Bad performance - Applying adjustments
So right now I'm trying to apply adjustments to about 250 photos in Capture One 12 and the performance is really bad.
I've checked every possible cause for this but my system is perfectly fine and there's enough space - in short: I can't find a reason for it.
However this is the first time I've used the Luma Range on a layer while working on so many images - so might that be the issue?
Does it really need to check every image for its Luma range?
I've checked every possible cause for this but my system is perfectly fine and there's enough space - in short: I can't find a reason for it.
However this is the first time I've used the Luma Range on a layer while working on so many images - so might that be the issue?
Does it really need to check every image for its Luma range?
0
-
Yes. I can't see how else it could work
That said the performance effect I have seen is noticeable but not dreadful considering what it is doing.
250 images seems quite a lot for a single pass.
If it has started to go slower and slower over time I would check the memory usage.
You may also find it useful to give the Support Team a heads up so that they can assess the log files to see if anything unexpected is being reported.
Given that this is a new feature there will, no doubt, be value in real use feedback compared to whatever might be discovered in testing - even user testing during the development period.
Grant0 -
It ate up 25GB of RAM and then crashed while shooting. That's not a good thing. 0 -
[quote="C-M-B" wrote:
It ate up 25GB of RAM and then crashed while shooting. That's not a good thing.
Definitely a Support Case matter then.
Just one layer with a Luma range selection or several?
Are you making these adjustments in batches while shooting tethered?
That's rather brave for a brand new feature.
Grant0 -
Just one layer - and it really slows down a lot during shooting tethered and takes forever to display the images. Usually it's super snappy...
I made the adjustments in a few breaks between shooting and applying them to the images took to long for me to stomach (had to abort, the client can't wait forever 🙄 ) and then I had the adjustments ready and wanted to automatically apply them to the images while shooting... but as I said, it takes a looong time.0 -
Update: Yep, it's the Luma Range setting. When I deleted it, it went back to its normal performance....
Already submitted a case, I'll keep you updated here if I find a solution/workaround. Seems like it doesn't matter where you put your Luma Range, performance just tanks as soon as you use any setting.0 -
Update 2: I've just heard back from Support and indeed this is to be expected and normal.
Luma Range has to be calculated for each and every image - which is understandable even though it might be a bit frustrating when you're not aware of it.0 -
[quote="C-M-B" wrote:
Update 2: I've just heard back from Support and indeed this is to be expected and normal.
Luma Range has to be calculated for each and every image - which is understandable even though it might be a bit frustrating when you're not aware of it.
David Grover ran a webinar intro to the Luma Range tool last week and it was quite clear that performance took a bit of a hit on a range of similar photos as the affected pixels had to be identified for every image individually as expected.
That said pasting the layer from one image to 5 others did not really take too long considering the results to be obtained in just a few seconds.
However if the whole process is "held" in memory in case of cancellation or reversal of the transaction and you are working with large number of large files then maybe the memory management is something to be checked. Certainly it should not crash but you may have inadvertently activated something that had not been specifically considered in the design nor tested by users during the beta.
I suspect that with 25GB memory used you were probably well into performance sapping memory swapping territory and that is never good for performance.
I suspect some finessing of the processes internal controls and limits may be desirable. All part of the development process.
Grant0 -
Yeah I agree - that's why I'm not angry about this situation, as it's a new feature there's always a pretty big room for improvements.
I've only used this feature with 1-2 single images before and that worked rather well - but as you said; it's probably not really made for a huge batch process (yet).0 -
Hmm.
I might test it .... just out of interest.0 -
after you have created your luma mask you can "rasterize"it. this will dramatically improve performance when you copy the adjustsments. 😊 0 -
[quote="djwga" wrote:
after you have created your luma mask you can "rasterize"it. this will dramatically improve performance when you copy the adjustsments. 😊
But is not going to give you what you want from a Luma mask copied across multiple images - or at least it will not do so in the majority of use cases.
If you can use the Luma based mask rasterized to a fixed mask for other images - and there are probably some use cases where that would be fine - then by all means use it that way. But then you are not copying and applying a Luma defined mask in a layer.
Grant0 -
[quote="SFA" wrote:
[quote="djwga" wrote:
after you have created your luma mask you can "rasterize"it. this will dramatically improve performance when you copy the adjustsments. 😊
But is not going to give you what you want from a Luma mask copied across multiple images - or at least it will not do so in the majority of use cases.
If you can use the Luma based mask rasterized to a fixed mask for other images - and there are probably some use cases where that would be fine - then by all means use it that way. But then you are not copying and applying a Luma defined mask in a layer.
Grant
Exactly! Specially when shooting people the rasterizing option is completely useless0 -
that is true, the rasterized version is static. For what its worth keeping sensitivity & Radius at a low value improves the batch performance 0 -
The thing is that you really want the facility to provide the effect across all of the selected images and the effect is probably the main consideration rather than how quickly something can be applied to a batch of images.
Reducing the level of detail will certainly allow processes to run more quickly and if that is OK for you as far as the results are concerned then it's fine to go for that approach. Indeed it would be sensible to do so.
But if the subtlety matters and the files are large and complex in terms of colour and structures, then there will be more data to manipulate and processing will take longer. Or the complexity of the selection can be reduced if that would still work for the changes required.
Grant0 -
[quote="SFA" wrote:
The thing is that you really want the facility to provide the effect across all of the selected images and the effect is probably the main consideration rather than how quickly something can be applied to a batch of images.
Reducing the level of detail will certainly allow processes to run more quickly and if that is OK for you as far as the results are concerned then it's fine to go for that approach. Indeed it would be sensible to do so.
But if the subtlety matters and the files are large and complex in terms of colour and structures, then there will be more data to manipulate and processing will take longer. Or the complexity of the selection can be reduced if that would still work for the changes required.
Grant
That might be fine for post-shoot editing the images but that does not help during tethered shooting with a client, who wants a fast but reliable (!) preview of the images you're taking.
I don't have a slow computer (desktop 7700K with 32GB RAM, 512GB NvME SSD and a GTX 1070), yet it takes an extraordinary amount of time for every single image to render properly during shooting - and that's "only" with 60MP. I can't image how long it would take with a 100MP or 150MP back.
If I know the performance drop in advance, that's fine - but that should be mentioned somewhere. Now that I know, I can work around that and find other solutions for it. But it wasn't a very ideal situation, no matter how forgiving your client might be.0 -
[quote="C-M-B" wrote:
[quote="SFA" wrote:
The thing is that you really want the facility to provide the effect across all of the selected images and the effect is probably the main consideration rather than how quickly something can be applied to a batch of images.
Reducing the level of detail will certainly allow processes to run more quickly and if that is OK for you as far as the results are concerned then it's fine to go for that approach. Indeed it would be sensible to do so.
But if the subtlety matters and the files are large and complex in terms of colour and structures, then there will be more data to manipulate and processing will take longer. Or the complexity of the selection can be reduced if that would still work for the changes required.
Grant
That might be fine for post-shoot editing the images but that does not help during tethered shooting with a client, who wants a fast but reliable (!) preview of the images you're taking.
I don't have a slow computer (desktop 7700K with 32GB RAM, 512GB NvME SSD and a GTX 1070), yet it takes an extraordinary amount of time for every single image to render properly during shooting - and that's "only" with 60MP. I can't image how long it would take with a 100MP or 150MP back.
If I know the performance drop in advance, that's fine - but that should be mentioned somewhere. Now that I know, I can work around that and find other solutions for it. But it wasn't a very ideal situation, no matter how forgiving your client might be.
I guess the question, for live use with tethered shooting, is whether one needs the level of edit detail that reaches almost a "final edit" situation for the client to understand the possibilities or whether, using a reduced mask for speed, the client can understand the intent and accept that further enhancement can be applied most effectively once they have made their selections.
I can just imagine a Papal official telling Michelangelo that waiting 4 years for a finished product was not an option and giving him a week to deliver a good approximation of the final result of his labours ... 😉0 -
Well when you have a "smooth" gradient over the whole luma spectrum.... then yes, that's quite a hefty differnece and not a "slight" difference which would be bearly noticeable.
edit:[quote="SFA" wrote:
I can just imagine a Papal official telling Michelangelo that waiting 4 years for a finished product was not an option and giving him a week to deliver a good approximation of the final result of his labours ... 😉
I'm pretty sure Art Directors see themselves quite a bit beyond a Papal official and much closer to a god 😉
I don't really want to find out what happens when I tell them it'll take me 4 years to show them the preview files0 -
This is an interesting discussion. I don't shoot tethered so I will not suffer the way you are but as I am reading I wonder...
Given what you have observed lets say I take 200 shots on a shoot and I want to use a luma mask to make the same adjustment to all of them will I bring my machine to a crawl when I apply an adjusted layer to all 200 images? If so then what about after I am done will things be slow every time I open that collection of 200 images?
I guess I am a bit confused. Is the problem that each edited image holds onto more memory than previous versions due to the use of the luma mask and as you add more and more images that you have edited things slow down? Or does this process take too long to complete for the last tethered shot such that it's not done when the next one is taken and therefore C1 gets further and further behind?0 -
[quote="IanL" wrote:
This is an interesting discussion. I don't shoot tethered so I will not suffer the way you are but as I am reading I wonder...
Given what you have observed lets say I take 200 shots on a shoot and I want to use a luma mask to make the same adjustment to all of them will I bring my machine to a crawl when I apply an adjusted layer to all 200 images? If so then what about after I am done will things be slow every time I open that collection of 200 images?
I guess I am a bit confused. Is the problem that each edited image holds onto more memory than previous versions due to the use of the luma mask and as you add more and more images that you have edited things slow down? Or does this process take too long to complete for the last tethered shot such that it's not done when the next one is taken and therefore C1 gets further and further behind?
The problem only arises when the Luma Range is applied to the image and calculated - and that's what takes so long. Afterwards it's as quick to open and close and display as a "regular" image. And the high memory use also only appears during application/calculation.
During tethering it slows down the rendering of the images considerably as CPU usage spikes as 100% (the files will get transfered though)0 -
Thanks - hopefully that is something they can work to improve for you then. 0
投稿コメントは受け付けていません。
コメント
20件のコメント