Exploring Deepfake Technology: Sin or Boon for Filmmaking
Overview
In an age where technology continuously blurs the line between reality and fiction, the emergence of deepfake technology has sparked both excitement and apprehension within the realm of cinema. Deepfake, a portmanteau of "deep learning" and "fake," refers to the manipulation of audiovisual content using advanced machine learning algorithms. Initially gaining notoriety for its potential misuse in spreading misinformation and deception, deepfake technology has also opened up a plethora of creative possibilities within the film industry. This article by Academic Block looks into the multifaceted role of deepfake technology in shaping the future landscape of cinema.
What is Deepfake Technology?
The term "deepfake" is derived from a combination of “deep learning” and “fake,” where deep learning refers to a subset of AI techniques. These techniques, particularly Generative Adversarial Networks (GANs), play a crucial role in the creation of deepfakes. GANs are made up of two neural networks: the generator and the discriminator. The generator’s job is to create synthetic data, like an image or video, while the discriminator evaluates its authenticity. The two networks work together in a feedback loop, continuously improving each other’s performance.
Deepfake technology typically involves the following steps:
-
Data Collection : To create a deepfake, the first step is gathering a large number of high-quality images or videos of the target individual. These images will serve as the foundation for building a digital model of the person’s face or voice.
-
Training the Model : Deep learning models are trained using the collected data. The model learns to understand the facial expressions, movements, and unique features of the subject, allowing it to replicate these characteristics on a different body or in a new scenario.
-
Synthesis of Fake Media : Once the model is trained, it can be used to generate realistic videos or audio clips, replacing the original subject with another individual or modifying their voice.
-
Fine-Tuning : To make the final output more realistic, the synthetic media undergoes further processing to improve the synchronization of lip movements, lighting, and background consistency.
The Role of Deepfake Technology in Filmmaking
Filmmakers have always sought innovative ways to tell compelling stories, often using groundbreaking technology to enhance the visual and auditory experience. Deepfake technology has introduced a new realm of possibilities in filmmaking, offering both creative opportunities and challenges.
1. Digital De-Aging and Resurrecting Actors
One of the most exciting uses of deepfake technology in filmmaking is the ability to digitally de-age actors or even resurrect deceased ones. Studios can now use AI to create young versions of actors, eliminating the need for casting younger performers or using makeup for aging effects.
For example, in movies like The Irishman (2019), Martin Scorsese utilized de-aging technology to make actors like Robert De Niro, Al Pacino, and Joe Pesci appear younger. While this wasn’t purely a deepfake, the technique shares similar principles, where digital models were used to manipulate the actors’ faces. With further advancements in deepfake technology, filmmakers can now create even more realistic digital de-aging, enhancing both the storytelling and production process.
Moreover, deepfake technology opens the door to resurrecting actors who have passed away. With the permission of the actor’s estate, filmmakers could potentially bring deceased stars back for special roles or cameo appearances. This capability raises ethical questions about consent and authenticity, but it could also transform how films are made in the future.
2. Costume and Makeup Alternatives
Traditional methods of special effects makeup and costume design can be time-consuming and costly. Deepfake technology offers an alternative, allowing filmmakers to digitally alter an actor’s appearance without the need for prosthetics, wigs, or extensive makeup.
For instance, an actor could wear minimal makeup and perform in front of a green screen, with their appearance later digitally altered to achieve the desired look. This not only saves time and money but also offers greater flexibility in terms of character design. Whether creating fantastical creatures, supernatural beings, or historical figures, deepfake technology can allow filmmakers to generate more realistic transformations.
3. Enhanced Visual Effects (VFX)
Deepfake technology can significantly enhance visual effects, providing filmmakers with the ability to create more convincing and immersive scenes. While traditional VFX work often requires creating digital doubles for complex stunts or action sequences, deepfakes can be used to seamlessly integrate real human faces into VFX-heavy environments.
For example, instead of relying on stunt doubles for dangerous scenes, a director could use deepfake technology to swap in an actor’s face onto the body of a stunt performer. This results in a more lifelike and convincing shot, without the need for complicated visual effects or risky stunts.
4. Voice Synthesis and Audio Manipulation
In addition to altering visual elements, deepfake technology is also capable of synthesizing realistic voice recordings. By training AI models on an actor's voice, filmmakers can create new lines of dialogue, ensuring that the actor’s voice matches the character’s speech patterns, even in situations where the actor cannot perform.
This capability is particularly useful when an actor is unavailable for reshoots or when a character needs to speak in a language the actor doesn’t know. By using voice deepfake technology, filmmakers can maintain consistency and authenticity without requiring the actor to return to the set.
5. Creating "What-If" Scenarios
Deepfake technology also offers filmmakers the ability to create "what-if" scenarios, allowing them to experiment with alternative takes, endings, or character portrayals without re-shooting entire scenes. For example, a filmmaker might be interested in seeing how a particular actor would have performed a specific scene but cannot afford the time or cost of a reshoot. Deepfakes make it possible to digitally swap out actors or alter performances in post-production, offering greater creative freedom.
This has the potential to significantly reduce production costs, particularly for studios working on large-scale projects with many reshoots or alternative versions.
Pros of Deepfake Technology
-
Creative Freedom in Filmmaking : Deepfake technology allows filmmakers to explore new creative possibilities, such as digital de-aging or resurrecting deceased actors, making storytelling more dynamic and innovative.
-
Cost and Time Efficiency : By using deepfake technology, filmmakers can reduce the costs and time associated with reshoots, stunt doubles, and makeup. This results in more efficient production processes.
-
Enhanced Special Effects : Deepfakes provide a more realistic integration of visual effects, such as facial replacements or seamless action sequences, making films more immersive.
-
Voice Synthesis and Audio Enhancement : AI-generated voice synthesis can replace or enhance actors' voices, allowing filmmakers to maintain consistency or even create new dialogue for a character without requiring the actor to return to the set.
-
Historical Accuracy : Deepfakes enable filmmakers to accurately portray historical figures or recreate iconic scenes with lifelike precision, enhancing the authenticity of biographical films and documentaries.
Cons of Deepfake Technology
-
Potential for Misinformation : Deepfakes can be used to manipulate videos or audio in misleading ways, creating false narratives or political manipulation, leading to widespread misinformation.
-
Ethical Concerns and Consent Issues : The use of deepfake technology without an actor’s consent, particularly with deceased actors, raises serious ethical questions and concerns over privacy rights.
-
Impact on Employment : As AI-generated content becomes more sophisticated, there is a concern about job displacement within the film and entertainment industries, as virtual actors and digital doubles could replace traditional roles.
-
Security Risks : Deepfakes can be exploited for identity theft, fraud, and cybercrime, as fake videos or audio can be used to impersonate individuals for malicious purposes.
-
Quality Control Challenges : Despite advancements, deepfakes are not always flawless, and creating convincing content requires high expertise. Low-quality deepfakes can lead to a loss of authenticity and credibility in media.
The Ethical Concerns of Deepfake Technology in Filmmaking
While deepfake technology offers exciting possibilities in filmmaking, it also raises significant ethical concerns. Here are some key issues filmmakers and audiences should consider:
1. Consent and Privacy
One of the most controversial aspects of deepfakes is the potential for misuse, particularly in the realm of consent and privacy. The ability to digitally recreate an individual’s likeness and voice raises concerns about the exploitation of public figures, including actors. Unauthorized use of an actor’s image or voice could lead to legal disputes, ethical dilemmas, and potential harm to a person’s reputation.
In filmmaking, studios must ensure that they have explicit consent from actors and their estates when using deepfake technology to alter their likeness or voice. This includes careful management of how and where the digital content is used, especially in the case of deceased actors.
2. Misinformation and Manipulation
The widespread use of deepfake technology also poses a threat to the authenticity of media content. As deepfakes become more realistic, it becomes increasingly difficult to distinguish between real and fake footage. This could lead to the spread of misinformation, particularly in political or social contexts.
Filmmakers need to be mindful of the potential consequences of creating deepfake content, particularly when using it to depict real-life events or individuals. While artistic freedom is crucial, filmmakers must weigh the ethical implications of producing content that could be easily misinterpreted as factual.
3. Impact on the Acting Industry
With the ability to digitally recreate actors and manipulate their performances, deepfake technology raises questions about the future of the acting profession. If studios can simply use AI to generate performances, it may impact the demand for traditional actors. This could lead to job losses and a shift in how films are made and consumed.
Furthermore, there are concerns about whether AI-generated performances can truly capture the nuance and emotional depth of human actors. While deepfake technology can create realistic appearances, the emotional resonance and complexity of human performances remain a crucial element of filmmaking.
The Future of Deepfake Technology in Filmmaking
Despite the ethical challenges, the future of deepfake technology in filmmaking looks promising. As AI continues to evolve, filmmakers will find even more innovative and creative ways to integrate this technology into their projects. Whether it's enhancing visual effects, improving storytelling, or creating unique digital experiences, deepfake technology is poised to play a significant role in the future of film production.
However, it is essential for the industry to establish clear ethical guidelines and regulations surrounding its use. By prioritizing consent, transparency, and responsible use, filmmakers can ensure that deepfake technology is used in ways that enhance the art of filmmaking while minimizing potential harm.
Final Words
Deepfake technology is transforming the way films are made, offering filmmakers new tools for creativity and storytelling. From de-aging actors to creating realistic visual effects and voice synthesis, the possibilities are vast. However, as with any powerful technology, it comes with a host of ethical and practical challenges that need to be addressed.
As deepfake technology continues to evolve, it will undoubtedly play a larger role in shaping the future of filmmaking. By embracing these innovations responsibly, filmmakers can continue to push the boundaries of storytelling, while maintaining a focus on ethics and authenticity. Hope you enjoyed reading with Academic Block, please provide your thoughts in the comment section to make this article better. Thanks for Reading!
This Article will answer your questions like:
Deepfake is a form of artificial intelligence that manipulates audio, video, or images to create realistic but entirely fake content. It uses machine learning algorithms, specifically deep learning models, to superimpose someone's face, voice, or actions onto another individual's, creating highly convincing fake visuals or sounds. The technology has gained attention for both its innovative uses and potential for abuse, especially in entertainment and media industries.
Deepfake technology in filmmaking allows filmmakers to digitally manipulate footage, creating highly realistic simulations of actors, locations, or scenes. This technique is commonly used for de-aging characters, resurrecting deceased actors, or enabling creative visual effects. It can streamline production by reducing the need for physical reshoots and makeup, while also enhancing special effects with more realistic integration of digital elements into live-action sequences.
Deepfakes themselves are not inherently illegal, but their use can violate laws depending on context. For instance, creating deepfakes to spread misinformation, defame someone, or infringe on intellectual property rights can result in legal consequences. Several countries have started passing legislation specifically targeting malicious deepfake uses, such as those related to harassment, political manipulation, and identity theft, making certain applications of deepfakes illegal.
Detecting a deepfake often involves examining visual inconsistencies such as unnatural facial expressions, irregular lighting, or pixel-level flaws in the video. Tools like deepfake detection software use AI to analyze characteristics that are difficult for deepfake algorithms to replicate, such as blink rates or lip movements. Manual inspection of the content’s context and cross-checking with credible sources can also help confirm whether a video is manipulated.
To detect if a photo is a deepfake, you should look for signs like inconsistent lighting, unnatural shadows, or odd distortions around the edges of the face. Deepfake photos might also exhibit blurred or mismatched details in the background. Using reverse image search or employing specialized deepfake detection tools can help confirm whether the image has been digitally altered. As deepfake technology improves, these signs become harder to detect without advanced AI tools.
Downloading deepfakes is not illegal in itself, but how they are used can lead to legal issues. If a deepfake is used to defraud, harass, or deceive, or if it violates copyright, it can result in criminal or civil charges. Additionally, downloading and sharing deepfakes that involve non-consensual manipulation of individuals' likenesses, such as in explicit content, may lead to severe legal consequences in many jurisdictions.
Deepfake technology offers numerous benefits, particularly in the entertainment and film industries. It allows for enhanced visual effects, such as realistic character de-aging or the resurrection of deceased actors. It can also improve the efficiency of film production by reducing the need for reshoots and makeup, while enabling filmmakers to create more dynamic and immersive storytelling. Additionally, deepfakes can be used in education and training simulations to create realistic scenarios.
The primary drawbacks of deepfake technology include ethical concerns, potential misuse for spreading misinformation or defamation, and the risk of eroding public trust in media. In the film industry, while deepfakes can streamline production, they could also replace traditional jobs, such as actors and makeup artists, leading to concerns about job displacement. Additionally, the rise of deepfakes contributes to the growing challenges in verifying the authenticity of digital content.
Yes, deepfake technology is increasingly being used in filmmaking to create realistic visual effects, such as de-aging characters or digitally resurrecting deceased actors. It is also used for enhancing special effects in action sequences and generating realistic stunt doubles. The flexibility of deepfake technology allows filmmakers to reduce production costs and time, offering new opportunities for storytelling and cinematic experiences.
Deepfake videos can be highly accurate, often indistinguishable from real footage, depending on the quality of the technology used and the source material. However, even the best deepfakes sometimes show subtle inconsistencies such as unnatural facial expressions, mismatched lighting, or distorted backgrounds. As deepfake technology improves, its ability to produce convincing and lifelike videos continues to evolve, making it increasingly difficult to detect without advanced detection tools.
Deepfake technology is transforming the entertainment industry by enabling more advanced visual effects and cost-efficient production. It allows filmmakers to create realistic scenes without reshoots or expensive prosthetics, offering creative flexibility. Additionally, it raises ethical concerns regarding the manipulation of actors' likenesses and the potential for misinformation. As a result, the industry must balance innovation with responsible use to ensure deepfake technology is utilized ethically and professionally.
Risks of using deepfake technology in the film industry include the potential for exploitation, such as using actors’ likenesses without consent, or creating misleading representations. It also poses a threat to job security in the film industry, replacing traditional techniques and actors. Additionally, deepfakes may contribute to public confusion and misinformation, as audiences might struggle to distinguish between real and manipulated content, raising concerns about trust in media and intellectual property.
Notable examples of films using deepfake technology include “The Irishman,” where the technology was used for de-aging actors like Robert De Niro. Another example is “Rogue One: A Star Wars Story,” where deepfake techniques were applied to bring back the likeness of Peter Cushing. These films illustrate how deepfake technology is used to enhance storytelling and create lifelike visual effects, pushing the boundaries of traditional filmmaking techniques.
Currently, there are limited regulations specifically governing the use of deepfake technology in filmmaking, although laws are being developed in various regions to address potential abuses. Issues such as consent, intellectual property rights, and digital manipulation are under scrutiny. Filmmakers and production companies are expected to follow ethical guidelines to ensure that deepfake technology is used responsibly, particularly concerning the manipulation of actors' likenesses and avoiding misinformation or harm.
Deepfake technology differs from traditional face-swapping techniques by utilizing machine learning algorithms to generate highly realistic images and videos. While face-swapping often relies on manual editing or simpler software, deepfakes use neural networks to analyze and replicate facial movements, expressions, and voices. This results in a more seamless and convincing output, making deepfake technology much more advanced and harder to detect than traditional face-swapping methods.
Famous instances of deepfake technology used in cinema
“Star Wars: The Rise of Skywalker” (2019): In this installment of the “Star Wars” saga, deepfake technology was utilized to digitally recreate the likeness of the late Carrie Fisher as Princess Leia. The technology was used to integrate Fisher’s likeness into scenes that were filmed after her passing, allowing her character to appear in key moments of the film.
“The Irishman” (2019): Directed by Martin Scorsese, “The Irishman” employed de-aging visual effects to portray actors Robert De Niro, Al Pacino, and Joe Pesci at various stages of their characters’ lives. While not strictly deepfake technology, this film utilized similar digital manipulation techniques to alter the appearance of the actors, creating a seamless transition between different time periods.
“Gemini Man” (2019): Directed by Ang Lee, “Gemini Man” featured extensive use of visual effects to create a younger digital double of actor Will Smith. While not explicitly labeled as deepfake technology, the film relied on advanced CGI and motion capture techniques to seamlessly integrate the digital double into action sequences alongside the real actor.
“Finding Jack” (TBA): In a controversial move, the producers of the film “Finding Jack” announced plans to digitally resurrect actor James Dean using deepfake technology to portray a supporting role. The decision sparked widespread debate over the ethics of using deepfake technology to recreate deceased actors’ likenesses for commercial purposes.
Experimental Shorts and Art Projects: Outside of mainstream cinema, there have been various experimental shorts and art projects that explore the creative potential of deepfake technology. These projects often use deepfake technology to manipulate existing footage or create entirely synthetic content, pushing the boundaries of traditional filmmaking techniques.
Benefits of Deepfake Technology for filmmakers
Cost Efficiency: Traditional filmmaking techniques often require significant investments in sets, props, costumes, and makeup effects. Deepfake technology provides a more cost-effective alternative by allowing filmmakers to achieve comparable results using digital manipulation. This can lead to substantial savings in production budgets, making ambitious projects more financially viable.
Time Savings: The process of creating practical effects and editing footage can be time-consuming, causing delays in production schedules. Deepfake technology streamlines the process by automating tasks such as face replacement and scene compositing, reducing the time needed to complete complex visual effects shots. This enables filmmakers to work more efficiently and meet tight deadlines without compromising on quality.
Flexibility in Casting: Deepfake technology offers filmmakers greater flexibility in casting choices, as actors can be digitally transformed to better fit their roles. This versatility allows filmmakers to explore diverse narratives and characters without being limited by the physical attributes of their performers. It also enables actors to portray multiple roles or personas within the same production, expanding their range and versatility.
Preservation of Performances: Deepfake technology enables the preservation and enhancement of performances by actors, both living and deceased. By digitally recreating their likeness, filmmakers can ensure that iconic performances are immortalized for future generations to enjoy. This can be particularly valuable in projects that require the participation of historical figures or actors who have passed away.
Visual Effects Innovation: Deepfake technology empowers filmmakers to push the boundaries of visual effects and computer-generated imagery (CGI). By seamlessly integrating digital elements with live-action footage, filmmakers can create immersive worlds, fantastical creatures, and larger-than-life spectacles with unprecedented realism. This opens up new creative possibilities and allows filmmakers to bring their wildest imaginations to the screen.
Historical and Educational Applications: In historical and educational films, deepfake technology can be used to recreate pivotal moments and figures with remarkable accuracy. By seamlessly integrating actors into archival footage or photographs, filmmakers can transport audiences back in time and provide immersive learning experiences. This enables filmmakers to explore complex historical events and personalities in ways that are engaging and informative.
Creative Freedom: Deepfake technology empowers filmmakers to experiment with unconventional storytelling techniques and visual aesthetics. By blurring the line between reality and fiction, filmmakers can craft compelling narratives that challenge societal norms and push the boundaries of audience expectations. This freedom allows filmmakers to explore bold ideas and unconventional narratives without being constrained by traditional filmmaking conventions.
Accessibility: The democratization of deepfake technology has made advanced visual effects tools more accessible to independent filmmakers and content creators. With user-friendly software and tutorials readily available online, aspiring artists can experiment with deepfake technology and bring their creative visions to life without the need for extensive training or resources. This levels the playing field and allows filmmakers of all backgrounds to compete on equal footing in the industry.
Drawbacks of deepfake technology for filmmakers
Ethical Concerns: Deepfake technology raises significant ethical concerns, particularly regarding the potential for misuse and manipulation. The ability to create highly realistic synthetic content raises questions about consent, privacy, and the authenticity of digital media. Filmmakers must navigate these ethical considerations carefully to ensure that their use of deepfake technology does not infringe upon the rights or dignity of individuals depicted in their work.
Legal Issues: The use of deepfake technology in filmmaking may raise legal issues related to copyright infringement, defamation, and misrepresentation. Filmmakers must ensure that they have the appropriate rights and permissions for any content they manipulate using deepfake technology, especially when it involves the likeness of real people or existing intellectual property.
Quality Concerns: While deepfake technology has advanced significantly in recent years, it still has limitations in terms of quality and realism. In some cases, deepfake videos may exhibit visual artifacts or inconsistencies that detract from the overall viewing experience. Filmmakers must carefully consider the limitations of deepfake technology and weigh the trade-offs between realism and technical feasibility in their work.
Overreliance on Technology: Relying too heavily on deepfake technology can stifle creativity and innovation in filmmaking. Filmmakers may become overly reliant on digital manipulation to achieve desired effects, rather than exploring alternative approaches or embracing the limitations of practical effects.
Security Risks: Deepfake technology poses security risks related to the manipulation and dissemination of digital media. Malicious actors may use deepfake technology to create fake videos for the purpose of spreading misinformation, defaming individuals, or committing fraud.
Social Implications: The proliferation of deepfake technology in filmmaking has broader social implications for how audiences perceive reality and truth. The prevalence of synthetic media may erode trust in traditional forms of communication and media, leading to increased skepticism and uncertainty in society.
Risk of deepfake technology in Cinema
Misinformation and Deception: One of the most significant risks of deepfake technology is its potential for misuse in spreading misinformation and deception. Deepfake videos can be used to create convincing fake news, hoaxes, or propaganda, leading to widespread confusion and distrust in the authenticity of digital media.
Privacy Infringement: Deepfake technology raises concerns about privacy infringement, as it enables the manipulation of individuals’ likeness and voice without their consent. Actors, public figures, and ordinary people alike may become targets of deepfake manipulation, leading to violations of their privacy and autonomy.
Identity Theft and Fraud: Deepfake technology poses risks of identity theft and fraud, as malicious actors can use manipulated videos to impersonate others or commit financial crimes. By convincingly imitating someone’s appearance and voice, deepfake videos can be used to deceive individuals or organizations for illicit purposes.
Defamation and Reputation Damage: Filmmakers must be aware of the potential legal risks associated with deepfake technology, including defamation and reputation damage. Deepfake videos that falsely depict individuals engaging in illegal, immoral, or scandalous behavior can cause significant harm to their reputation and livelihood.
Cultural Sensitivity and Representation: Deepfake technology raises concerns about cultural sensitivity and representation in cinema. The manipulation of actors’ appearances, voices, and identities using deepfake technology may perpetuate stereotypes or reinforce harmful narratives, leading to backlash from marginalized communities and audiences.
Erosion of Trust: The proliferation of deepfake technology in cinema may erode trust in traditional forms of media and communication. Audiences may become increasingly skeptical of the authenticity of digital media, leading to a loss of confidence in the information presented to them.
Psychological Impact: Deepfake videos have the potential to cause psychological harm to individuals who are targeted or depicted without their consent. The knowledge that one’s likeness and voice can be manipulated without their control may lead to feelings of anxiety, insecurity, and vulnerability.
Legal and Regulatory Challenges: Filmmakers must navigate complex legal and regulatory frameworks when using deepfake technology in their work. Issues such as copyright infringement, defamation, and privacy rights may arise, requiring filmmakers to obtain appropriate permissions.
Rules and Regulations for the use of deepfake technology
Copyright Law: Deepfake technology often involves the manipulation of existing audiovisual content, which may raise issues related to copyright infringement. Filmmakers using deepfake technology must ensure that they have the appropriate rights and permissions for any copyrighted material they incorporate into their work.
Privacy and Consent: Deepfake technology can be used to manipulate individuals’ likeness and voice without their consent, raising concerns about privacy infringement. Filmmakers must obtain consent from individuals before using their likeness in deepfake videos, especially if the videos are intended for commercial or public distribution.
Defamation and Misrepresentation: Deepfake technology poses risks of defamation and misrepresentation, as manipulated videos can falsely depict individuals engaging in illegal, immoral, or scandalous behavior. Filmmakers must ensure that their use of deepfake technology does not violate the rights or reputation of individuals depicted in their work.
Fraud and Misuse: Deepfake technology can be used for fraudulent purposes, such as impersonating others or creating fake videos for malicious intent. Filmmakers must refrain from using deepfake technology for illegal or unethical purposes and adhere to ethical standards and guidelines in their work.
Academic References on the Deepfake Technology
- Li, Y., & Lyu, S. (2020). DeepFake Video Detection Using Recurrent Neural Networks. IEEE Transactions on Image Processing, 29(11), 7832-7841.
- Wu, B., & Cosman, P. C. (2020). Deepfake detection using temporal information. Journal of Electronic Imaging, 29(2), 1-8.
- Nguyen, T., & Hoang, A. (2020). Deepfake videos detection based on MesoNet model. Journal of Electronic Imaging, 29(6), 1-8.
- Shen, X., & Tian, Y. (2019). Attention-aware compositional network for video-based person re-identification. IEEE Transactions on Image Processing, 28(11), 5541-5552.
- Rossler, A., et al. (2019). FaceForensics++: Learning to Detect Manipulated Facial Images. IEEE Transactions on Information Forensics and Security, 15, 537-553.
- Zhou, Y., & Ye, Q. (2020). Learning deep features for discriminative localization. IEEE Transactions on Image Processing, 29(1), 3706-3717.
- Guera, D., & Martinez, A. M. (2020). On the effectiveness of forensic techniques to detect deepfakes. IEEE Transactions on Information Forensics and Security, 15(1), 308-319.
- Piryonesi, S. M., & Gabbouj, M. (2020). Video forensic detection using convolutional long short-term memory networks. IEEE Transactions on Information Forensics and Security, 15(1), 369-382.
- Rössler, A., et al. (2020). FaceForensics++: Learning to Detect Manipulated Facial Images. IEEE Transactions on Information Forensics and Security, 15(1), 537-553.
- Tolosana, R., et al. (2020). Deep learning for digital media forensic: recent advances and challenges. IEEE Access, 8, 25461-25492.
- Nie, S., et al. (2020). A lightweight detection method for deepfake videos based on optical flow and convolutional neural networks. IEEE Access, 8, 166294-166303.
- Afchar, D., et al. (2018). Mesonet: a compact facial video forgery detection network. Proceedings of the IEEE International Conference on Computer Vision, 4895-4904.
- Zhou, Y., & Ye, Q. (2018). Learning deep features for discriminative localization. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2921-2929.
- Zhou, Z., & Wang, C. (2019). Deep learning on irregularly sampled signals: a survey. IEEE Transactions on Signal Processing, 67(5), 1305-1322.