Fine tuning EDN

About corrections

The quality of the corrections calculated by Easy Digital Images depends largely on the quality of the coating of the light-sensitive solution on the paper. Namely, in the case of a good coating, the program calculates an almost perfect correction of the transparent negative, regardless of the type of photographic technique.

Automatic corrections 

A slightly worse correction can be corrected with the Combine 1st and 2nd Correction command, located in the Easy Digital Negatives program. We simply merge the first LUT correction file (1) and the second LUT correction file (2). The result is often, but only with a proper application of the solution onto the paper, a much-improved correction (3). Once again. The best solution for making an excellent correction is a quality of coating on a paper or some other image carrier.

Manual correction of corrections

In most cases, software correction of corrections is of high quality, and we can produce high-quality negatives with such calculation. But some users want to make almost 100% perfect corrections. That’s is impossible, as due to the structure of the paper and errors in reading the color with scanners, cameras, or spectrometers, we always encounter about 2% deviation from the ideal values. The theoretically straight line of scanned data is thus always jagged. 

The line is jagged even when we read values that are printed with a printer directly on photographic paper (4). 

From what has just been said, we can now conclude that it is impossible to correct each color’s position, as we will never hit its real value. 

Programers of various photo processing corrections have also found that the optimal number of points we use for corrections is about 16. A smaller number of these checkpoints gives only approximate results, and too many checkpoints lead to increasing errors. Therefore, a more significant amount of control points does not improve the quality of the correction.

But with an already made correction (1), we can also use a more significant number of points. 

In Easy Digital Negatives, curve correction and gradient maps are made with 15 “standard” points, as this technology rarely allows a more significant number of control points. 

Otherwise, Easy Digital Negatives uses 21 checkpoints to calculate corrections (5). 

Corrections via curves and gradient maps are used in Adobe Photoshop and GIMP. 

The 21 points calculated by EDN, can only be changed manually via the LUT 1D.cube file. 

There are two main LUT files available in the program used for manual corrections, LUT 1D with 21 checkpoints (6) and LUT 1D with 52 checkpoints (7). If someone is brave enough, it can also change the 256 control points stored in the LUT 1D 256 file (8).  

Automatic correction

We calculate corrections of corrections (1) using two correction files. First, we correct the basic step wedge table, and then we print the new step wedge table with the just calculated correction. From this sample, we make a further, second correction.   

1. When we print a negative with the procedure we described in the initial chapter; the program will calculate the first correction (1). For correction of correction, we can only use LUT 1D files. 

2. In our example, we save the correction with a name CORRECTION_1.cube. 

3. With this first correction, we print another step wedge table (2). Of course, we can also use data for spectrometers. 

4. In using a step wedge, we get an image with an almost perfect distribution of tones (2). 

5. Now we open this, almost complete perfect sample (2) in Easy Digital Negatives. The program calculates a new correction.

6. We save this LUT 1D file with the name  CORRECTION_2.cube. 

7. If we have applied a quality coating of the light-sensitive solution to the paper, the correction will be almost excellent (3). 

8. Otherwise, we get a slightly worse correction (4), which we will fix automatically. 

9. In the Easy Digital Negatives program, we check the Combine 1st and 2nd Correction option. 

10. Using the mouse, we select both the first and second correction files, i.e., CORRECTION_1.cube and CORRECTION_2.cube. We drag the files to the Choose Files button.

11. The program instantly calculates a correction (5) that differs slightly from the first correction (1). But the resulting image will be much better.

12. We can now save the correction file to any file type. 

Manual correction

More demanding users are often dissatisfied with fixes that are not nearly perfect. 

In this case, demanding users can change the corrections manually in any correction file. 

This section will focus on corrections in Adobe Photoshop and corrections that we manually change in the LUT 1D file. 

Correcting Adobe Photoshop gradient map

1. We print the first sample step wedge table EDN_RGB_256.tif (1) and make the first correction in the EDN program. 

2. We save the correction to an Adobe Gradient Map file (2). We save it with the name CORRECTION_1.grd. 

3. In the digital image processing program, we add the just calculated correction to the EDN_RGB_256.tif file. Then we print a new sample image (3). 

4. When we calculate a new correction for a file that we just corrected, we notice that the correction is quite good. But in the middle part of the correction, we see a slight deviation of the tones (4). We will correct this deviation manually. 

5. As mentioned, we use 15 checkpoints in gradient and curve type corrections. These are placed in the positions shown in Table 1. In our graph, we marked these points with orange dots (4 and 6). 

6. The linearized result in the Easy Digital Negatives graph (6) shows us that we need to reduce the value in point 40. 

7. First, we observe that the offset of the curve in the table with 101 values is about 5, which means that in the table with 256 values, the offset is about 2.5 times higher, i.e., 5 x 2.5 = 12.5. The exact correction value can also be obtained in the Scanned Normalized Data file (5). 

8. In Adobe Photoshop we open our gradient map with the first correction CORRECTION_1.grd (7). Then we change the value at position 40 (8). 

9. As mentioned, we will reduce the present value by 12. That means that we lower the value of all three colors from 130 to 118 (9). 

10. We click on the OK buttons to save the corrections, after which we can reprint the sample with the corrected correction of CORRECTION_1.grd. 10. The result is now much better (10).

File Scanned Normalized Data

If we want to know the exact offsets of the scanned tones from the ideal values, we can use the Scanned Normalized Data file (5). 

This file stores both ideal and scanned data for all 256 fields of step wedge table. 

We open the CSV file in any data processing program (Excel, Numbers, etc.) or any text editor (12 and 13). If we are interested in shifting our point at position 40, we must first convert this number to the value for use in a table with 256 fields. In our Table1 table above, we notice that this point has a value of 102. 

Now, in Excel or Numbers, we find this ideal value at position 102. 

Once again. Do not use curves to make corrections. Changing the value of a single control point will also change the values at adjacent points (11).

Once we find it, we notice its real, scanned value in the right column (12). In our case, the offset is 13.67, as the value of the scanned color is 115.67.

Corrections in the LUT 1D file

This paragraph will only briefly mention the manual change of points in the LUT 1D file. 

Such corrections will most likely be dealt with only by very demanding users, who, in most cases, are already familiar with the basics of corrections. 

We open the file with LUT 1D corrections, be it a file with 21, 52, or 256 checkpoints, in any simple text editing program (13). 

We notice that the LUT file is a simple text file. In it, the RGB values of the colors are written in three columns. 

The color values are written in the range from 0 to 1. That means that we have to multiply these results by 255 if we want to get color values from 0 to 255. 

If we want to change the color values from 0 – 256 to 0 to 1, the number is simply divided by the value 255. 

In our example above, we need to reduce the value in point 40 by 12.5. The number 12.5 is therefore divided by 255. The result is 0.04090. 

The values in all three columns, i.e., columns R, G, and B, which are delimited by a space, are always the same. The correction is made in gray. 

The values in the LUT file are also not written with 15 points, but with 21 and 52 correction points, respectively. 

The ideal points are now evenly distributed, as we can see in the graphical representation (14), and are no longer centered around light and dark tones (6). 

In a 21-point file, the typical values increase by 0.05 in each step. In the range from 0 to 255, the step increases by 12.75. It is the same as the steps in the curve and the gradient map, but with some intermediate values. 

The values in a file with 52 fields change in steps of 5, i.e., 0, 5, 10, 15, 20. 

If, for example, we want to change the value of the point at position 40 in the LUT file with 21 control points, we have to change all three values of the ninth point. The values should change by 12.5 / 255 = 0.04090.

A brief description of corrections in the LUT file

1. We make the first correction of the step wedge table. 

2. We save the correction to the LUT 1D file.  

3. With this new correction, we create a new positive image of the step wedge table. 

4. From this, almost corrected image, we make a new correction file of LUT 1D type. 

5. We search for the most significant deviations on the graph of the EDN program (14). Exact deviations from ideal values can be found in the Scanned Normalized Data file. We divide the value of the found difference between the ideal and the scan value by 255.

6. We open the first correction file and change the color values of the selected control point in all three columns. 

7. We save the file and print a new sample. Etc.