However, I have now seen striping pattern noise and crosshatching pattern noise with cameras of different brands and sensor technologies, including the Sony A7R III, Sony A7R IV, Fujifilm GFX100, and now the Nikon D850. I submitted a bug report to Adobe back in July, but I have not heard of any progress on the matter as yet.
- Sony A7R IV Image Quality: Crosshatching Pattern Noise with Adobe Camera Raw Enhance Details
- Fujifilm GFX Image Quality: Crosshatching Pattern Noise with Adobe Camera Raw Enhance Details
- Nikon D850 Image Quality: Striping and Crosshatching Issues with Adobe Camera Raw Enhance Details
With much regret, I must question the general viability of Enhance Details. While striping pattern noise and crosshatching pattern noise are frequently masked by sensor noise along with subject matter detail, using Enhance Details might be unacceptable with many images, particularly for large prints. Plus, using frame averaging makes any pattern noise pop out like a sore thumb, since once noise is reduced, any latent pattern is revealed.
Potential cause: sensitivity difference between rows of sensor?
My hypothesis is that the striping problem is not created by Enhance Details but already latent in the original raw files, with Enhance Details making that latent pattern visible. If so, the issue should affect all cameras, excepting those that have very fine tuned image processing that entirely neutralizes differences across rows of the sensor (some high-end cameras might do so).
In particular, my hypothesis is that the striping pattern noise or crosshatching pattern noise that emerges is the result of a small sensitivity differential between the rows of Bayer matrix cameras.
The RawDigger folks tell me that Monochrome2DNG compensates for inter-row sensitivity differences when processing raw files into monochrome DNG files. Seemingly, Adobe Camera Raw Enhance Details is not taking this differential into account (or cannot or is not doing it well, for reasons unknown to me). Which makes it a bug in any practical sense.
The pattern is seen whether processing the Enhanced DNG file with ACR into color or black and white, so it is inherent to the DNG produced by Enhance Details. Which also points out that processing into DNG with Enhance Details and then discarding the original raw files is a Very Very Bad Idea, since there is at least some hope for a fix.
With the Nikon D850, the predominant pattern is horizontal striping pattern noise (shot in landscape orientation), which tends to confirm the inter-row differential theory. There is also a certain crosshatching pattern composed of a checkerboard style arrangement of darker black dots. My guess is that these dots are another manifestation of sensitivity differential: the red or green or blue pixels in one row are slightly different in sensitivity to those in nearby rows.
Below, enlarged crops from a Nikon D850 raw file (14-bit lossless compressed). Toggle to compare the three.
ACR = Adobe Camera Raw processing of the original raw file, ACR_Enhanced = ACR processing of the DNG from Enhance Details.
The pattern of checkerboard spots is readily visible here in the trees against the sky, with the spots along with the striping pattern noise seen in the rocks.
The sky is affected, and looks particularly bad in black and white when darker.