The linearization process refines the ink curves used to print digital negatives so that your printed densities match the target output response, ensuring consistent and accurate tonal rendering in alternative processes. The QTP-DN app automatically performs linearization when a valid set of measurement data is loaded. This process compares your printed densities against a known target and adjusts your curves to produce more accurate tones across highlights, midtones, and shadows. If no measurement data is loaded, QTP-DN uses a default input range based on ideal values—from LAB L* 99.99 down to 0.01. This fallback ensures that any `.quad` file opened will undergo only minimal modification, preserving its original tone structure unless true measurement data is provided. #### When Does The Linearization Actually Happen - **If you load measurement data** from a scanner, spectrophotometer, or densitometer: - The measurements are smoothed using the dynamic averaging method. - QTP-DN calculates the difference between these measurements and a target tone response. - A new correction curve is generated and ready to be applied to the .quad file used the generate the negative the measurements came from. - **If no measurement data is provided**: - QTP-DN applies a default linear Lab L* curve from 99.99 to 0.01. - This ensures the output remains linear with minimal adjustment to the original file. The resulting correction curve is then used to adjust how your ink is printed, ensuring more accurate tone reproduction for digital negatives. ### Linearization (Updated for New QTP-DN App) ![[Linerar Lab L with Auto Install.png]] ### Printing the negative from the starter curve There are a few different step wedge targets of different patch sizes that come with the application. Choosing a target with a greater number of steps will result in a more accurate correction curve in the linearization stage, and most of the benefits will occur at the extreme highlights and shadows. You will need to balance the ease of measurement and the accuracy of the correction curve. If you are using a flatbed scanner and the Photoshop color picker, or a manual reflective densitometer then it is easier to use a 21 or 51 step target. If you are using a measurement device that is capable of operating in scan mode, then I would recommend a 101, 128, or 256 step target. All the targets I provide have already been converted to a negative with the correct gray gamma working space that will match what I recommend in inverting and flipping your image in Photoshop. Print the step wedge image onto the transparency material and thoroughly dry it by letting it sit overnight or by using a hair dryer until he milkiness is gone when viewing it from the base side of the material. Then print the negative with your standardized coating, exposure, and processing that you will use going forward. Make sure the print is thoroughly dried before moving on to the measurement step. Any additional dry down or perceived increase in Dmax from a partially dried print will affect the linearization calculations. #### Import the Measurements into QTP-DN 1. Open the QTP-DN app. 2. From the main profile interface, click the Linearize button. 3. In the Linearization window, click the Open Measurements File button - Alternatively, can drag and drop the file into the linearization window or the app icon in the Dock. ![[Smoothing Settings.png]] #### Interpreting the Measurements Graph ![[Raw Measurement Data.png]] After importing the measurements, the left graph will visualize the measurement data. Here’s how to read it: - The y-axis and left-hand labels is the Lab L\* values from Light at the top to Dark at the bottom. - the right-hand y-axis labels is for the Lab a\* and b\* with neutral being 0 in the center - The x-axis shows the **positive** K% from 0 as white on the left to 100 as black with the dMax on the right - Black Dots: The original measured densities from the printed target. - Extreme outliers can be manually selected and moved with the mouse or by editing the values in the table on the left. - Red Line: A smoothed measurements that are used in the linearization lookup functions. This eliminates noise and serves as a reference for generating new linearized curves. - [[2 Linearization#How The Measurement Data Smoothing Works Dynamic Sliding Window Averaging|More information on the smoothing method]] - Green Line: The ideal output target densities - Different contrast ranges can be set with the Output Curve Type dropdown menu - Linear Lab L - Perceptual rendering to have better screen to print matching - Manual curve control where you can load a photoshop .acv file (see soft proofing and manual Linearizing to Photoshop Curves) - Spot color tone value correction for linearizing the color tint for CMY separations - Black Line: A reference line for with linear steps from Dmin to Dmax. From this smoothed measurement data, you can now export a .cube LUT or .acv file to apply to an image in Photoshop (for use as a correction or soft proof curve). ## Opening the Quad File for QTR Linearization You must then load the .quad file that was used to make the negative for the print you measured and loaded the data above. (It will usually be the starter curve you made with the blocking density setting if this is your first iteration) You can also load .quad files for positives for straight inkjet prints, photogravure, or inverted PiezoDN curves but you must click the checkbox for using a positive correction curve. The correction curve from the smoothed measurement data is automatically applied the the ink values in the quad file as soon as it is opened, and will update in real time when adjusting the smoothing or output curve settings. You can also smooth the ink values independently of the measurement values for adjusting for any slight bumps in the ink curves. This is generally not needed, but uses a similar method as the measurement data smoothing formula (see below.) ### Linearization Export Options File saving options in the linearization module. ![[QTP-DN Export Options.png]] ![[Save with Auto-install Output.png]] ##### Linearization and Correction Curves - .quad file - normal linearized quad file - .quad file with auto-install - Linearized quad file and then automatically runs the QTR installxxxxx.command script based on the name of the Quad folder you saved the file - You will get a popup alert with the list of successfully installed curves with information about any errors of files that were not able to be installed. - (do not run this when something is currently being printed with that Quad printer. ) - .cube LUT - 256-step correction curve that can be used for non-QTR workflows - .acv Photoshop curve - 16-step correction curve that can be used for non-QTR workflows ##### Soft Proof .cube or .acv curves - Measured Soft proof: creates RGB values based on the smoothed LAB data (red line in the measurement graph) - Final Soft proof: creates the RBG values based from an internally created lookup table based on the correction curve and target output LAB data (green line in the measurement graph) ##### Additional Export Options - Smoothed measurement data: creates a 256 step txt file with the smoothed LAB data that can be useful if you manually adjusted the original measurement data - GutenPrint XML ink curve files that can be used with the the gutenprint 5.3+ when saved to the "User Defined" folder --- ### Using the Soft Proof Options If you want to see what you print will look like with the Linear Lab L* output curve, you can export a final soft proof as a Photoshop .acv curve preset file. Then open Photoshop and convert your image to sRGB Then create a new Curves Adjustment Layer in Photoshop then load the .acv file you just saved as a curves preset. This will show you the expected color and tonal range and dMax from the original test print. ![[Linear Softproof Curve.png]] You can use this soft proof to make any additional adjustments to the image with adjustment layers created **below** the soft proof layer, then turn off the soft proof layer before flattening, converting to grayscale as gray gamma 2.2, and then inverting and flipping to a negative. Alternatively, you can create a new adjustment layer **below** the soft proof layer to make a custom perceptual rendering curve for a better screen to print match that you can then open in the Manual Curve Control option and apply it directly to the ink values in the .quad file so you do not have to apply an additional contrast adjustment to every image prior to creating the negative. When you are happy with the contrast adjustment, save it as a curves preset to your measurements data folder. ![[Perceptual Rendering Correction Curve.png]] ### Linearizing to the Manual Curve ![[Better Screen to Print Matching with Manual Curve Control.png]] Then go back to the linearization screen that has the original measurement data and quad file and select the manual curve control option in the output curve type dropdown menu, and press the open .acv button to load the Photoshop curve you just created. It will automatically re-calculate the new target densities and re-linearize. You might need to adjust the curve points to get a smooth output curve, and then press ok, and save a new .quad file with an indication for the different contrast range. (I generally only us Process-Crv-Name-MC.quad for manual curve) --- ### Confirming Linearization To ensure the new linearized curve is printing correctly, you should make a test print of your standard test image and include a bullseye gradient that will show any obvious reversals or problems with the linearization. I also recommend including a 21-step grayscale target to measure and confirm linearity. The other important factor to check is the smoothness of the mid-tones. There is a grayscale torture test image in the resources folder that includes two bullseye gradients as well as linear gradients broken up into 11, 21, 51, 101, and 256 steps. I uses this image to make sure there is no posterization or banding in the final linearized curves. Any problems or banding are usually caused by inconsistent coating or not enough smoothing being applied to the measurement data. You can usually see that by examining the printed gradient and comparing it to any bumps or spikes in the measurements and linearized quad curves. ### How The Measurement Data Smoothing Works: Dynamic Sliding Window Averaging To reduce noise and inconsistencies in your measurement data, QTP-DN uses a smoothing algorithm called **dynamic sliding window averaging**. This method enhances the reliability of the correction curves by adjusting how much data is averaged at each step, based on its position along the tonal scale. - In the **highlight and shadow regions**, where tone changes are subtle and highly sensitive, the window is kept small to preserve detail. - As the algorithm moves toward the **midtones**, the window **gradually increases in size** up to a defined maximum. This larger averaging window smooths out minor measurement noise while maintaining the overall shape of the curve. - After passing the midpoint, the window size **gradually decreases again** as it approaches the deepest shadows, ensuring detail in the Dmax region is preserved. This approach prevents overfitting noisy data in the midtones and avoids flattening delicate tonal information at the ends of the scale.