The rapid advancement in machine learning and artificial intelligence has significantly enhanced capabilities in multimedia content creation, particularly in the domain of deepfake generation. Deepfakes leverage complex neural networks to create hyper-realistic manipulated audio-visual content, raising profound ethical, societal, and security concerns. This paper presents a comprehensive survey of contemporary trends in deepfake video research, focusing on both generation and detection methodologies. The study categorizes deepfakes into three primary types: facial manipulation, lip-synchronization, and audio deepfakes, further subdividing them into face swapping, face generation, attribute manipulation, puppeteering, speech generation, and voice conversion. For each type, the paper reviews cutting-edge generation techniques, including StyleGANs, variational autoencoders, and various speech synthesis models. It also presents an in-depth analysis of detection methods, highlighting both traditional handcrafted feature-based approaches and modern deep learning frameworks utilizing CNNs, RNNs, attention mechanisms, and hybrid transformer models. The paper evaluates these methods in terms of performance, generalization, robustness, and limitations against evolving deepfake techniques. The survey identifies significant challenges such as vulnerability to adversarial attacks, lack of generalized models, and dependency on high-quality training data. The insights provided aim to aid researchers and practitioners in developing more robust detection mechanisms and understanding the landscape of deepfake threats and countermeasures. Ultimately, this study contributes to the growing body of literature by mapping current trends and suggesting potential avenues for future research in combating deepfake proliferation.