Details in over-exposed areas of a Sony raw (.arw) photo that can be recovered by reducing highlights or exposure of the area effected (I use Lightroom Classic primarily), cannot be recovered in the output .dng produced by Topaz Photo AI from the same .arw photo.
Is there some procedure or setting I can use to correct this problem or is there a problem with .arw to .dng conversion?
I’ve uploaded a .arw photo that exhibits the issue.
In Lightroom reduce exposure and hightlights to reduce brightness and reveal the details of the ocean froth on the left side of photo. Shows that detail is present in the .arw.
Process .arw in Topaz Photo AI.
In output .dng adjust exposure, or whatever, to attempt to achieve same result as in step 1. I have been unable to succeed here. Appears to be no dependency on denoise or sharpen settings in Photo AI (sharpening subject only or no sharpening at all).
Converting the .arw to .dng in Lightroom directly without using Topaz Photo AI results in an image with all detail recoverable with proper exposure adjustments. Indicates there is an issue in Topaz Photo AI.
We generally recommend processing in Photo AI first before making edits in Lightroom, even if you use the plugin. We don’t read RAW edits so that might be causing the conflict you’re experiencing.
Immediately after you import the image to LR, use the File > Plug-in Extras > Process with Topaz Photo AI workflow to process the image to Photo AI and then export it back to LR. Let me know if that works.
I understand that you don’t read the raw edits. I have tested what you suggest, and I tested it again now, but the problem still appears. The point in step 1 in adjusting the exposure was just to demonstrate that the original .arw file actually contains the detailed information. The test steps can be modified to move step 1 to the end for verification purposes. I am still unable to make adjustments of .dng output from Topaz Photo AI that will recover the detail that can be recovered in the .arw file.
I have encountered the problem in at least one other image so it is not specific to this one file.
