Tom Hogarty Interview

by Michael Clark

Last Friday an interview with Tom Hogarty went live on Inside Lightroom. To check that interview out click on the link below:

http://www.oreillynet.com/pub/a/oreilly/digitalmedia/2007/08/17/interview-with-tom-hogarty.html

In the interview, Tom was gracious enough to speak with me about future improvements to Lightroom, Lightroom's digital asset management (DAM) capabilities, a little explanation on the new sharpening improvements, and about many other topics as well as a lot of other very interesting comments. I was very impressed with how consumer feedback influences Adobe's direction concerning Lightroom. If you ask for it they will at the least consider it.

One of the things I found very interesting were Tom's comments on using Lightroom as a digital asset management tool. At the moment, I have about 5,200 images imported into Lightroom. This week I plan to import about 20,000 images just to see how much the program slows down and if it slows down at all. As a test to see how fast or slow Lightroom is working, I will render 1:1 previews for one folder that contains 287 images as it is now (with 5,200 images in the catalog), then repeat this test once I have imported the new images. With 5,200 images total in my catalog it took 38 minutes 54 seconds to render the 1:1 previews for a folder containing 287 images. I will also of course work on images and see if there are other ways in which the program slows down or if it seems to run just the same as it does currently.

Of course Lightroom can be used as a DAM tool right now no matter how many images you have. With the catalog structure one can have multiple catalogs of 10, 20, 30 or even 50,000 images. That is kind of a workaround but I'll soon find out just how well it performs. If you have any suggestions please feel free to offer them up and I'll try to incorporate those into my testing.

I'll report back on my findings next week. Until then enjoy the Tom Hogarty interview. That's it for this session. See you next week....

Adios, Michael Clark

7 Comments

Mark Sirota
2007-08-20 10:37:29
Your 1:1 rendering test sounds like a good objective measure, but I'm not sure it'll be showing you a useful measure of Lightroom performance.


What matters to the user in terms of performance is responsiveness in frequent activities. It's not often I render 1:1's for 287 images at a time. And that test will be impacted strongly by the disk read performance on the originals, disk write performance on the previews database, and compute power for rendering, as well as database access to acquire the list of modifications to be rendered. Only that last part is likely to be significantly affected by the size of the catalog.


If the goal is to test catalog performance at scale, I would recommend a test that works the catalog exclusively. Perhaps make a metadata change on those 287 images rather than creating 1:1 previews, for example.

Tim
2007-08-20 11:11:32
When you import your images, I suggest you do it in small bites. When 1.0 and 1.1 were released, Adobe's LR forum had people reporting how LR crashed, took forever or failed to import alot of photos when it finished. Many were trying to import a large number of images. It occurred to me the database isn't designed to digest tens of thousands of images at one time. Typical use is a few hundred at a time.


Good Luck

Geoff
2007-08-20 12:07:08
Thanks for posting up your Tom Hogarty interview. It inspired me to submit a suggestion regarding the DAM functionality of Lightroom. As mentioned by Tom, DAM can now be partly achieved with smaller catalogs of images (keeping performance to a maximum), however searching your keywords would only return results of the currently loaded catalog (or so I believe). How about the Lightroom database tracking and recording keywords for images in all catalogs? For example, a keyword search would show thumbnails of matching images in the current catalog, but would also show a list, or maybe thumbs (but only thumbs) of matching images from other catalogs. These thumbs could then be dynamically linked to the catalog and could load on double click. Any thoughts?
Gary
2007-08-20 12:29:09
I have about 23,000 image in LR 1.1 on an Intel Mac Pro, dual Xeon (2.3GHz I think) and I often import and render 1:1 simultaneously quite often up to 600 or so images, I don't ever remember it taking that long.


The thing I find slowest is when I change the metadata of a large group of images, and it is writing the changes.

Michael Clark
2007-08-20 12:48:14
I guess I should clarify (and I will in part 2 of this blog post next week) that I am working on an Apple G5 Dual 2 Ghz tower with 4.5 GB of RAM. I decided to do the test on my imaging machine because I felt it would be closer to an above average but middle of the road computer. I also have a MacBook Pro which seems to run about the same speed as my G5 tower but I decided to stick with the tower - I can export the catalog and compare speeds on the intel laptop vs the tower.


Thanks to all for the good ideas - keep them coming. There are many criteria that I'll have to lay out which might explain the longer times on my machine - I am importing raw images shot with a Nikon D2x and those can be quite large files. Also, I have my presets set to roduce the highest quality 1:1 previews because of my large monitor so that also will affect the preview rendereing times. I'll lay out all of the details next week when I have finished up the comparison but so far at 10,000 images it works just fine.

GIo
2007-08-22 09:04:21
"How about the Lightroom database tracking and recording keywords for images in all catalogs?"
Surely the priority should be to ensure performance is such that no-one feels such a multi-catalogue workaround is necessary? While there are valid reasons for breaking up catalogues, performance shouldn't be one of them.
Geoff
2007-08-23 11:54:40
"Surely the priority should be to ensure performance is such that no-one feels such a multi-catalogue workaround is necessary? While there are valid reasons for breaking up catalogues, performance shouldn't be one of them."


Well I have to agree with you there, in an ideal world ;)
However the functionality would still be valid to those of us who find a multi-catalogue workflow useful.