT what it does in reality is to look for "edges" (small regions with drop in brightness between pixels) and then increase local contrast. You can use this function in IrfanView with a standard setting, or with a customizing dialogue where you define how finely or coarsely edges are to be detected, over how many pixels of area, and to increase contrast by how much.
In short: That function finds edges that are really there, and manipulates them to make them even more edgy.
This is typically not done to alter the content of the picture, just make it look better.
So, if a nifty forensic analyst detects signs that edges in my cat photos have been manipulated (sharpened by algorithm), is that evidence of malfeasance? Does it suggest the cat has been added to an originally cat-free image by impostors?
All video of the 9/11 events that we find on YouTube have necessarily gone through a series of conversions, and in each conversions, "manipulations" take place:
In 2001, many video cameras worked in TV-mode, with interlacing and what have you. Such videos would often be stored on a tape reel, and someone would make (lossy!) copies of the tape real to a copy tape, to give to the FBI or NIST for analysis.
These would then be converted (lossily!) to digital files using whatever codec and encoding format was vogue at the time.
In the best case, someone got hold of that 3rd-generation incarnation of the video and upload it to YouTube. But that is yet another lossy conversion, for YouTube does its own routine manipulations, though I can't tell you which, and what options there are or were available. Surely, compression is involved. Surely, adjustment to some standard image format is involved. Surely, there is a change to some different encoding format and/or codec.
Generally, however, we don't know if a video we find on YT has been uploaded from the very 1st generation of digital file.
It could be that the digital copy of the original analogue tape has been enhanced, compressed, adjusted, transformed to a different aspect ratio or image size. It could be that the uploader had downloaded it from an earlier YT video and re-uploaded, with now two (or more) sets of manipulations associated with the YT upload.
Do we know that routines such as "sharpening" have not been done on a given video at some point of its evolution?
So, someone playing around with a tool that forensicists may also use finds evidence of "artefacts".
Is that reason for concern, or fully expected and a nothingburger?
My hunch is it's the latter.
The burden of evidence to show that the particular artefact found is indeed indicative of malfeasance falls squarely on that someone. They'd have to show that...
...on one hand, this artefact often appears where an actual manipulation (that is well-defined) has taken place, even in YT videos of unknown pefigree
...on the other hand, the same artefact appears essentially never in YT videos, no matter their pedigree, that have not undergone that same sort of manipulation.
Our "someone", nt1, so far does not seem to have started to show proof of either hand of that task.
Oystein
Penultimate Amazing
- Joined
- Dec 9, 2009
- Messages
- 18,941
Better yet: post all text as text, not as images of text.
