I’ve been profiling exporting from Negative Lab Pro to see where I can improve export speeds. I have noticed that other exports from lightroom seems to be faster than my NLP exports. I also recently started using “add to catalog” when I export, and noticed exports again took longer. So, here’s a bit of profiling information.
@Nate has mentioned the many limitations of the Lightroom SDK plugins. Working with what we’ve got, I was interested in understanding the performance implications of different exporting options.
Based on this profiling, my personal workflow from now on will be:
Never use “Add to Catalog”. Just import it afterwards if I really want it in my catalog.
If i’m delivering images where I do not want metadata, I’ll use the generic export.
I’ll continue to use NLP exports for my personal library, where I want the metadata information.
@Nate perhaps you can give us some insight into the differences between using NLP versus generic exports, and what we might lost with not using NLP to export in the cases where metadata does not matter to us. Thanks for all the hard work you do!
To be completely honest, I don’t know what sort of project I would need, to pay attention if export speed is 55 or 59 frames per second. You data show speed variation about 10% or less - certainly I would not even notice (ok, I do it in much smaller batches) . IMHO the specific settings, hard drive configuration, caching and CPU may have larger impact than NLP optimizations. I would think that export to internal SSD drive would be faster than to classic hard drive or to external USB drive (I figure would be the slowest). No offense , for me personally it’s much more important that @nate release the workflow for slides than optimize export speed
Cheers!
I think you misread my table. The “FPS” is not the export speed. The export duration is the seconds column on the right that’s the total time to export. All the experiments is the same 34 images. This table shows there is a 6x performance difference between different settings.
For context, I work part-time at a film lab, and we export over a thousand images from lightroom ever day, with a decent percentage going through negative lab pro. At lab-scale, optimizing your export speed starts becoming impactful. I made this post to be helpful to the community so people can have some sense of how their export settings will impact their export duration. We switched our workflow to not add exports to catalog and use generic lightroom exports since we don’t send metadata to clients.
I agree with you that I don’t expect Nate to prioritize optimizing export speed. That being said, I made this post to share our findings and workflow changes to help the community. If others try to optimize their export speeds, hopefully they find this post as a starting point
Oh yes certainly, if you observe differences in factor of 6x this is certainly a showstopper - especially at lab. That’s interesting factoid by itself. I just read FPS as frames per second and was wandering how did you even noticed the difference. Never mind, all the best!
The only differences using the NLP export is that 1) the settings are already pre-set, 2) it uses ExifTool to update the metadata attached to the image depending on your metadata options, 3) when you “add to catalog” it’s also adding back in NLP specific custom metadata that is attached to the Lightroom catalog itself (which is separate from exif data). So for instance, the custom NLP metadata that allows you to see some of the NLP settings used, or filter your catalog. This is separate from the standard EXIF, and requires calls to the Lightroom catalog, which is probably why you are seeing the increase in time.
I’m not sure why you would see a difference in speed when you are not changing metadata or adding to catalog… there may be something else going on (like a different application of hardware acceleration internally) that I’m not aware of.
