Introduction

While chdkptp is primarily intended as a PTP client, it also includes some commands useful for working with CHDK DNG files.

DNG CLI commands

DNG related CLI commands all start with dng. Use help for a list.

Most dng commands operate on files loaded into memory with dngload filename

More than one file can be loaded at once. By default, commands operate on the most recently loaded file, but images may also be specified by numbers obtained with the dnglist command, and the default can be changed with the dngsel command. The * in dnglist output indicates the default file. Files remain loaded until you unload them with the dngunload command

To work on a large number of files, you can use the dngbatch command. This loads each file, runs the specified commands and unloads. Directories can be recursively processed and patterns may be used to restrict the operation to certain files. To save the files, you must include a save command in the list. If you want to modify the original file, you must use save -over.

Warning: if you use save -over, no backups are made, and errors or interrupting the command may result in data loss. You may use -odir in the batch command to send output to a different directory.

The -pretend option can be used to see which files will be processed

Example: To print the dimensions and version of all DNG files in the current directory:

dngbatch . { info }

Note that command executed withing the dngbatch command must not include the dng prefix.

Bad pixel handling

Overview

CHDK creates DNG version 1.3 DNG files by default. In DNG 1.3, bad pixels are not fixed on the camera. Instead, DNG Opcodes are used to specify values that indicate which pixels are bad. This reduces the amount of memory required to shoot DNG, improves shooting speed, and avoids the need to create BADPIXEL.BIN on the camera. However, many programs do not process DNG opcodes, which leaves the bad pxiels as brightly colored or dark spots.

Patching bad pixels

Use dngmod -patch to interpolate over pixels with value less than or equal to a certain value. The default is 0, which is appropriate for fixing badpixels on most cameras. To batch all files in a directory called mydir , saving the patched files to a new directory called patched, use something like

dngbatch mydir -odir=patched { mod -patch ; save }

Generating badpixel lists

You can generate lists of pixels with certain values using the dnglistpixels command. This can be used to generate badpixel lists for programs like raw therapee, dcraw or CHDK BADPIXEL.TXT

To generate a list of 0 valued pixels in bad.txt in a format suitable for raw therapee use

dngload myfile.dng
dnglistpixels -max=0 -out=bad.txt -fmt=rt -coords=rel

To generate a badpixel.txt of “hot” pixels for chdk BADPIXEL.TXT on a 12 bit camera use
dnglistpixels -min=4095 -out=badpixel.txt -fmt=chdk

Displaying DNG information

The dnginfo command can display various information about dng files. By default, it shows the dimensions, bit depth, bayer pattern, dng version, chdk version, and camera name.

Using the -ifd option, you can display the contents of tags in the various IFDs. By default, the first IFD is displayed. IFDs may be specified, like -ifd=0.0 to display the first sub-IFD of the first IFD. Use -r to display all sub IFDs of the specified IFD. The exif IFD is 0.exif. To display values that aren’t inline in the IFD entry itself, use -vals.

Scripting

Using the ! command, you can use Lua code to query or modify DNG files. The files loaded in the cli are available from dngcli.list, and the currently selected file is available from dngcli.selected. There are some examples of code that modifies tags a dng in lua/extras: hackcfa.lua and hackdngactive.lua

Determining active area for a new port

Sensors generally have masked areas that do not contain image data. The DNG active area should be set to the part of the raw buffer which contains image data.

Shoot a DNG. The subject not be too underexposed at the edges, so you can see where the active area ends.
load a dng with dngload
dngdump -raw -rfmt=8pgm
open the resulting .pgm file in gimp or similar
Measure the corners of the active area
- The values should be divisible by 2 to so the CFA will not change between the active area and whole sensor.
- The X2 and Y2 should be should be the first pixel outside the active area, e.g. if the active area covers the entire sensor, X2 and Y2 would be the sensor width and height, not the coordinates of the lower right corner.
- If there are some borders that are darker, but still include recognizable image data, they should be included.