October 2008 - Pro-apps-dev

Re: Releasing after -getBitmap -- memory not recovered?
by Steve Christensen 17 Oct '08

17 Oct '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com On Oct 17, 2008, at 9:33 AM, Matt Rhodes wrote: if (_needToCheckForImageWellImage) { FxBitmap* image; _imageWellHasImage = NO; if ([getParamAPI getBitmap:&image ...]) { if (image != nil) { _imageWellHasImage = YES; [image release]; } } _needToCheckForImageWellImage = NO; } steve _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/site_archiver%40lists.a… Two questions: 1. Is there a way to check that an Image Well has a bitmap/ texture without calling -getBitmap or -getTexture? I only need to check if it's empty. I don't know of any way to do that. If it is empty, -getBitmap and -getTexture return nil, as I'm sure you know. If it's not empty, you're eventually going to want to use the image, right? You could just keep it around until you need it. I guess I don't understand why you'd want to check if it's empty then not use the image it returns if it's not empty. It's a speed issue. There happens to be a situation where we only need to see what's populated (and we don't care what's in there). Not a big deal - just noticed it during my "where am I allocating memory and why" search. I guess it depends on where you're testing the state of the image well. If you're calling -getBitmap on every call to your plugin's rendering method(s) then, yeah, you're going to take a speed hit. How about doing something like this in -renderOutput:...? Obviously you'd need to set _needToCheckForImageWellImage=YES in your class' -initWithAPIManager: and -parameterChanged: methods. Even if you follow Paul's suggestion of calling - durationOfImageParm:forEffect:, you're still asking for information about the image only when there is a change. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Releasing after -getBitmap -- memory not recovered?
by Paul Schneider 17 Oct '08

17 Oct '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com Hi Matt, On Oct 17, 2008, at 11:33 AM, Matt Rhodes wrote: m _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/pschneider%40apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/site_archiver%40lists.a… I believe you can quickly check to see if an image parameter is empty using the FxTimingAPI. -durationOfImageParm: will return kFxDuration_Undefined if the parameter is empty. It's not the most intuitive, but it should be faster than grabbing the image. /*! @method durationOfImageParm:forEffect: @abstract Returns the duration of the clip assigned to the given image parameter for the given effect. @param parmId The ID of the image parameter. @param effect The plug-in object. @result The clip's duration, expressed in timeline frames, or the special value kFxDuration_Undefined if the image is empty. */ - (double)durationOfImageParm:(UInt32)parmId forEffect:(id<FxBaseEffect>)effect; Two questions: 1. Is there a way to check that an Image Well has a bitmap/texture without calling -getBitmap or -getTexture? I only need to check if it's empty. I don't know of any way to do that. If it is empty, -getBitmap and - getTexture return nil, as I'm sure you know. If it's not empty, you're eventually going to want to use the image, right? You could just keep it around until you need it. I guess I don't understand why you'd want to check if it's empty then not use the image it returns if it's not empty. It's a speed issue. There happens to be a situation where we only need to see what's populated (and we don't care what's in there). Not a big deal - just noticed it during my "where am I allocating memory and why" search. 2. I'm tracking down a memory leak in some software and I've found that if I watch the before and after memory usage during a -getBitmap and -release <bitmap>, I don't seem to recover everything allocated by -getBitmap. If a call to [parmsApi getBitmap:myMap ...] increases memory usage by 5 Megs, should I expect my 5 Megs back if I immediately call [myMap release]? Has anyone else had memory issues with -getBitmap? I'm running Motion v3.0.2. Motion will keep around some image-related data until rendering is done. For textures, it's not a lot of data, but for bitmaps, it is. But then, your users probably have significantly more system RAM than VRAM, so it's usually much less of a problem. I'm watching usage with Activity Monitor (probably not the best way to watch memory, but I do see my internal image memory being released back properly). You can also try the Leaks tool. In Xcode under the "Run" menu is the option "Start with Performance Tool," which has "Leaks" in the submenu. Thanks. I was able to do enough testing to rule out the Map get/ release. This email sent to pschneider(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: Releasing after -getBitmap -- memory not recovered?
by Matt Rhodes 17 Oct '08

17 Oct '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com Thanks. I was able to do enough testing to rule out the Map get/release. m _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/site_archiver%40lists.a… Two questions: 1. Is there a way to check that an Image Well has a bitmap/texture without calling -getBitmap or -getTexture? I only need to check if it's empty. I don't know of any way to do that. If it is empty, -getBitmap and - getTexture return nil, as I'm sure you know. If it's not empty, you're eventually going to want to use the image, right? You could just keep it around until you need it. I guess I don't understand why you'd want to check if it's empty then not use the image it returns if it's not empty. It's a speed issue. There happens to be a situation where we only need to see what's populated (and we don't care what's in there). Not a big deal - just noticed it during my "where am I allocating memory and why" search. 2. I'm tracking down a memory leak in some software and I've found that if I watch the before and after memory usage during a -getBitmap and -release <bitmap>, I don't seem to recover everything allocated by -getBitmap. If a call to [parmsApi getBitmap:myMap ...] increases memory usage by 5 Megs, should I expect my 5 Megs back if I immediately call [myMap release]? Has anyone else had memory issues with -getBitmap? I'm running Motion v3.0.2. Motion will keep around some image-related data until rendering is done. For textures, it's not a lot of data, but for bitmaps, it is. But then, your users probably have significantly more system RAM than VRAM, so it's usually much less of a problem. I'm watching usage with Activity Monitor (probably not the best way to watch memory, but I do see my internal image memory being released back properly). You can also try the Leaks tool. In Xcode under the "Run" menu is the option "Start with Performance Tool," which has "Leaks" in the submenu. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Releasing after -getBitmap -- memory not recovered?
by Darrin Cardani 17 Oct '08

17 Oct '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com On Oct 16, 2008, at 5:27 PM, Matt Rhodes wrote: Darrin -- Darrin Cardani dcardani(a)apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/site_archiver%40lists.a… Two questions: 1. Is there a way to check that an Image Well has a bitmap/texture without calling -getBitmap or -getTexture? I only need to check if it's empty. I don't know of any way to do that. If it is empty, -getBitmap and - getTexture return nil, as I'm sure you know. If it's not empty, you're eventually going to want to use the image, right? You could just keep it around until you need it. I guess I don't understand why you'd want to check if it's empty then not use the image it returns if it's not empty. 2. I'm tracking down a memory leak in some software and I've found that if I watch the before and after memory usage during a -getBitmap and -release <bitmap>, I don't seem to recover everything allocated by -getBitmap. If a call to [parmsApi getBitmap:myMap ...] increases memory usage by 5 Megs, should I expect my 5 Megs back if I immediately call [myMap release]? Has anyone else had memory issues with -getBitmap? I'm running Motion v3.0.2. Motion will keep around some image-related data until rendering is done. For textures, it's not a lot of data, but for bitmaps, it is. But then, your users probably have significantly more system RAM than VRAM, so it's usually much less of a problem. I'm watching usage with Activity Monitor (probably not the best way to watch memory, but I do see my internal image memory being released back properly). You can also try the Leaks tool. In Xcode under the "Run" menu is the option "Start with Performance Tool," which has "Leaks" in the submenu. This email sent to site_archiver(a)lists.apple.com

1 0

Premultiplication and fxscript
by Fiorani Gallotta Pierluigi 17 Oct '08

17 Oct '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/site_archiver%40lists.a… This email sent to site_archiver(a)lists.apple.com

1 0

Releasing after -getBitmap -- memory not recovered?
by Matt Rhodes 17 Oct '08

17 Oct '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com Hey Team. I'm running Motion v3.0.2. Thanks a lot. m _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/site_archiver%40lists.a… Two questions: 1. Is there a way to check that an Image Well has a bitmap/texture without calling -getBitmap or -getTexture? I only need to check if it's empty. 2. I'm tracking down a memory leak in some software and I've found that if I watch the before and after memory usage during a -getBitmap and -release <bitmap>, I don't seem to recover everything allocated by -getBitmap. If a call to [parmsApi getBitmap:myMap ...] increases memory usage by 5 Megs, should I expect my 5 Megs back if I immediately call [myMap release]? Has anyone else had memory issues with -getBitmap? I'm watching usage with Activity Monitor (probably not the best way to watch memory, but I do see my internal image memory being released back properly). I may be barking put the wrong tree, but I'm wondering if other developers have experience anything similar. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Final Cut Pro XML clip encodingand duration
by Helena Ju 14 Oct '08

14 Oct '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com Hey Bjørn, Best Luck! Helena On Oct 14, 2008, at 2:16 PM, Bjørn Holmgren wrote: _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/hju%40apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/site_archiver%40lists.a… The way the XML import works 1. it makes the whole project, all offline. 2. it reconnects the files if specified to in the XML import dialog or in the <importoptions>. At reconnect time, the durations would get updated, as long as the reconnect succeeds (which might not happen if there is a problem). But if the reconnect failed, or the XML did anything other than making browser clips, then a bogus duration won't work. For example, a sequence referencing items with bogus durations won't work since it happens in step 1 (before reconnect) and it is unclear how to edit things, or even if the source media is long enough. So, the updating of durations is intentional, but only at that one phase of the XML import, which you can guarantee we do by specifying it in the <importoptions>, but you can't guarantee that it succeeds (I've heard reports of specific servers preventing the reconnect from succeeding, for example). You're going to have to make the call about whether to use that hack or not. I am developing an application that searches through a folder on the network containing many thousands of clips, and generates a FCP XML with the search results. The problem is that I have no easy and fast way of determining the duration of the found clips (mdls does not work; no spotlight index). The manual for the XML interchange format indicates that duration is mandatory in the XML, and testing confirms that it is. The good news is that FCP seems to ignore the duration encoded in the XML, so I can write an arbitrary value on all clips, and FCP get the real duration from the source files. My question is: Is this behaviour intentional, or can I expect it to change at any time, thereby breaking my app? This is how I encode the clips: <clip id="asdf"> <name>asdf</name> <duration>25</duration> <rate> <ntsc>FALSE</ntsc> <timebase>25</timebase> </rate> <file id="asdf 2"> <name>asdf</name> <pathurl>file://localhost/Volumes/disk/dir/asdf</pathurl> </file> </clip> This email sent to hju(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Final Cut Pro XML clip encodingand duration
by Bjørn Holmgren 14 Oct '08

14 Oct '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/site_archiver%40lists.a… I am developing an application that searches through a folder on the network containing many thousands of clips, and generates a FCP XML with the search results. The problem is that I have no easy and fast way of determining the duration of the found clips (mdls does not work; no spotlight index). The manual for the XML interchange format indicates that duration is mandatory in the XML, and testing confirms that it is. The good news is that FCP seems to ignore the duration encoded in the XML, so I can write an arbitrary value on all clips, and FCP get the real duration from the source files. My question is: Is this behaviour intentional, or can I expect it to change at any time, thereby breaking my app? This is how I encode the clips: <clip id="asdf"> <name>asdf</name> <duration>25</duration> <rate> <ntsc>FALSE</ntsc> <timebase>25</timebase> </rate> <file id="asdf 2"> <name>asdf</name> <pathurl>file://localhost/Volumes/disk/dir/asdf</pathurl> </file> </clip> This email sent to site_archiver(a)lists.apple.com

1 0

Re: setting keyframes?
by Micah Sharp 13 Oct '08

13 Oct '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/site_archiver%40lists.a… This email sent to site_archiver(a)lists.apple.com

1 0

Re: setting keyframes?
by Paul Schneider 13 Oct '08

13 Oct '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/site_archiver%40lists.a… This email sent to site_archiver(a)lists.apple.com

1 0