The Statistical Footprint of AI-Generated Images - Revision history

Mikesomm: Changed into redirect page; can be deleted.

2025-12-13T21:39:14Z

Changed into redirect page; can be deleted.

← Older revision		Revision as of 21:39, 13 December 2025
Line 1:		Line 1:
	~~=== The Blur Between Real and Artificial ===~~		#REDIRECT [[The Statistical Fingerprint of AI-Generated Images]]
	[[~~File:Sony_Awards_The_Electrician.png\|thumb\|right\|300px\|Figure 1: "~~The ~~Electrician" by Boris Eldagsen, the AI-generated image that won the Creative category at the 2023 Sony World Photography Awards.]]~~

	In 2023, the Sony World Photography Awards awarded a prize in the Creative category to an image titled ''"The Electrician"''. The image presented a haunting, black-and-white portrait of two women. However, the artist, Boris Eldagsen, refused the award, revealing that the work was not a photograph at all, but a synthetic creation generated by AI. This event marked a significant turning point: synthetic imagery had crossed a threshold of fidelity where even expert judges could no longer distinguish pixels captured by photons from artificially generated pixels.[[#ref1\|<sup>[1]</sup>]] This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just 1.5 years, generative AI models have produced as many images as traditional photography produced in its first 150 years (approximately 15 billion images).[[#ref2\|<sup>[2]</sup>]] We are rapidly entering an era where a significant portion of digital visual data is artificially generated, rather than captured on a physical image sensing system.

	~~[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics illustrating the scale of AI image generation distributed across different models.]]~~

	~~=== Research Question and Project Goals ===~~
	For the general public, the inability to distinguish real from synthetic raises concerns about misinformation, as well as a loss of creativity intrinsic to humans. However, for Image Systems Engineering, artificially generated images also provoke a serious question of engineering safety. Developers of autonomous systems, such as autonomous vehicles (AVs), are increasingly turning to generative AI to create training data for perception systems to bridge the gap between expensive real-world data and the need for massive edge-case datasets, such as car accidents and severe weather. Although these digital twins may appear realistic to human observers, they often fail to behave like real sensors, lacking the correct noise, spectral, or optical properties. Thus, artificially generated data introduces a high-risk domain gap where autonomous systems are trained on hallucinations rather than physics. This poses a critical question: '''Can we discern the real from the artificial?'''

	This project conducts a forensic analysis to determine whether AI-generated images can be distinguished from camera-simulated images based on radiometric statistics. In this investigation, we ignore the semantic content (whether the car looks like a car) and focus entirely on the physical statistics (how the image was formed from the radiance source). By comparing a '''physical radiance''' dataset (ISET3D physically based ray tracing[[#ref3\|<sup>[3]</sup>]]) and '''camera photograph''' dataset (ISETCam[[#ref4\|<sup>[4]</sup>]]) against an '''artificially generated image''' dataset (Stable Diffusion v1.5[[#ref5\|<sup>[5]</sup>]]), we aim to quantify the statistical fingerprints of ~~the~~ AI ~~across four domains:~~
	* '''Spatial Statistics:''' Texture and frequency distribution.
	* '''Photometric Statistics:''' Signal-~~dependent noise response.~~
	* '''Spectral Statistics:''' Inter-channel color correlation.
	* '''Optical Statistics:''' Point spread function (PSF) and diffraction.

	~~== References ==~~
	~~<div id="ref1">[1] Paul Glynn, "Sony World Photography Award 2023: Winner refuses award after revealing AI creation," BBC News, 18 April 2023.</div>~~
	~~<div id="ref2">[2] Everypixel Journal, "AI Image Statistics Report," August 2023.</div>~~
	~~<div id="ref3">[3] Wandell, B. and Liu, Z. (2024). "ISET HDR Driving scene light groups." Stanford Digital Repository.</div>~~
	~~<div id="ref4">[4~~] ~~Farrell, J. E., et al. (2012). "ISET: The Image Systems Engineering Toolbox." Proceedings of the Society for Imaging Science and Technology.</div>~~
	~~<div id="ref5">[5~~] ~~Rombach, R., et al. (2022). "High-Resolution Image Synthesis with Latent Diffusion Models." CVPR.</div>~~

Mikesomm: Updated third paragraph and added some references.

2025-12-13T21:33:24Z

Updated third paragraph and added some references.

← Older revision		Revision as of 21:33, 13 December 2025
Line 1:		Line 1:
	~~== Introduction ==~~

	=== The Blur Between Real and Artificial ===		=== The Blur Between Real and Artificial ===
	[[File:Sony_Awards_The_Electrician.png\|thumb\|right\|300px\|Figure 1: "The Electrician" by Boris Eldagsen, the AI-generated image that won the Creative category at the 2023 Sony World Photography Awards.]]		[[File:Sony_Awards_The_Electrician.png\|thumb\|right\|300px\|Figure 1: "The Electrician" by Boris Eldagsen, the AI-generated image that won the Creative category at the 2023 Sony World Photography Awards.]]
Line 11:		Line 9:
	For the general public, the inability to distinguish real from synthetic raises concerns about misinformation, as well as a loss of creativity intrinsic to humans. However, for Image Systems Engineering, artificially generated images also provoke a serious question of engineering safety. Developers of autonomous systems, such as autonomous vehicles (AVs), are increasingly turning to generative AI to create training data for perception systems to bridge the gap between expensive real-world data and the need for massive edge-case datasets, such as car accidents and severe weather. Although these digital twins may appear realistic to human observers, they often fail to behave like real sensors, lacking the correct noise, spectral, or optical properties. Thus, artificially generated data introduces a high-risk domain gap where autonomous systems are trained on hallucinations rather than physics. This poses a critical question: '''Can we discern the real from the artificial?'''		For the general public, the inability to distinguish real from synthetic raises concerns about misinformation, as well as a loss of creativity intrinsic to humans. However, for Image Systems Engineering, artificially generated images also provoke a serious question of engineering safety. Developers of autonomous systems, such as autonomous vehicles (AVs), are increasingly turning to generative AI to create training data for perception systems to bridge the gap between expensive real-world data and the need for massive edge-case datasets, such as car accidents and severe weather. Although these digital twins may appear realistic to human observers, they often fail to behave like real sensors, lacking the correct noise, spectral, or optical properties. Thus, artificially generated data introduces a high-risk domain gap where autonomous systems are trained on hallucinations rather than physics. This poses a critical question: '''Can we discern the real from the artificial?'''

	This project conducts a forensic analysis to determine whether AI-generated images can be distinguished from camera-simulated images based on radiometric statistics. In this investigation, we ignore the semantic content (whether the car looks like a car) and focus entirely on the physical statistics (how the image was formed from the radiance source). By comparing a ~~Physical Ground Truth~~ dataset (~~Physically Based Ray Tracing~~) and ~~Camera~~ dataset (~~simulated in~~ ISETCam) against ~~a AI Generated~~ dataset (~~generated via~~ Stable Diffusion v1.5), we aim to quantify the statistical fingerprints of the AI across four domains:		This project conducts a forensic analysis to determine whether AI-generated images can be distinguished from camera-simulated images based on radiometric statistics. In this investigation, we ignore the semantic content (whether the car looks like a car) and focus entirely on the physical statistics (how the image was formed from the radiance source). By comparing a '''physical radiance''' dataset (ISET3D physically based ray tracing[[#ref3\|<sup>[3]</sup>]]) and '''camera photograph''' dataset (ISETCam[[#ref4\|<sup>[4]</sup>]]) against an '''artificially generated image''' dataset (Stable Diffusion v1.5[[#ref5\|<sup>[5]</sup>]]), we aim to quantify the statistical fingerprints of the AI across four domains:
	* '''Spatial Statistics:''' Texture and frequency distribution.		* '''Spatial Statistics:''' Texture and frequency distribution.
	* '''Photometric Statistics:''' Signal-dependent noise response.		* '''Photometric Statistics:''' Signal-dependent noise response.
Line 20:		Line 18:
	<div id="ref1">[1] Paul Glynn, "Sony World Photography Award 2023: Winner refuses award after revealing AI creation," BBC News, 18 April 2023.</div>		<div id="ref1">[1] Paul Glynn, "Sony World Photography Award 2023: Winner refuses award after revealing AI creation," BBC News, 18 April 2023.</div>
	<div id="ref2">[2] Everypixel Journal, "AI Image Statistics Report," August 2023.</div>		<div id="ref2">[2] Everypixel Journal, "AI Image Statistics Report," August 2023.</div>
			<div id="ref3">[3] Wandell, B. and Liu, Z. (2024). "ISET HDR Driving scene light groups." Stanford Digital Repository.</div>
			<div id="ref4">[4] Farrell, J. E., et al. (2012). "ISET: The Image Systems Engineering Toolbox." Proceedings of the Society for Imaging Science and Technology.</div>
			<div id="ref5">[5] Rombach, R., et al. (2022). "High-Resolution Image Synthesis with Latent Diffusion Models." CVPR.</div>

Mikesomm at 21:27, 13 December 2025

2025-12-13T21:27:55Z

← Older revision		Revision as of 21:27, 13 December 2025
Line 11:		Line 11:
	For the general public, the inability to distinguish real from synthetic raises concerns about misinformation, as well as a loss of creativity intrinsic to humans. However, for Image Systems Engineering, artificially generated images also provoke a serious question of engineering safety. Developers of autonomous systems, such as autonomous vehicles (AVs), are increasingly turning to generative AI to create training data for perception systems to bridge the gap between expensive real-world data and the need for massive edge-case datasets, such as car accidents and severe weather. Although these digital twins may appear realistic to human observers, they often fail to behave like real sensors, lacking the correct noise, spectral, or optical properties. Thus, artificially generated data introduces a high-risk domain gap where autonomous systems are trained on hallucinations rather than physics. This poses a critical question: '''Can we discern the real from the artificial?'''		For the general public, the inability to distinguish real from synthetic raises concerns about misinformation, as well as a loss of creativity intrinsic to humans. However, for Image Systems Engineering, artificially generated images also provoke a serious question of engineering safety. Developers of autonomous systems, such as autonomous vehicles (AVs), are increasingly turning to generative AI to create training data for perception systems to bridge the gap between expensive real-world data and the need for massive edge-case datasets, such as car accidents and severe weather. Although these digital twins may appear realistic to human observers, they often fail to behave like real sensors, lacking the correct noise, spectral, or optical properties. Thus, artificially generated data introduces a high-risk domain gap where autonomous systems are trained on hallucinations rather than physics. This poses a critical question: '''Can we discern the real from the artificial?'''

	This project conducts a forensic analysis of AI-generated ~~sensor data to determine its viability for simulation~~. We ignore the ''semantic'' content (~~e.g.,~~ whether the car looks like a car) and focus entirely on the ''physical'' statistics (how the image was formed). By comparing a ~~'''~~Physical Ground Truth~~'''~~ (simulated ~~via [[~~ISETCam]]) against a ~~'''Generative Reconstruction'''~~ (generated via Stable Diffusion v1.5), we aim to quantify the statistical "fingerprints" of the AI across four domains:		This project conducts a forensic analysis to determine whether AI-generated images can be distinguished from camera-simulated images based on radiometric statistics. In this investigation, we ignore the semantic content (whether the car looks like a car) and focus entirely on the physical statistics (how the image was formed from the radiance source). By comparing a Physical Ground Truth dataset (Physically Based Ray Tracing) and Camera dataset (simulated in ISETCam) against a AI Generated dataset (generated via Stable Diffusion v1.5), we aim to quantify the statistical fingerprints of the AI across four domains:
	* '''Spatial Statistics:''' Texture and frequency distribution.		* '''Spatial Statistics:''' Texture and frequency distribution.
	* '''Photometric Statistics:''' Signal-dependent noise response.		* '''Photometric Statistics:''' Signal-dependent noise response.

Mikesomm: Updated second paragraph.

2025-12-13T21:12:47Z

Updated second paragraph.

← Older revision		Revision as of 21:12, 13 December 2025
Line 9:		Line 9:

	=== Research Question and Project Goals ===		=== Research Question and Project Goals ===
	For the general public, the inability to distinguish real from synthetic is a ~~question~~ of ~~misinformation~~. However, for Image Systems Engineering, ~~it is~~ a question of ~~'''~~safety~~'''~~. ~~Autonomous Vehicle~~ (AV) ~~developers~~ are increasingly turning to generative AI to create training data for perception systems to bridge the gap between expensive real-world data and the need for massive "edge-case" datasets ~~(e.g.~~, accidents, severe weather). ~~This poses a critical question: '''"Can we discern the real from the artificial?"''' If~~ these ~~"Digital Twins"~~ appear realistic to human observers ~~but~~ fail to behave like real ~~sensors—lacking~~ the correct noise, spectral, or optical ~~properties—they~~ risk ~~introducing a '''Domain Gap'''~~ where ~~AVs~~ are trained on hallucinations rather than physics.		For the general public, the inability to distinguish real from synthetic raises concerns about misinformation, as well as a loss of creativity intrinsic to humans. However, for Image Systems Engineering, artificially generated images also provoke a serious question of engineering safety. Developers of autonomous systems, such as autonomous vehicles (AVs), are increasingly turning to generative AI to create training data for perception systems to bridge the gap between expensive real-world data and the need for massive edge-case datasets, such as car accidents and severe weather. Although these digital twins may appear realistic to human observers, they often fail to behave like real sensors, lacking the correct noise, spectral, or optical properties. Thus, artificially generated data introduces a high-risk domain gap where autonomous systems are trained on hallucinations rather than physics. This poses a critical question: '''Can we discern the real from the artificial?'''

	This project conducts a forensic analysis of AI-generated sensor data to determine its viability for simulation. We ignore the ''semantic'' content (e.g., whether the car looks like a car) and focus entirely on the ''physical'' statistics (how the image was formed). By comparing a '''Physical Ground Truth''' (simulated via [[ISETCam]]) against a '''Generative Reconstruction''' (generated via Stable Diffusion v1.5), we aim to quantify the statistical "fingerprints" of the AI across four domains:		This project conducts a forensic analysis of AI-generated sensor data to determine its viability for simulation. We ignore the ''semantic'' content (e.g., whether the car looks like a car) and focus entirely on the ''physical'' statistics (how the image was formed). By comparing a '''Physical Ground Truth''' (simulated via [[ISETCam]]) against a '''Generative Reconstruction''' (generated via Stable Diffusion v1.5), we aim to quantify the statistical "fingerprints" of the AI across four domains:

Mikesomm: Updated first paragraph.

2025-12-13T21:07:15Z

Updated first paragraph.

← Older revision		Revision as of 21:07, 13 December 2025
Line 4:		Line 4:
	[[File:Sony_Awards_The_Electrician.png\|thumb\|right\|300px\|Figure 1: "The Electrician" by Boris Eldagsen, the AI-generated image that won the Creative category at the 2023 Sony World Photography Awards.]]		[[File:Sony_Awards_The_Electrician.png\|thumb\|right\|300px\|Figure 1: "The Electrician" by Boris Eldagsen, the AI-generated image that won the Creative category at the 2023 Sony World Photography Awards.]]

	In 2023, the Sony World Photography ~~Awards—one of the most prestigious competitions in the field—awarded~~ a prize in the Creative category to an image titled ''"The Electrician"''. The image presented a haunting, black-and-white portrait of two women. However, the artist, Boris Eldagsen, refused the award, revealing that the work was not a photograph at all, but a synthetic creation generated by AI. This event marked a significant turning point: synthetic imagery had crossed a threshold of fidelity where even expert judges could no longer distinguish pixels captured by photons from pixels ~~hallucinated by neural networks~~.[[#ref1\|<sup>[1]</sup>]] This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just ~~'''~~1.5 years~~'''~~, generative AI models have produced as many images as traditional photography produced in its first ~~'''~~150 years~~'''~~ (approximately 15 billion images).[[#ref2\|<sup>[2]</sup>]] We are rapidly entering an era where a significant portion of digital visual data is ~~synthetic~~.		In 2023, the Sony World Photography Awards awarded a prize in the Creative category to an image titled ''"The Electrician"''. The image presented a haunting, black-and-white portrait of two women. However, the artist, Boris Eldagsen, refused the award, revealing that the work was not a photograph at all, but a synthetic creation generated by AI. This event marked a significant turning point: synthetic imagery had crossed a threshold of fidelity where even expert judges could no longer distinguish pixels captured by photons from artificially generated pixels.[[#ref1\|<sup>[1]</sup>]] This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just 1.5 years, generative AI models have produced as many images as traditional photography produced in its first 150 years (approximately 15 billion images).[[#ref2\|<sup>[2]</sup>]] We are rapidly entering an era where a significant portion of digital visual data is artificially generated, rather than captured on a physical image sensing system.

	[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics illustrating the scale of AI image generation distributed across different models.]]		[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics illustrating the scale of AI image generation distributed across different models.]]

Mikesomm: Updated Figure 2 caption.

2025-12-12T20:04:16Z

Updated Figure 2 caption.

← Older revision		Revision as of 20:04, 12 December 2025
Line 6:		Line 6:
	In 2023, the Sony World Photography Awards—one of the most prestigious competitions in the field—awarded a prize in the Creative category to an image titled ''"The Electrician"''. The image presented a haunting, black-and-white portrait of two women. However, the artist, Boris Eldagsen, refused the award, revealing that the work was not a photograph at all, but a synthetic creation generated by AI. This event marked a significant turning point: synthetic imagery had crossed a threshold of fidelity where even expert judges could no longer distinguish pixels captured by photons from pixels hallucinated by neural networks.[[#ref1\|<sup>[1]</sup>]] This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just '''1.5 years''', generative AI models have produced as many images as traditional photography produced in its first '''150 years''' (approximately 15 billion images).[[#ref2\|<sup>[2]</sup>]] We are rapidly entering an era where a significant portion of digital visual data is synthetic.		In 2023, the Sony World Photography Awards—one of the most prestigious competitions in the field—awarded a prize in the Creative category to an image titled ''"The Electrician"''. The image presented a haunting, black-and-white portrait of two women. However, the artist, Boris Eldagsen, refused the award, revealing that the work was not a photograph at all, but a synthetic creation generated by AI. This event marked a significant turning point: synthetic imagery had crossed a threshold of fidelity where even expert judges could no longer distinguish pixels captured by photons from pixels hallucinated by neural networks.[[#ref1\|<sup>[1]</sup>]] This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just '''1.5 years''', generative AI models have produced as many images as traditional photography produced in its first '''150 years''' (approximately 15 billion images).[[#ref2\|<sup>[2]</sup>]] We are rapidly entering an era where a significant portion of digital visual data is synthetic.

	[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics ~~comparing~~ the ~~timeline~~ of AI image generation ~~vs. traditional photography~~.]]		[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics illustrating the scale of AI image generation distributed across different models.]]

	=== Research Question and Project Goals ===		=== Research Question and Project Goals ===

Mikesomm: Condensed the introducton sections.

2025-12-12T20:02:28Z

Condensed the introducton sections.

← Older revision		Revision as of 20:02, 12 December 2025
Line 1:		Line 1:
	== Introduction ==		== Introduction ==

	=== The Blur Between Real and ~~Synthetic~~ ===		=== The Blur Between Real and Artificial ===
	[[File:Sony_Awards_The_Electrician.png\|thumb\|right\|300px\|Figure 1: "The Electrician" by Boris Eldagsen, the AI-generated image that won the Creative category at the 2023 Sony World Photography Awards.]]		[[File:Sony_Awards_The_Electrician.png\|thumb\|right\|300px\|Figure 1: "The Electrician" by Boris Eldagsen, the AI-generated image that won the Creative category at the 2023 Sony World Photography Awards.]]

	In 2023, the Sony World Photography Awards—one of the most prestigious competitions in the field—awarded a prize in the Creative category to an image titled ''"The Electrician"''. The image presented a haunting, black-and-white portrait of two women. However, the artist, Boris Eldagsen, refused the award, revealing that the work was not a photograph at all, but a synthetic creation generated by AI. This event marked a significant turning point: synthetic imagery had crossed a threshold of fidelity where even expert judges could no longer distinguish pixels captured by photons from pixels hallucinated by neural networks.[[#ref1\|<sup>[1]</sup>]]		In 2023, the Sony World Photography Awards—one of the most prestigious competitions in the field—awarded a prize in the Creative category to an image titled ''"The Electrician"''. The image presented a haunting, black-and-white portrait of two women. However, the artist, Boris Eldagsen, refused the award, revealing that the work was not a photograph at all, but a synthetic creation generated by AI. This event marked a significant turning point: synthetic imagery had crossed a threshold of fidelity where even expert judges could no longer distinguish pixels captured by photons from pixels hallucinated by neural networks.[[#ref1\|<sup>[1]</sup>]] This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just '''1.5 years''', generative AI models have produced as many images as traditional photography produced in its first '''150 years''' (approximately 15 billion images).[[#ref2\|<sup>[2]</sup>]] We are rapidly entering an era where a significant portion of digital visual data is synthetic.

	~~=== The Scale of Generation ===~~
	[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics comparing the timeline of AI image generation vs. traditional photography.]]		[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics comparing the timeline of AI image generation vs. traditional photography.]]

	~~This blurring~~ of ~~reality~~ is ~~compounded by~~ the ~~unprecedented scale of production~~. ~~Recent reports indicate that in just~~ '''~~1.5 years~~''', ~~generative AI models have produced as many images as traditional photography produced in its first~~ '''~~150 years~~''' ~~(approximately 15 billion images).[[#ref2\|<sup>[2]</sup>]] We~~ are ~~rapidly entering an era where a significant portion of digital visual data is synthetic~~.		=== Research Question and Project Goals ===
			For the general public, the inability to distinguish real from synthetic is a question of misinformation. However, for Image Systems Engineering, it is a question of '''safety'''. Autonomous Vehicle (AV) developers are increasingly turning to generative AI to create training data for perception systems to bridge the gap between expensive real-world data and the need for massive "edge-case" datasets (e.g., accidents, severe weather). This poses a critical question: '''"Can we discern the real from the artificial?"''' If these "Digital Twins" appear realistic to human observers but fail to behave like real sensors—lacking the correct noise, spectral, or optical properties—they risk introducing a '''Domain Gap''' where AVs are trained on hallucinations rather than physics.

	~~=== The Research Question ===~~		This project conducts a forensic analysis of AI-generated sensor data to determine its viability for simulation. We ignore the ''semantic'' content (e.g., whether the car looks like a car) and focus entirely on the ''physical'' statistics (how the image was formed). By comparing a '''Physical Ground Truth''' (simulated via [[ISETCam]]) against a '''Generative Reconstruction''' (generated via Stable Diffusion v1.5), we aim to quantify the statistical "fingerprints" of the AI across four domains:
	~~For the general public, the inability to distinguish real from synthetic is a question of misinformation. However, for Image Systems Engineering, it is a question of '''safety'''.~~

	Autonomous Vehicle (AV) developers are increasingly turning to generative AI to create training data for perception systems to bridge the gap between expensive real-world data and the need for massive "edge-case" datasets (e.g., accidents, severe weather). This poses a critical question: '''"Can we discern the real from the artificial?"'''

	If these "Digital Twins" appear realistic to human observers but fail to behave like real sensors—lacking the correct noise, spectral, or optical properties—they risk introducing a '''Domain Gap''' where AVs are trained on hallucinations rather than physics.

	~~=== Project Goals ===~~
	This project conducts a forensic analysis of AI-generated sensor data to determine its viability for simulation. We ignore the ''semantic'' content (e.g., whether the car looks like a car) and focus entirely on the ''physical'' statistics (how the image was formed).

	By comparing a '''Physical Ground Truth''' (simulated via [[ISETCam]]) against a '''Generative Reconstruction''' (generated via Stable Diffusion v1.5), we aim to quantify the statistical "fingerprints" of the AI across four domains:
	* '''Spatial Statistics:''' Texture and frequency distribution.		* '''Spatial Statistics:''' Texture and frequency distribution.
	* '''Photometric Statistics:''' Signal-dependent noise response.		* '''Photometric Statistics:''' Signal-dependent noise response.

Mikesomm: Added draft for the introduction and skeleton for references.

2025-12-12T07:20:41Z

Added draft for the introduction and skeleton for references.

← Older revision		Revision as of 07:20, 12 December 2025
Line 4:		Line 4:
	[[File:Sony_Awards_The_Electrician.png\|thumb\|right\|300px\|Figure 1: "The Electrician" by Boris Eldagsen, the AI-generated image that won the Creative category at the 2023 Sony World Photography Awards.]]		[[File:Sony_Awards_The_Electrician.png\|thumb\|right\|300px\|Figure 1: "The Electrician" by Boris Eldagsen, the AI-generated image that won the Creative category at the 2023 Sony World Photography Awards.]]

	In 2023, the Sony World Photography Awards—one of the most prestigious competitions in the field—awarded a prize in the Creative category to an image titled ''"The Electrician"''. The image presented a haunting, black-and-white portrait of two women. However, the artist, Boris Eldagsen, refused the award, revealing that the work was not a photograph at all, but a synthetic creation generated by AI. This event marked a significant turning point: synthetic imagery had crossed a threshold of fidelity where even expert judges could no longer distinguish pixels captured by photons from pixels hallucinated by neural networks.<~~ref~~>~~Insert citation for Sony World Photography Awards / Boris Eldagsen controversy here.~~</~~ref~~>		In 2023, the Sony World Photography Awards—one of the most prestigious competitions in the field—awarded a prize in the Creative category to an image titled ''"The Electrician"''. The image presented a haunting, black-and-white portrait of two women. However, the artist, Boris Eldagsen, refused the award, revealing that the work was not a photograph at all, but a synthetic creation generated by AI. This event marked a significant turning point: synthetic imagery had crossed a threshold of fidelity where even expert judges could no longer distinguish pixels captured by photons from pixels hallucinated by neural networks.[[#ref1\|<sup>[1]</sup>]]

	=== The Scale of Generation ===		=== The Scale of Generation ===
	[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics comparing the timeline of AI image generation vs. traditional photography. ~~<ref name="everypixel">Everypixel Journal, "AI Image Statistics Report," August 2023.</ref>~~]]		[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics comparing the timeline of AI image generation vs. traditional photography.]]

	This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just '''1.5 years''', generative AI models have produced as many images as traditional photography produced in its first '''150 years'''.<~~ref name="everypixel"~~ /> We are rapidly entering an era where a significant portion of digital visual data is synthetic.		This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just '''1.5 years''', generative AI models have produced as many images as traditional photography produced in its first '''150 years''' (approximately 15 billion images).[[#ref2\|<sup>[2]</sup>]] We are rapidly entering an era where a significant portion of digital visual data is synthetic.

	=== The Research Question ===		=== The Research Question ===
	For the general public, the inability to distinguish real from synthetic is a question of misinformation. However, for Image Systems Engineering, it is a question of '''safety'''.		For the general public, the inability to distinguish real from synthetic is a question of misinformation. However, for Image Systems Engineering, it is a question of '''safety'''.

	Autonomous Vehicle (AV) developers are increasingly turning to generative AI to create training data for perception systems to bridge the gap between expensive real-world data and the need for massive edge-case datasets. This poses a critical question: '''"Can we discern the real from the artificial?"'''		Autonomous Vehicle (AV) developers are increasingly turning to generative AI to create training data for perception systems to bridge the gap between expensive real-world data and the need for massive "edge-case" datasets (e.g., accidents, severe weather). This poses a critical question: '''"Can we discern the real from the artificial?"'''

	If these "Digital Twins" appear realistic to human observers but fail to behave like real sensors—lacking the correct noise, spectral, or optical properties—they risk introducing a '''Domain Gap''' where AVs are trained on hallucinations rather than physics.		If these "Digital Twins" appear realistic to human observers but fail to behave like real sensors—lacking the correct noise, spectral, or optical properties—they risk introducing a '''Domain Gap''' where AVs are trained on hallucinations rather than physics.
Line 28:		Line 28:

	== References ==		== References ==
	~~{{Reflist}}~~		<div id="ref1">[1] Paul Glynn, "Sony World Photography Award 2023: Winner refuses award after revealing AI creation," BBC News, 18 April 2023.</div>
			<div id="ref2">[2] Everypixel Journal, "AI Image Statistics Report," August 2023.</div>

Mikesomm at 07:09, 12 December 2025

2025-12-12T07:09:51Z

← Older revision		Revision as of 07:09, 12 December 2025
Line 7:		Line 7:

	=== The Scale of Generation ===		=== The Scale of Generation ===
	[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics comparing the timeline of AI image generation vs. traditional photography. <ref name="everypixel">Everypixel Journal, ~~[https://journal.everypixel.com/ai-image-statistics~~ "AI Image Statistics Report"], August 2023.</ref>]]		[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics comparing the timeline of AI image generation vs. traditional photography. <ref name="everypixel">Everypixel Journal, "AI Image Statistics Report," August 2023.</ref>]]

	This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just '''1.5 years''', generative AI models have produced as many images as traditional photography produced in its first '''150 years'''.<ref name="everypixel" /> We are rapidly entering an era where a significant portion of digital visual data is synthetic.		This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just '''1.5 years''', generative AI models have produced as many images as traditional photography produced in its first '''150 years'''.<ref name="everypixel" /> We are rapidly entering an era where a significant portion of digital visual data is synthetic.
Line 28:		Line 28:

	== References ==		== References ==
	~~<references />~~		{{Reflist}}

Mikesomm at 07:03, 12 December 2025

2025-12-12T07:03:13Z

← Older revision		Revision as of 07:03, 12 December 2025
Line 1:		Line 1:
	~~Test~~		== Introduction ==

			=== The Blur Between Real and Synthetic ===
			[[File:Sony_Awards_The_Electrician.png\|thumb\|right\|300px\|Figure 1: "The Electrician" by Boris Eldagsen, the AI-generated image that won the Creative category at the 2023 Sony World Photography Awards.]]

			In 2023, the Sony World Photography Awards—one of the most prestigious competitions in the field—awarded a prize in the Creative category to an image titled ''"The Electrician"''. The image presented a haunting, black-and-white portrait of two women. However, the artist, Boris Eldagsen, refused the award, revealing that the work was not a photograph at all, but a synthetic creation generated by AI. This event marked a significant turning point: synthetic imagery had crossed a threshold of fidelity where even expert judges could no longer distinguish pixels captured by photons from pixels hallucinated by neural networks.<ref>Insert citation for Sony World Photography Awards / Boris Eldagsen controversy here.</ref>

			=== The Scale of Generation ===
			[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics comparing the timeline of AI image generation vs. traditional photography. <ref name="everypixel">Everypixel Journal, [https://journal.everypixel.com/ai-image-statistics "AI Image Statistics Report"], August 2023.</ref>]]

			This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just '''1.5 years''', generative AI models have produced as many images as traditional photography produced in its first '''150 years'''.<ref name="everypixel" /> We are rapidly entering an era where a significant portion of digital visual data is synthetic.

			=== The Research Question ===
			For the general public, the inability to distinguish real from synthetic is a question of misinformation. However, for Image Systems Engineering, it is a question of '''safety'''.

			Autonomous Vehicle (AV) developers are increasingly turning to generative AI to create training data for perception systems to bridge the gap between expensive real-world data and the need for massive edge-case datasets. This poses a critical question: '''"Can we discern the real from the artificial?"'''

			If these "Digital Twins" appear realistic to human observers but fail to behave like real sensors—lacking the correct noise, spectral, or optical properties—they risk introducing a '''Domain Gap''' where AVs are trained on hallucinations rather than physics.

			=== Project Goals ===
			This project conducts a forensic analysis of AI-generated sensor data to determine its viability for simulation. We ignore the ''semantic'' content (e.g., whether the car looks like a car) and focus entirely on the ''physical'' statistics (how the image was formed).

			By comparing a '''Physical Ground Truth''' (simulated via [[ISETCam]]) against a '''Generative Reconstruction''' (generated via Stable Diffusion v1.5), we aim to quantify the statistical "fingerprints" of the AI across four domains:
			* '''Spatial Statistics:''' Texture and frequency distribution.
			* '''Photometric Statistics:''' Signal-dependent noise response.
			* '''Spectral Statistics:''' Inter-channel color correlation.
			* '''Optical Statistics:''' Point spread function (PSF) and diffraction.

			== References ==
			<references />

← Older revision		Revision as of 21:07, 13 December 2025
Line 4:		Line 4:
	[[File:Sony_Awards_The_Electrician.png\|thumb\|right\|300px\|Figure 1: "The Electrician" by Boris Eldagsen, the AI-generated image that won the Creative category at the 2023 Sony World Photography Awards.]]		[[File:Sony_Awards_The_Electrician.png\|thumb\|right\|300px\|Figure 1: "The Electrician" by Boris Eldagsen, the AI-generated image that won the Creative category at the 2023 Sony World Photography Awards.]]

	In 2023, the Sony World Photography ~~Awards—one of the most prestigious competitions in the field—awarded~~ a prize in the Creative category to an image titled ''"The Electrician"''. The image presented a haunting, black-and-white portrait of two women. However, the artist, Boris Eldagsen, refused the award, revealing that the work was not a photograph at all, but a synthetic creation generated by AI. This event marked a significant turning point: synthetic imagery had crossed a threshold of fidelity where even expert judges could no longer distinguish pixels captured by photons from pixels ~~hallucinated by neural networks~~.[[#ref1\|<sup>[1]</sup>]] This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just ~~'''~~1.5 years~~'''~~, generative AI models have produced as many images as traditional photography produced in its first ~~'''~~150 years~~'''~~ (approximately 15 billion images).[[#ref2\|<sup>[2]</sup>]] We are rapidly entering an era where a significant portion of digital visual data is ~~synthetic~~.		In 2023, the Sony World Photography Awards awarded a prize in the Creative category to an image titled ''"The Electrician"''. The image presented a haunting, black-and-white portrait of two women. However, the artist, Boris Eldagsen, refused the award, revealing that the work was not a photograph at all, but a synthetic creation generated by AI. This event marked a significant turning point: synthetic imagery had crossed a threshold of fidelity where even expert judges could no longer distinguish pixels captured by photons from artificially generated pixels.[[#ref1\|<sup>[1]</sup>]] This blurring of reality is compounded by the unprecedented scale of production. Recent reports indicate that in just 1.5 years, generative AI models have produced as many images as traditional photography produced in its first 150 years (approximately 15 billion images).[[#ref2\|<sup>[2]</sup>]] We are rapidly entering an era where a significant portion of digital visual data is artificially generated, rather than captured on a physical image sensing system.

	[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics illustrating the scale of AI image generation distributed across different models.]]		[[File:AI_Generation_Statistics.png\|thumb\|right\|300px\|Figure 2: Statistics illustrating the scale of AI image generation distributed across different models.]]