The most difficult part for the BF (in something like this) would be to train the network through hundreds and hundreds of renders (in order to generate the needed parameters so the code can resolve any situation and any scene you throw at it). Realistically when will this feature get built into Blender? Or at least thats the way i understand it. They use machine learning to train a neural network algorithm to learn the final average color value of each pixel after i guess some rough starting data like 10 samples, and then it uses the algorithm to skip straight to the predicted final value for each pixel, resulting in no noise and a render that looks like its gone through thousands of samples. The maxed out settings is still much better but also needs to run 30-60 minutes depending on resolution.It’s the new best thing since sliced bread for rendering. The sweet spot on credits per quality is the medium setting. That said - it can still produce amazing art in “low” setting. As the Cronos Blend uses more compute resources than Hyperion, the free-tier setting for Cronos is “low quality”. The Cronos Blend allows you to select a quality level. There are diminishing returns after 500 though - the image will not get much better anymore in most cases. Where smaller numbers will make it more abstract. The sweet spot seems to be between 200 and 500. Those are the number of steps performed to optimized the image against the input text prompt. The Hyperion Blend has the concept of iterations. As this AI is an uscaler, it does not result in major changes of the image composition but only adds finer details to the output. It is also possible to use a seocndary AI to upscale that 4 times. This reflect the native internal resolution of the AI model. You can select a low resolution - typical 256 x 256 pixel depending on the aspect ration, or a high resolution, 512 x 512 pixel. This settings is currently not available for the Cronos Blend which will always be square. Default is a square image, a 16:9 landscape format and 9:16 portrait format are also available. The Hyperion Blend allows you to select an aspect ratio. So it does help to produce multiple images with the exact same input. Other than the default Hyperion Blend, the Cronos Blend produces very different results even with the exact same input. Good results can be amazing but sometimes it also produces results that seem not too much related to the input. Cronos BlendĬronos Blend is a new model that can be a bit of a hit-and-miss. Hyperion Blend is the deault settings and brings you predictible good results. Each model has it’s own unique character. Selecting a “blend” allows you to select a different kind of AI model that NeuralBlender uses.
#NEURAL BLENDER FREE#
NeuralBlender has a free tier that allows anyone to create images for free.ĭepending on the settings you select, credits might be required and will automatically select the “premium tier”. Screenshot of the main NeuralBlender settings: You can learn about it later step by step once you got the basics. If you are new to NeuralBlender I would suggest to just enter a text in the text input field - for example: “a verry happy cat” - and press “go!”. This article will show you how the NeuralBlender settings work.
0 kommentar(er)
