” In this video, I utilized artificial intelligence to generate an animated music video for the song Canvas by Resonate. This tool allows anyone to generate beautiful images using only text as the input. My question was, what if I used song lyrics as input to the AI, can I make perfect music synchronized videos automatically with the push of a button? Let me know how you think the AI did in this visual interpretation of the song.
After getting caught up in the excitement around DALL·E2 (latest and greatest AI system, it’s INSANE), I searched for any way I could use similar image generation for music synchronization. Since DALL·E2 is not available to the public yet, my search led me to VQGAN + CLIP (Vector Quantized Generative Adversarial Network and Contrastive Language–Image Pre-training), before settling more specifically on Disco Diffusion V5.2 Turbo. If you don’t know what any of these words or acronyms mean, don’t worry, I was just as confused when I first started learning about this technology. I believe we’re reaching a turning point where entire industries are about to shift in reaction to this new process (which is essentially magic!).
Disco Diffusion (DD) is a Google Colab Notebook which leverages an AI Image generating technique called CLIP-Guided Diffusion to allow you to create compelling and beautiful images from just text inputs. Created by Somnai, augmented by Gandamu, and building on the work of RiversHaveWings, nshepperd, and many others.
Spectral sensitivity of eye is influenced by light intensity. And the light intensity determines the level of activity of cones cell and rod cell. This is the main characteristic of human vision. Sensitivity to individual colors, in other words, wavelengths of the light spectrum, is explained by the RGB (red-green-blue) theory. This theory assumed that there are three kinds of cones. It’s selectively sensitive to red (700-630 nm), green (560-500 nm), and blue (490-450 nm) light. And their mutual interaction allow to perceive all colors of the spectrum.
Most software around us today are decent at accurately displaying colors. Processing of colors is another story unfortunately, and is often done badly.
To understand what the problem is, let’s start with an example of three ways of blending green and magenta:
Perceptual blend – A smooth transition using a model designed to mimic human perception of color. The blending is done so that the perceived brightness and color varies smoothly and evenly.
Linear blend – A model for blending color based on how light behaves physically. This type of blending can occur in many ways naturally, for example when colors are blended together by focus blur in a camera or when viewing a pattern of two colors at a distance.
sRGB blend – This is how colors would normally be blended in computer software, using sRGB to represent the colors.
Let’s look at some more examples of blending of colors, to see how these problems surface more practically. The examples use strong colors since then the differences are more pronounced. This is using the same three ways of blending colors as the first example.
Instead of making it as easy as possible to work with color, most software make it unnecessarily hard, by doing image processing with representations not designed for it. Approximating the physical behavior of light with linear RGB models is one easy thing to do, but more work is needed to create image representations tailored for image processing and human perception.
A LUT (Lookup Table) is essentially the modifier between two images, the original image and the displayed image, based on a mathematical formula. Basically conversion matrices of different complexities. There are different types of LUTS – viewing, transform, calibration, 1D and 3D.
Note: In Foundry’s Nuke, the software will map 18% gray to whatever your center f/stop is set to in the viewer settings (f/8 by default… change that to EV by following the instructions below).
You can experiment with this by attaching an Exposure node to a Constant set to 0.18, setting your viewer read-out to Spotmeter, and adjusting the stops in the node up and down. You will see that a full stop up or down will give you the respective next value on the aperture scale (f8, f11, f16 etc.).
One stop doubles or halves the amount or light that hits the filmback/ccd, so everything works in powers of 2.
So starting with 0.18 in your constant, you will see that raising it by a stop will give you .36 as a floating point number (in linear space), while your f/stop will be f/11 and so on.
If you set your center stop to 0 (see below) you will get a relative readout in EVs, where EV 0 again equals 18% constant gray.
In other words. Setting the center f-stop to 0 means that in a neutral plate, the middle gray in the macbeth chart will equal to exposure value 0. EV 0 corresponds to an exposure time of 1 sec and an aperture of f/1.0.
This will set the sun usually around EV12-17 and the sky EV1-4 , depending on cloud coverage.
To switch Foundry’s Nuke’s SpotMeter to return the EV of an image, click on the main viewport, and then press s, this opens the viewer’s properties. Now set the center f-stop to 0 in there. And the SpotMeter in the viewport will change from aperture and fstops to EV.
A way to approximate complex lighting in ultra realistic renders.
All SH lighting techniques involve replacing parts of standard lighting equations with spherical functions that have been projected into frequency space using the spherical harmonics as a basis.
DISCLAIMER – Links and images on this website may be protected by the respective owners’ copyright. All data submitted by users through this site shall be treated as freely available to share.
