{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/precomputed-real-time-texture-synthesis-with","title":"Precomputed Real-Time Texture Synthesis with Markovian Generative Adversarial Networks","arxiv_id":"1604.04382","date":"2016-04-15","proceeding":null,"authors":["Chuan Li","Michael Wand"],"abstract":"This paper proposes Markovian Generative Adversarial Networks (MGANs), a\nmethod for training generative neural networks for efficient texture synthesis.\nWhile deep neural network approaches have recently demonstrated remarkable\nresults in terms of synthesis quality, they still come at considerable\ncomputational costs (minutes of run-time for low-res images). Our paper\naddresses this efficiency issue. Instead of a numerical deconvolution in\nprevious work, we precompute a feed-forward, strided convolutional network that\ncaptures the feature statistics of Markovian patches and is able to directly\ngenerate outputs of arbitrary dimensions. Such network can directly decode\nbrown noise to realistic texture, or photos to artistic paintings. With\nadversarial training, we obtain quality comparable to recent neural texture\nsynthesis methods. As no optimization is required any longer at generation\ntime, our run-time performance (0.25M pixel images at 25Hz) surpasses previous\nneural texture synthesizers by a significant margin (at least 500 times\nfaster). We apply this idea to texture synthesis, style transfer, and video\nstylization.","url_abs":"http://arxiv.org/abs/1604.04382v1","url_pdf":"http://arxiv.org/pdf/1604.04382v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"precomputed-real-time-texture-synthesis-with","repo_url":"https://github.com/chuanli11/MGANs","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"torch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"precomputed-real-time-texture-synthesis-with","repo_url":"https://github.com/soumik12345/Pix2Pix","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"style-transfer","task_name":"Style Transfer"},{"task_slug":"texture-synthesis","task_name":"Texture Synthesis"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1604.04382","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}