Excellent artistic work frequently showcases the individual inventiveness of the creator, and many artists place a high value on producing images of superior quality. A remarkable success has been achieved by GANs in producing photorealistic 2-D images. Because traditional GANs generate 3D images directly from 2D, their images frequently lack the perfect quality and consistency of multiple views from a single image. With varied degrees of success, numerous attempts have been made to address these issues. With the use of a 3D Variational Autoencoder (VAE) that integrated Improved Wasserstein Generative Adversarial Networks (3D VAE-IWGAN), this research aims to produce high-quality and consistent multi-view three-dimensional (3D) artistic images. The experiments were conducted using the COCO Africa Mask dataset, and the suggested method produced high-quality images that were more closely aligned with the aesthetic qualities of authentic works of art than other complex methods. In comparison to other methods, the proposed method has a high Inception Score(IS) score of 26.37, a low fréchet inception distance (FID) score of 12.09, a peak signal-to-noise ratio (PSNR) of 31.35, a mean squared error (MSE) of 65.32, a chamfer distance (CD) of 0.243, Learned Perceptual Image Patch Similarity (LPIPS) score of 0.134 and a structural similarity index measure (SSIM) of 0.772. These results are better than those obtained with other methods on the COCO Africa Mask dataset. The suggested approach can help artists showcase, develop, and improve the originality of their work, which will increase their financial gains.