Machine-Learning

OpenAI ChatGPT Reference DeepMind Reference InstructGPT Reference The most talked about topic I’ve seen on Twitter lately is ChatGPT, and many people are raving about it. For a creator to use AI in their creative process to quickly prototype their work is certainly a great thing in the business world. chatGPT can add inspiration to screenwriters, give problem solvers ideas, and give translators help with more accurate language translations. So does ChatGPT count as AI in the true sense of the word, i.e. does it pass the Turing test. Although we don’t know the answer, we can be sure that ChatGPT must have collected huge amount of Internet data for model training, and the parameters of the final model are amazingly huge. ...

Book’s Contents Weta workshop Weta Digital Recently I spent a month reading the two books of weta digital && weta workshop, in which weta digital is mainly engaged in computer 3D image creation, while weta workshop is mainly engaged in pre-film concept creation, such as creature and character sculpture, weapon and prop design, actual concept design of the scene, all these are done by weta workshop. And then weta digital makes CG characters and creatures according to the concept design of workshop, and makes these CG move, what interests me more about workshop is their special effects makeup and fine creature sculpting technology, and weta digital we need to learn their all techniques. The main focus of this article is the digital part, and I have pasted screenshots of the technical points of concern into this article, because each key technical point in time and cost breakthrough will bring a qualitative leap to film post-production, and by breakthrough I mean that the technology can achieve the height of the previous failure at low cost. ...

Paper Deep Learned Super Resolution for Feature Film Production Learning Recently I have been learning about 3D modeling rendering and production, and I saw a paper about Pixar’s release in 2020, the main content is about how Pixar’s internal engineers and developers use deep learning to improve the resolution of rendered images, fine rendering of a 4k image frame may consume a lot of capital cost, if the final rendering of 2k resolution image then use deep neural network to boost it to 4k, the image is more detailed and richer, then it can help companies reduce a lot of the cost of rendering. Unlike the previous image super-resolution enhancement model, the film industry typically uses images that contain data with high dynamic range of lighting, and true reproduction of color, which makes the current mainstream super-resolution models simply unusable in the film domain. This is an area of interest to me at the moment, as using deep learning for industrial applications is more practical than theoretical. ...

Back in 2020 I wrote a blog post Renderman24 Will Support Blender,at the end of which I documented my intention to purchase a commercial version of Renderman. after more than a year, I realized my idea, configured myself with an image workstation, purchased a commercially copyrighted Renderman, and most importantly, I also purchased a set of EIZO CG319x monitors for myself. The initial reason for preparing all this was mainly because of the rise of web 3.0. I envisioned a one-person freelance career, because the film industry is created through a very large group of people, and it is very time-consuming for one person to complete a live-action film or an animated feature film, or even a short film, independently, because there are many technical as well as artistic stumbling blocks in front of you, and you have to overcome them little by little to complete every detail of the film. Then I thought about the possibility of using the machine’s algorithm to set up the characters’ environment, to generate life and environment in the computer’s world, which is called modeling and animation and scene production in conventional film production. Dare to imagine if the computer can complete these creative processes, such as animal jumping and running, model materials, keying, human walking, we are no longer through the drawing of key frames, debugging materials, manual green screen keying, but through the computer’s learning of the real world to actively explore these key frames, to achieve the simulation of real-world object materials, the precise keying of the characters in the picture, then it seems that a person It is no problem to create a wonderful movie. Of course, this is just an idea, and the technology is not yet so sci-fi. ...

Abstract This is an article from 2020. After recording the title of the article, I didn’t write the content, but now I’m taking advantage of the vacation time to give a detailed description of Matte Painting, a magical field in which we create fantasy worlds inside our own minds. This article is about the history of Matte Painting and the techniques of Matte Painting. History Wikipedia Matte Painting Matte painting is the painting of landscapes, sets, and vistas, often allowing the filmmaker to create environments in the film that do not exist in the location, combining the painted image with live footage. ...