Windows11 (25H2), display resolution 3840x2160, dpi scale 150%
Settings - Formats - Read
request to add Jpegli decoder by WinnieW deserves more attention
Jpegli decodes standard JPEG files with slightly better visual quality than traditional decoders by prioritizing mathematical accuracy and psychovisual fidelity over sheer processing speed. By processing the image data using 32-bit floating-point precision, it minimizes rounding errors, which subtly smooths gradients, sharpens edges, and reduces chroma artifacts in areas like skin tones and saturated colors (especially in low-quality JPEGs).
However, while this high-fidelity reconstruction reduces banding and ringing, the decoder remains fully JPEG-compatible and cannot magically restore permanently lost details or completely eliminate heavy compression blocks.
https://github.com/google/jpegli#decoder-improvements wrote:Decoder improvements
- Support for 16-bit unsigned and 32-bit floating point output buffers.
- Non-zero DCT coefficients are dequantized to the expectation value of their respective quantization intervals assuming a Laplacian distribution of the original unquantized DCT coefficients.
- After dequantization, inverse DCT, chroma upsampling and color space conversions are all done in floating point precision, the conversion to integer samples happens only in the final output phase (unless output to floating point was requested).
Suggestion
- add Jpegli decoder
- add 'Use Jpegli' checkbox in Settings>Formats>Read>JPEG (like in Write>JPEG)
- maybe with bitdepth (per channel) combobox [8, 16]
Test
- Test image
testorig.jpg (4:2:0) from jpegsr10.zip (ijg.org)
- Decoders
- libjpeg-turbo
- commands
Code: Select all
djpeg.exe -png -outfile testorig_djpeg_dct-int.png testorig.jpg djpeg.exe -dct float -png -outfile testorig_djpeg_dct-float.png testorig.jpg
- switches (usage.txt)
There are few decode switches available, but only one can potentially improve default decode quality: dct
Code: Select all
-dct int Use accurate integer DCT method (default). -dct fast Use less accurate integer DCT method [legacy feature]. When the Independent JPEG Group's software was first released in 1991, the decompression time for a 1-megapixel JPEG image on a mainstream PC was measured in minutes. Thus, the fast integer DCT algorithm provided noticeable performance benefits. On modern CPUs running libjpeg-turbo, however, the decompression time for a 1-megapixel JPEG image is measured in milliseconds, and thus the performance benefits of the fast algorithm are much less noticeable. On modern x86/x86-64 CPUs that support AVX2 instructions, the fast and int methods have similar performance. On other types of CPUs, the fast method is generally about 5-15% faster than the int method. If the JPEG image was compressed using a quality level of 85 or below, then there should be little or no perceptible quality difference between the two algorithms. When decompressing images that were compressed using quality levels above 85, however, the difference between the fast and int methods becomes more pronounced. With images compressed using quality=97, for instance, the fast method incurs generally about a 4-6 dB loss in PSNR relative to the int method, but this can be larger for some images. If you can avoid it, do not use the fast method when decompressing images that were compressed using quality levels above 97. The algorithm often degenerates for such images and can actually produce a more lossy output image than if the JPEG image had been compressed using lower quality levels. -dct float Use floating-point DCT method [legacy feature]. The float method does not produce significantly more accurate results than the int method, and it is much slower. The float method may also give different results on different machines due to varying roundoff behavior, whereas the integer methods should give the same results on all machines.
- commands
- jpegli* (libjxl)
*project migrated to separate repo, but binaries available only in the parent repo so far
- commands
Code: Select all
djpegli.exe testorig.jpg testorig_djpegli_bitdepth8.png djpegli.exe testorig.jpg testorig_djpegli_bitdepth16.png --bitdepth=16
- switches (help from binary)
There is only one switch available that can potentially improve default decode quality: bitdepthCode: Select all
Usage: D:\jpegli\bin\djpegli.exe INPUT OUTPUT [OPTIONS...] INPUT The JPEG input file. OUTPUT The output can be PPM, PNM, PFM, PAM, PGX, PNG, APNG, JPEG --disable_output No output file will be written (for benchmarking) --bitdepth=8|16 Sets the output bitdepth for integer based formats, can be 8 (default) or 16. Has no impact on PFM output. --num_reps=N Sets the number of times to decompress the image. Used for benchmarking, the default is 1. --quiet Silence output (except for errors). -h, --help Prints this help message. All options are shown above.
- commands
- libjpeg-turbo
- Comparison
- djpeg -dct int|float
difference is marginal, so default '-dct int' is used
- djpeg vs djpegli
difference is visible, jpegli looks better
- djpegli --bitdepth=8|16
there is diff, but it is not visually noticeable in the test image
- djpeg -dct int|float
Related
- issues
- fixed -
1.8.6 - settings - read - camera raw - settings ignored (1.8.7)
- suggestions
Settings - read - JPEG - add Jpegli decoder (current post)
Viewer - read - RAW - add open dialog
Viewer - read - Camera RAW - add command to loop sizes
Settings - read - Camera RAW - display preview while loading full size
Settings - formats - rework Saving Read Write tabs
- done -
Settings - write - PDF, TIFF - update default settings (1.8.7)
Settings - read - Camera RAW - clarify settings (1.8.7)
