Skip to main content
K
Kevin25756719nnjs
Participant
May 28, 2024
Question

Is there a way to resume batch AI Denoise after a crash without duplicating work already done?

  • May 28, 2024
  • 2 replies
  • 739 views

I am using AI Denoise on roughly 700 images at a time. This usually takes 12-40 hours to acheive, and usually it works, sometimes the program crashes, and I have to restart.

 

The trouble is that I haven't found a simple way to denoise only the ARW images that have not yet been processed. When I open Lightroom again, it appears that roughly half of the ARW images have accomanying DNG images that have been denoised, but they don't appear to have been processed in any particular order, so I can't simply select the second half of the image set and resume from there. And if I select all ARW images and resume, there does not appear to be any option to skip processing images that already havea  corresponding DNG.

 

Is there any way to simply select all ARW images that do not already have corresponding DNG images? Or to sort ARW images in the same order that Lightroom processes them so I can simply select all the ones after the last one that had processed? Or to resume batch denoise on all ARW images while skipping the ones that already have corresponding DNGs?

 

Thanks!

This topic has been closed for replies.

2 replies

dj_paige
dj_paige
Legend
May 28, 2024

You can search for photos among those 700 that do not have the keyword Denoise applied to them. Those are the ones that still need to be denoised. (This assumes the option is turned on to add this keyword to your photos that are denoised — I believe this option is in Preferences but I can't check right now as I am at work and don't have access to Lightroom Classic)

johnrellis
johnrellis
Legend
May 28, 2024

"You can search for photos among those 700 that do not have the keyword Denoise applied to them. Those are the ones that still need to be denoised."

 

I hadn't used that option before, but I did some testing, and the Enhanced > Denoise keyword is applied to the generated DNGs, not to the original raws. So I don't think it isn't possible to directly use keyword search to find raws with no corresponding denoised DNG.

 

You could use the Syncomatic plugin to copy the Denoise keywords from the DNGs to their corresponding raws. But that would be about the same number of steps as the method I proposed (which doesn't require a plugin).

dj_paige
dj_paige
Legend
May 29, 2024

@johnrellis 

 

Syncomatic not required. It is my observation that the Denoise operation goes sequentially through the selected photos. You search for keyword Denoise, you find that of the 700 photos, DSC_7328 through DSC_7407 have been denoised. Which photos still haven't been denoised? The ones that are not in that range. So, you select the RAW images that are not in that range, and continue the Denoise operation. This last step is not done via the filter bar or any other Lightroom Classic tool, it is done via human paying attention to what is seen on the screen.

 

This assumes my standard workflow, where photos are sorted by either capture time or file name.

johnrellis
johnrellis
Legend
May 28, 2024

1. Run Denoise with the Create Stacks option, which will stack the denoised DNGs on top of the ARWs.

 

2. After a crash, in Library Grid view do the menu command Photo > Stacking > Collapse All Stacks.

 

3. In the Library Filter bar, open the Metadata browser with the column File Type and select Raw.  This will filter for raws that are not at the bottom of the collapsed stacks, i.e. that don't have a denoised DNG stacked on top of them.

 

You can save that setting of the filter bar as a filter preset.

 

(This leverages one of the differences between filters and smart collections -- filters only work on the tops of collapsed stacks, while smart collections search all photos in the catalog, regardless of stacking.)

    Terms & ConditionsAccessibility statement
    Using the community Terms of use Privacy Cookie preferences Do not sell or share my personal information ad choicesAdChoices