Lightroom 3.2 Release Candidate Available on Adobe Labs

Report · Aug 09, 2010

This release includes camera support, bug fixes and new features. Details here:

http://blogs.adobe.com/lightroomjournal/2010/08/lightroom-3-2-and-camera-raw-6-2-available-on-adobe-...

Regards,

Tom Hogarty

Lightroom, Camera Raw and DNG Product Manager

Report · Aug 16, 2010

What is odd is that loading even unedited pictures was considerably faster in LR 2.7. In fact almost everything was faster in 2.7, with the same files from the same camera. It's as if memory and cache management has been changed in 3, but not properly optimised therafter. To be fair, I do accept that using perspective correction will involve a major performance hit. It's just one of those things, I expect, you don't get something for nothing. But the rest of LR should not be affected when lens corrections and perspective are not in use.

John

Report · Aug 16, 2010

John_R_Smith wrote:
What is odd is that loading even unedited pictures was considerably faster in LR 2.7. In fact almost everything was faster in 2.7, with the same files from the same camera. It's as if memory and cache management has been changed in 3, but not properly optimised therafter. To be fair, I do accept that using perspective correction will involve a major performance hit. It's just one of those things, I expect, you don't get something for nothing. But the rest of LR should not be affected when lens corrections and perspective are not in use.
John

John,

Since you're working with medium format size files, and since we're on the Lens Correction side of the discussion, I wondered if you could try something. I've opened a support ticket about export times being impacted in 3.2RC and that both 2010 Process and, even more so, Lens Correction add a lot to the export times.

Could you select 10 or 15 images and try exporting, perhaps try with 2003 Process and no Lens Correction as a starting point, then add the others back in? In my case each added a not so insignificant amount. I am exporting to an eSATA connected external drive.

I'm just more curious because of the size of the files from the perspective of both items being "off" vs. both "on" when exporting.

Thanks!

Jay S.

Report · Aug 16, 2010

After running about 10,000 new photos through 3.2, addiding to a bloated 217,000 image catalog, I see definate improvements. Thanks Adobe for killing some bugs. I am sure the next version will be better still.

Better

* Brush performance improved

* Program opens and default catalog loads more quickly

* Simultaneous preview rendering and import from card reader no longer freezing program

* Herky-jerky "move folders within LR" issues gone

Don't like

* No default keyword option on import

Report · Aug 17, 2010

Well, chaps, I am going to have to hang my head in shame and eat humble pie here. I have been horribly guilty of making subjective judgements not based on proper scientific data.

Last night I purged the ACR Cache and did some testing in the Develop Module loading files, timing everything just using the second hand on my wrist watch. The ACR cache is working properly, and it does make a significant difference. The results are -

Load uncached, unedited image - 15 secs

Load same image, with cache - 10 secs

This is consistent, and is 30% faster. So why didn't I notice this? Well, the reason is simple. I always work on one photograph at a time, and never do any sort of batch processing - every image has different edits. The only reason I ever load a file into the develop module is to edit it, so I never load it again cached and without edits. Editing the file makes the load slower, so that even with the cache it takes 12-13 secs, which is so similar that I perceived it as the same.

The next horrible error was to think that LR 2.7 was faster. I thought it was, but running the same test the load results were the same - 15s with no cache and 10s with the cache. So why did it seem faster? Once again, the reason is subjective. In Win 7, LR2 just displays a "Loading" notice. LR3 displays the "spinning circle" as well. Because the spinning circle sort of hesitates and freezes at intervals during the load, it made it seem to me that LR3 was slower.

So to sum up - the ACR cache makes a 30% reduction in load time with an unedited image. But once the image is edited, the difference is much less, about 10%. When this represents either 15s or 13.5s, you really hardly notice. I think that for me, at any rate, once a process of any sort on the PC goes above 10 sec it just seems slow. Slight differences of 2 or 3 secs above this are of little consequence.

John

Report · Aug 17, 2010

John,

I sense you've slapped yourself around a bit already so I won't venture to do it more, and anyway I find your candor refreshing.

15 uncached & 10 cached seems pretty good for a 39MP file. It seems like Lr3 is working about as it should for you, I guess.

Not sure how to reconcile Bob Frost's subsecond rendering with all this, but his data is definitely not in with the clustering (meaning it is unusual).

I'm starting to suspect my 3:1 ratio (cached versus non-cached) is also unusual, although I don't have enough data for a curve yet.

At some point I'm gonna get tired of all this talking about it, but at this point I'm still curious what others times are for comparison.

Rob

Report · Aug 17, 2010

John_R_Smith wrote:
Well, chaps, I am going to have to hang my head in shame and eat humble pie here. I have been horribly guilty of making subjective judgements not based on proper scientific data.
...
...
...
John

Hi John,

No need for humble pie.. Happens to everyone here at some point. TOO many things to try and retry . Speaking of (Notice the clever use of a segue here). Did you happen to try the Export image request I made to you a few posts back? I know you've been busy thrashing cache'ing but if you get a chance it would be helpful.

Thanks.

Jay S.

Report · Aug 17, 2010

Jay

No I have not tried that yet. So far I have never exported more than one file at once, so this would be a bit of an odd experience for me.

John

Report · Aug 17, 2010

John_R_Smith wrote:
Jay
No I have not tried that yet. So far I have never exported more than one file at once, so this would be a bit of an odd experience for me.
John

John,

It really shouldn't matter, one or five, it's just that with a group, the opportunity to measure the time is a bit easier. The bigger question is the impact of Lens Correction and 2010 Process model to export. As I said, Adobe is looking at it, and additional info wouldn't hurt. 🙂

Interesting that you only do one image at a time on export. I do most editing on all images in a series, then export the bunch to jpeg.

Thanks..

Jay S.

Report · Aug 18, 2010

Here you go Jay, this is the nitty-gritty -

Export 3 files (MF 39MP images, export to 16-bit TIFFs at 360 dpi, file size 230 MB)

PC Win 7 64-bit, Intel Dual-Core 2.1 Mhz, 4GB RAM, 500MB HD

As PV 2003, no edits - 45 secs

As PV 2010, no edits - 50 secs

As PV 2010, edited with perspective corrections on all three - 94 secs

Which is pretty much what I would have expected, really. I never work on batches of photos, each picture is edited individually, and the export to TIFF is for archival purposes at the end of my editing session which could take two or three evenings for a given picture. So the time taken to export is really not an issue for me personally. If you were a busy wedding photographer I can see that it might be a real concern.

John

Report · Aug 18, 2010

John_R_Smith wrote:
Here you go Jay, this is the nitty-gritty -
Export 3 files (MF 39MP images, export to 16-bit TIFFs at 360 dpi, file size 230 MB)
PC Win 7 64-bit, Intel Dual-Core 2.1 Mhz, 4GB RAM, 500MB HD
As PV 2003, no edits - 45 secs
As PV 2010, no edits - 50 secs
As PV 2010, edited with perspective corrections on all three - 94 secs
Which is pretty much what I would have expected, really. I never work on batches of photos, each picture is edited individually, and the export to TIFF is for archival purposes at the end of my editing session which could take two or three evenings for a given picture. So the time taken to export is really not an issue for me personally. If you were a busy wedding photographer I can see that it might be a real concern.
John

John,

Thanks much.. Pretty much what I saw as well. The difference between 2010 and 2003 is small. I don't know how many edits you made on the last run with the Len Correction, but that seems to be where a much larger time element comes in, and yes, it is the cases where you're exporting a large number that this really begins to add up..

Appreciate the effort John.. I'll pass along the additional data to Adobe. They'll have a Windows and Mac example now.

Jay S.

Report · Aug 17, 2010

I have to say that compared to the original v0 download version that this 3.2rc is an exponential

improvement.

There are still some issues with rendering, and rerendering .. and rerendering.. but it's been mentioned in previous posts, and I'm sure it'll get fixed prior to the official upgrade.

Thanks Adobe. (everything I had been P _ _ _ ed off about has been given a sizable speed increase)

Good Work!

Report · Aug 17, 2010

Be careful about that assessment just yet. I thought the same thing until put it through it's paces. 1 week later I'm right back where I started... 15-20 second load times just moving from one image to the next. Basically, the software is useless at this point and I'm currently looking for alternatives.

Report · Aug 17, 2010

(I must admit, I've been treating it very carefully, and taking it a lot slower than I was with LR 2.7 because this is only partly tested.. so you're right in my early euphoria about some things being fixed, might be too quick to judge properly)

Report · Aug 17, 2010

(deleted (is there a delete button?))

Report · Aug 17, 2010

pgmartphoto wrote:
(3) lens profile correction do not seem to be as exact as using the old cyan/magenta slider method to correct CA. using a single slider to determine both kinds of ca is a bit dodgy because they don't always both happen, and I'm finding that as the profile eliminates some part of one, I'm getting fringing of another. not sure this was tested well, or even well thought out in the first place.

When using the default profile for your lens, it's quite possible that the Lens Correction CA under-corrects a bit (even if the amount of red/cyan and blue/yellow are essentially correct). For many of my lenses, I need to bump up the CA Amount to 125-130 for eliminate the CA entirely. When I do that I save out a new Lens Default so that the increased amount is applied. Also note that Lens Correction still can't do much about highlight fringing which still needs to be applied in the Manual tab of the Lens Correction panel. And yes, it's well thought out and tested...it's just that not unlike camera profiles, there's only so much you can do when testing a small sample. Canon to be sure has pretty wide build variation in many of their lenses. What may work for one specimen might not be prefect for every lens model.

Report · Aug 18, 2010

sorry if my words are a bit harse sometimes.. I get a bit worked up. Yeah I've been heading over to the manual tab a bit more often than not these days.

thanks for that. its good info

Report · Aug 18, 2010

And yes, it's well thought out and tested...it's just that not unlike camera profiles, there's only so much you can do when testing a small sample. Canon to be sure has pretty wide build variation in many of their lenses. What may work for one specimen might not be prefect for every lens model.

In this case I would expect to have both undercorrection and sometimes overcorrection. Still, it's always an undercorrection that requires a 120-130 values. According to Eric Chan (see this post) it's also a flaw in the Lens profile Creator.

Report · Aug 18, 2010

ok thanks. that's very helpful advice.

thanks

Report · Aug 18, 2010

dorin_nicolaescu wrote:
In this case I would expect to have both undercorrection and sometimes overcorrection. Still, it's always an undercorrection that requires a 120-130 values. According to Eric Chan (see this post) it's also a flaw in the Lens profile Creator.

I have seen over-correction not so much in CA but sometimes in distortion and sometimes in vignetting (not on the same lens so far). I have a Canon 10-22mm lens that the default distortion correction is too strong and "fixes" the normal barrel distortion and goes too far and turns it into a pincushion distortion. Backing the distortion amount down fixes that.

I know that Eric and Simon are working at improving both the Lens Profile Creator as well as improving some of the previously installed lens profiles. I'm pretty sure LR 3.2 and ACR 6.2 came with a few updated profiles for lenses that were less good on the original profiles as well as new profiles.

Report · Aug 18, 2010

Jeff Schewe wrote:
dorin_nicolaescu wrote:
In this case I would expect to have both undercorrection and sometimes overcorrection. Still, it's always an undercorrection that requires a 120-130 values. According to Eric Chan (see this post) it's also a flaw in the Lens profile Creator.
I have seen over-correction not so much in CA but sometimes in distortion and sometimes in vignetting (not on the same lens so far). I have a Canon 10-22mm lens that the default distortion correction is too strong and "fixes" the normal barrel distortion and goes too far and turns it into a pincushion distortion. Backing the distortion amount down fixes that.
I know that Eric and Simon are working at improving both the Lens Profile Creator as well as improving some of the previously installed lens profiles. I'm pretty sure LR 3.2 and ACR 6.2 came with a few updated profiles for lenses that were less good on the original profiles as well as new profiles.

Jeff,

Thanks for these updates. Ian has posted a few notes as well. The one area of confusion for me, is still the "how to get there" aspect of lens correction. In my case, while EXIF nailed the combo of a 70-200 Canon 2.8 + Canon 1.4 TC, Lens Correction was totally off base as to what Canon Lens it "thought" I had on. I understand there are obviously underlying operations for the corrections, but it would still seem that the starting point should be fixed by what EXIF is reporting, shouldn't it? It just seems that any calculations around trying to figure out what type of lens is on would be "wasteful" ? I'm sure I'm missing some aspect of this, but just am not seeing it.

Thanks again,

Jay S.

Report · Aug 17, 2010

Self-Delete Post?

I've considered reporting my own posts as abuse but I'm afraid I might get a buger next to my name somewhere.

Ian???

_R

Report · Aug 18, 2010

thanks.

I tried to delete it because I realised it didn't actually help and I was just frustrated, but then I got a worthy reply,

so I feel a bit embarrassed now, to say the least.

Report · Aug 18, 2010

pgmartphoto,

Well, if you can't get embarrassed on the Lightroom Forum, then where can you get embarrassed (OK - too many places to mention).

No worries,

Rob

Report · Aug 24, 2010

Is there any new ifnormation on Lightroom 3.2 official? I'm still dying for a cure here. I've all but given up on Adobe. Clients have all but given up on me.

Report · Aug 24, 2010

Just checked a few minutes ago ... still "only" Lightroom 3.2 RC - which is what I'm doing my production work with ... so far I'm still pretty happy with it ...

Best regards,

Harald