Evaluating Color

From RMIT Visual Effects
Jump to: navigation, search

Color Evaluation in the Panels

Color evaluation can be done within Nuke's panels, without calling on any nodes.

The Viewer and the RGB values

At the bottom of the Viewer window is the value information bar (pictured on this page). It displays a set of numbers that represent the color values of the region of the image that lies directly under the cursor. A shift drag in the viewer window will create a box from which the average values can be derived (click ones again in the viewer window to get rid of the red square).

The Viewer and the f-stop and Gamma sliders

These two sliders in the Viewer window (pictured on this page) change their two eponymous associated values. The change is limited to what is presented in the viewer window and has no affect upon the actual rendered output. If the f-stop value is moved really high, all but the very darkest values will be 'blown out'. If the gamma is moved really low, all but the highest values will be lost to black. These two actions are great for checking the dark and light points of the RGB and of check for noise in alphas. Warning: don't forget to set them back to their default values after using them.

The Pixel Analyser

The color values of a region of an image may be acquired from within the Viewer window (see here for details). The Pixel Analyser does the same thing, but offers far more functionality.

The pixel analyser will present the color values of any pixel that lays underneath a Command click in the Viewer window. This value is visualised in the 'current' preview box, located on the left hand side of the analyser. The RGB and HSL values are presented as numerical values. More usefully, it will also present the minimum, maximum, average and median values. For single pixel values, this is irrelevant. However, a region of colors can be selected by Command Shift clicking in the viewer window. Also... a Command Drag will return a trail of points, which is useful for non-rectangular selections.

Pixel analyzer.png

The Scopes

The Scopes panel is relatively new to Nuke. It offers a standard set of tools for evaluating the color of an image.

The Vectorscope

This visualises the hue and saturation values of an image. In the middle there is a lump that extends from the centre. The distance that the lump extends from the centre represents the saturation of the image: the further out, the more saturated the image is. The orientation of the lump (i.e. what 'o'clock' it is) represents the hue values. In the example shown, the image is clearly very blue and cyan, with some purple and a tiny bit of yellow.


The Vectorsope

This visualises both the RGB values of an image, and its luminance (i.e. its brightness). I most commonly use it for the later. I will describe how to use this tool in class.


The Histogram

Histograms are found in multiple applications. They are most commonly used to evaluate the lightness values of an image. Histograms are described in this page (from my Photoshop wiki).


Color Evaluation Nodes

The following nodes may be used to evaluate the color values of an image. Some of them are designed for that specific purpose, others may be co-opted into that role.

The CurveTool node

The CurveTool node (below) presents both numerical and visual feedback on the color values of an image, a section of an image or (usefully) a moving image. The results that it gives are outputted as a set of animation curves which can be referenced by other nodes that actively work upon the image or image sequence (e.g. the lightness curve from flickering footage can be inverted so as to remove the flicker). The most important tab in the CurveTool is the MaxLumaData which shows the values and locations of the brightest and darkest points in the image. To use set the pull down menu in the CurveTool tab to Max Luma Pixel and press the big Go! button, then navigate to the MaxLumaData tab where the data will be found. This node gives absolute numerical values, not ‘best judgment’ values so the results can sometimes be a bit twitchy.

Feedback 4.png

The Sampler node

The Sampler node gives visual feedback on the R,G and B values (below) that lay along a horizontal cross section of the image the position of which is determined by a little widget that appears in the viewer window and whose position you can change. It is very useful and we shall be using it to visualize the effects that color manipulations have upon the image. An example of it at work can be found in the Sampler node page.

The Histogram node

This functions a lot like any other histogram (e.g. the one in Photoshop). It will return the familiar distribution mapping of color values from black to white along a straight line (below). It can, like Photoshops's histogram, also re-map these values but I would recommend strongly against it's use for this purpose as in the process it destroys all values below zero and above one. Besides which, its a dog to use. In its default state it only reads luminance values. You can, with the judicious use of a Shuffle of Copy node move your channels around so that they are isolated one by one.

Feedback 6.png

The RGB vertical lines (in ColorLookup node)

This is a very powerful color manipulation tool the use of which is covered separately. however, it can also be used to give visual feedback on any area or point within the image by displaying three vertical lines representing the red, green and blue values of whatever lies under the mouse cursor. It also returns the average values for any area that has been shift selected in the Viewer window.

Feedback 7.png

The ClipTest node

This identifies clearly in an image all instances of a particular color range in an image. It will display all instances of that color as a striped zebra pattern (below). Its common use is to check the range of the lower and upper registers of an image.

Feedback 8.png

The Clamp node

Then Clamp node, like the ColorLookup node, is not explicitly designed to provide feedback on images, however it can easily be used to do so. It functions in much the same way as the clip test node: it flattens out the user-defined lower and upper lightness range of the image and, if the MinClampTo and MaxClampTo parameters are ticked, it can replace these values with user defined colors. Properly used it can provide clearer information than the ClipTest node.

Feedback 9.png