July 2008 - Pro-apps-dev - lists.apple.com

Re: note on problem reports
by Paul Schneider 22 Jul '08

22 Jul '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com Hi Kristian, Both FCP and Motion expect premultiplied output from FxPlugs. - Paul On Jul 22, 2008, at 5:42 AM, Kristian Harms wrote: Thanks for your answers, both Steve and you :-) -- Kristian Harms _______________________________________________ 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… Okay, but it seems that to provide consistent output for both Motion and FCP, I'll have to premultiply in one host but not the other. Can you confirm that this is the normal way to do this? Also, since I don't seem to get any information with which I can differentiate between the two cases, except -hostIsMotion et al. FCP will often convert your output back to straight to apply more processing. Motion works premultiplied internally, so in a simple test (one layer, just your generator in a timeline) it will just display your color channels directly to the screen. But you will still see problems when Motion starts using your alpha channel in more interesting ways (adding a shape, or working with more layers). Anyway, isn't half the point of making a generator instead of a filter that I should simply include an alpha channel and leave it to the host app to blend with lower layers? The host composites your output with the layers below, whether you are a filter or a generator (or a transition). But the rules for compositing premultiplied images are different than those for compositing straight images, as I'm sure you're aware. One reason why many compositing applications prefer working premultiplied is that the typical "over" operation is so much simpler; compositing two premultiplied images can be done by simply adding the color channels. Den 21. juli. 2008 kl. 18.32 skrev Darrin Cardani: The third problem is only in Motion: If an FxPlug effect generates pixels that are fully transparent, but have the R, G and B channels set to some non-zero value (e.g. 127 if we're talking UInt8), then Motion won't treat these pixels as fully transparent, but will instead show any underlying layer as "washed out", and the black default background will shine through as grey. This is with the normal blending modes. (Apple guys, the problem ID is #6079465.) As Steve mentioned, this is a problem on your end (and you're not the first person to have this problem). You just need to premultiply your pixel values, and everything will work as expected. Okay, but it seems that to provide consistent output for both Motion and FCP, I'll have to premultiply in one host but not the other. Can you confirm that this is the normal way to do this? Also, since I don't seem to get any information with which I can differentiate between the two cases, except -hostIsMotion et al. Anyway, isn't half the point of making a generator instead of a filter that I should simply include an alpha channel and leave it to the host app to blend with lower layers? This email sent to pschneider(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: note on problem reports
by Kristian Harms 22 Jul '08

22 Jul '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: note on problem reports
by Kristian Harms 22 Jul '08

22 Jul '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com Thanks for your answers, both Steve and you :-) -- Kristian Harms _______________________________________________ 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… Den 21. juli. 2008 kl. 18.32 skrev Darrin Cardani: The third problem is only in Motion: If an FxPlug effect generates pixels that are fully transparent, but have the R, G and B channels set to some non-zero value (e.g. 127 if we're talking UInt8), then Motion won't treat these pixels as fully transparent, but will instead show any underlying layer as "washed out", and the black default background will shine through as grey. This is with the normal blending modes. (Apple guys, the problem ID is #6079465.) As Steve mentioned, this is a problem on your end (and you're not the first person to have this problem). You just need to premultiply your pixel values, and everything will work as expected. Okay, but it seems that to provide consistent output for both Motion and FCP, I'll have to premultiply in one host but not the other. Can you confirm that this is the normal way to do this? Also, since I don't seem to get any information with which I can differentiate between the two cases, except -hostIsMotion et al. Anyway, isn't half the point of making a generator instead of a filter that I should simply include an alpha channel and leave it to the host app to blend with lower layers? This email sent to site_archiver(a)lists.apple.com

1 0

Re: note on problem reports
by Steve Christensen 21 Jul '08

21 Jul '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: Auxiliary timecode and XML
by Bruce Sharpe 21 Jul '08

21 Jul '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:message-id:date:from:to :subject:cc:in-reply-to:mime-version:content-type :content-transfer-encoding:content-disposition:references; bh=MhfIong0Q0t5SUt1Ze8PICkxPaW9Alu8Yh/NnDMCh68=; b=Hz0a5+MHh4QInLRWfrlNzAkN8hvOYVtwzZSM6UbTMQfEpPBddE3nh806NuuwmK5CWh J/sMmTLBlCfHj+ZOsnJfM0meMMlFAr2wvH18RaEVnIWj3yT/j3GB8US5xGe0nAwgHrWk DHmpxAE7umCD5abSfFu8chL6WI2kn9RZrbDNI= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=message-id:date:from:to:subject:cc:in-reply-to:mime-version :content-type:content-transfer-encoding:content-disposition :references; b=ENxVmSir03kkXRjtvo8VpTVERDJivNPU/wwsSUy+IeqtHgXlHSXJmHJF1ODQsvUR4r WLI5fCPmZmvljBNF+rKw/Zi1doMx0sWjuL/NqEDYchN5OvUJ+sTAhBPo98l4i2ZpRyXj 5zpvwStcPjugkn+wEYGUTECSqIgrPQ2qK1G+Q= Great, thanks for clarifying that, Helena. My mental model of a multiclip continues to evolve... Bruce On Mon, Jul 21, 2008 at 9:58 AM, Helena Ju <hju(a)apple.com> wrote: > Hi Bruce! > Actually, in the encoding of the active angle's clipitem (the one that > contains <multiclip>) there is also a <mediadelay>. This should be the same > as the one for the active angle encoded at >> >> .../multiclip/angle/clip/media/video/track/clipitem/mediadelay > > Cheers, > Helena > On Jul 19, 2008, at 3:42 PM, Bruce Sharpe wrote: > >> Hi Helena, >> >> By the way, I think there is an error in the example usage of >> mediadelay in Listing 3-9 of the PDF (page 23). It shows mediadelay >> as a sibling of multiclip, but the correct position in the hierarchy >> is: >> .../multiclip/angle/clip/media/video/track/clipitem/mediadelay >> >> If you agree, I'll file a bug report. >> >> Bruce >> >> On Fri, Jul 18, 2008 at 9:18 PM, Bruce Sharpe <bruce.sharpe(a)gmail.com> >> wrote: >>> >>> Hi Helena, >>> >>> Yup, I think mediadelay is going to do the job. >>> >>> Is it exposed in the UI anywhere or is it only accessible via XML? >>> >>> Thanks! >>> Bruce >>> >>> On Fri, Jul 18, 2008 at 11:31 AM, Helena Ju <hju(a)apple.com> wrote: >>>> >>>> Hi Bruce! Pattrick! >>>> To do the syncronization you've both talked about where various clips >>>> you'd like to use in a multiclip may not overlap, please check out the >>>> "mediadelay" key. (described on page 42 of the PDF in figure 3-3. I believe >>>> it fills the need you have. Certainly that's what FCP is using when we have >>>> multiclips like that. >>>> Cheers! >>>> Helena >>>> PS if that doesn't fill the need, please let me know and give me an >>>> example to play with. >>>> On Jul 18, 2008, at 8:56 AM, Bruce Sharpe wrote: >>>> >>>> Filed as bug number 6085957. >>>> >>>> Thanks, >>>> Bruce >>>> >>>> On Thu, Jul 17, 2008 at 1:30 PM, Helena Ju <hju(a)apple.com> wrote: >>>>> >>>>> Hi Bruce, >>>>> >>>>> I hear you, but that one is much harder. Lots more edge cases. So it'll >>>>> be more difficult to get in. Please write up a bug and send me the bug >>>>> number. >>>>> >>>>> Cheers, >>>>> Helena >>>>> >>>>> On Jul 17, 2008, at 11:14 AM, Bruce Sharpe wrote: >>>>> >>>>>> It would be great to have it for media with timecode too. Example >>>>>> use case: a music video or a commercial with choreography that is shot >>>>>> as multiple takes on one camera. You want to use the media file time >>>>>> code for capturing purposes but an auxiliary time code not embedded in >>>>>> the media file for purposes of synchronization. >>>>>> >>>>>> Bruce >>>>>> >>>>>> On Thu, Jul 17, 2008 at 10:27 AM, Helena Ju <hju(a)apple.com> wrote: >>>>>>> >>>>>>> Hi Michael, Andreas, Bruce, >>>>>>> I hear you. Wes has also repeatedly asked for this one. For those of >>>>>>> you >>>>>>> asking, the reference bug number is 5519050. I would have the >>>>>>> documentation >>>>>>> changed, but I think instead, we'd really like to have the >>>>>>> application let >>>>>>> you change the timecode, at least for timecode-less media, right? >>>>>>> No promises, but I'll bring it up again and mention that y'all were >>>>>>> all >>>>>>> asking. >>>>>>> Thanks! >>>>>>> -Helena >>>>>>> On Jul 17, 2008, at 10:07 AM, Michael Miller wrote: >>>>>>> >>>>>>> We had the exact same problem here. I think either the featureset or >>>>>>> the >>>>>>> documentation must be misleading because this seems to be a common >>>>>>> question. >>>>>>> If an MOV doesn't have a TC track embedded is there any reason FCP >>>>>>> can't >>>>>>> just assign an arbitrary one to it based on what is in the XML? >>>>>>> >>>>>>> -Michael Miller >>>>>>> >>>>>>> On Thu, Jul 17, 2008 at 1:03 AM, Andreas Kiel >>>>>>> <andreas.kiel(a)spherico.com> >>>>>>> wrote: >>>>>>>> >>>>>>>> No. It doesn't work. It's not possible to use XML to "overwrite" or >>>>>>>> create >>>>>>>> file data. >>>>>>>> >>>>>>>> Andreas >>>>>>>> >>>>>>>> On 17.07.2008, at 06:53, Bruce Sharpe wrote: >>>>>>>> >>>>>>>>> I would like to add or modify auxililary timecode via XML, but am >>>>>>>>> having no success. For example, if a clip has an auxiliary >>>>>>>>> timecode >>>>>>>>> value I export the clip to XML, edit the timecode value in the XML >>>>>>>>> and >>>>>>>>> re-import it. The new timecode values is ignored. >>>>>>>>> >>>>>>>>> Is it supposed to work? >>>>>>>>> >>>>>>>>> Bruce >>>>>>>>> _______________________________________________ >>>>>>>>> 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/andreas.kiel%40spherico… >>>>>>>>> >>>>>>>>> This email sent to andreas.kiel(a)spherico.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/michaelmiller%40gmail.c… >>>>>>>> >>>>>>>> This email sent to michaelmiller(a)gmail.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/hju%40apple.com >>>>>>> >>>>>>> This email sent to hju(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/bruce.sharpe%40gmail.com >>>>>>> >>>>>>> This email sent to bruce.sharpe(a)gmail.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: Auxiliary timecode and XML
by Helena Ju 21 Jul '08

21 Jul '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: Auxiliary timecode and XML
by Helena Ju 21 Jul '08

21 Jul '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com Cheers, Helena On Jul 19, 2008, at 3:42 PM, Bruce Sharpe wrote: Hi Helena, By the way, I think there is an error in the example usage of mediadelay in Listing 3-9 of the PDF (page 23). It shows mediadelay as a sibling of multiclip, but the correct position in the hierarchy is: .../multiclip/angle/clip/media/video/track/clipitem/mediadelay If you agree, I'll file a bug report. Bruce Hi Helena, Yup, I think mediadelay is going to do the job. Is it exposed in the UI anywhere or is it only accessible via XML? Thanks! Bruce Filed as bug number 6085957. Thanks, Bruce Hi Bruce, Cheers, Helena On Jul 17, 2008, at 11:14 AM, Bruce Sharpe wrote: Bruce -Michael Miller Andreas On 17.07.2008, at 06:53, Bruce Sharpe wrote: Is it supposed to work? Bruce _______________________________________________ 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/andreas.kiel%40spherico… _______________________________________________ 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/michaelmiller%40gmail.c… _______________________________________________ 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 This email sent to hju(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/bruce.sharpe%40gmail.com This email sent to bruce.sharpe(a)gmail.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… Hi Bruce! Actually, in the encoding of the active angle's clipitem (the one that contains <multiclip>) there is also a <mediadelay>. This should be the same as the one for the active angle encoded at .../multiclip/angle/clip/media/video/track/clipitem/mediadelay On Fri, Jul 18, 2008 at 9:18 PM, Bruce Sharpe <bruce.sharpe(a)gmail.com> wrote: On Fri, Jul 18, 2008 at 11:31 AM, Helena Ju <hju(a)apple.com> wrote: Hi Bruce! Pattrick! To do the syncronization you've both talked about where various clips you'd like to use in a multiclip may not overlap, please check out the "mediadelay" key. (described on page 42 of the PDF in figure 3-3. I believe it fills the need you have. Certainly that's what FCP is using when we have multiclips like that. Cheers! Helena PS if that doesn't fill the need, please let me know and give me an example to play with. On Jul 18, 2008, at 8:56 AM, Bruce Sharpe wrote: On Thu, Jul 17, 2008 at 1:30 PM, Helena Ju <hju(a)apple.com> wrote: I hear you, but that one is much harder. Lots more edge cases. So it'll be more difficult to get in. Please write up a bug and send me the bug number. It would be great to have it for media with timecode too. Example use case: a music video or a commercial with choreography that is shot as multiple takes on one camera. You want to use the media file time code for capturing purposes but an auxiliary time code not embedded in the media file for purposes of synchronization. On Thu, Jul 17, 2008 at 10:27 AM, Helena Ju <hju(a)apple.com> wrote: Hi Michael, Andreas, Bruce, I hear you. Wes has also repeatedly asked for this one. For those of you asking, the reference bug number is 5519050. I would have the documentation changed, but I think instead, we'd really like to have the application let you change the timecode, at least for timecode-less media, right? No promises, but I'll bring it up again and mention that y'all were all asking. Thanks! -Helena On Jul 17, 2008, at 10:07 AM, Michael Miller wrote: We had the exact same problem here. I think either the featureset or the documentation must be misleading because this seems to be a common question. If an MOV doesn't have a TC track embedded is there any reason FCP can't just assign an arbitrary one to it based on what is in the XML? On Thu, Jul 17, 2008 at 1:03 AM, Andreas Kiel <andreas.kiel(a)spherico.com > wrote: No. It doesn't work. It's not possible to use XML to "overwrite" or create file data. I would like to add or modify auxililary timecode via XML, but am having no success. For example, if a clip has an auxiliary timecode value I export the clip to XML, edit the timecode value in the XML and re-import it. The new timecode values is ignored. This email sent to andreas.kiel(a)spherico.com This email sent to michaelmiller(a)gmail.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: note on problem reports
by Kristian Harms 21 Jul '08

21 Jul '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: note on problem reports
by Darrin Cardani 21 Jul '08

21 Jul '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: note on problem reports
by Steve Christensen 21 Jul '08

21 Jul '08

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com On Jul 21, 2008, at 4:06 AM, Kristian Harms wrote: What are you doing to determine the length of the clip? 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… I just wanted to give a quick note on some problems that I believe I've come across in FCP and Motion, and that I reported on bugreports.apple.com last week. Hopefully this information can help other developers as well. The first problem is fairly easy to grasp: If you have an FxPlug effect (I've only verified it for software generators) which varies over time and which somehow smoothly animates *proportionally to* where one is within the effect (e.g., imagine a bar that starts off as nothing and then grows to reach the full width of the frame at the instant the effect ends), then FCP won't handle partial re- renderings correctly. Imagine you have such an effect, and drag the end point further out to the right, then only the new part of the effect will be marked as red (needing render), and the rendered result you then get will have the bar grow to the full width of the frame at the point where the effect previously ended, then instantly fall back to a smaller size and once again grow to the full width of the frame. Similar incorrect results are obtained by lengthening the effect using the start point, as well as by shortening the effect. (Apple devs, the problem ID is #6082523.) The second problem is probably the most serious: Let's say you have an FxPlug effect that not only has a custom parameter, but uses its value to determine how it should render (e.g. the string "red" would give red frames, the string "blue" would give blue frames). You're in FCP and send two arbitrary effects to motion using the right-click menu item, and apply this effect that has the custom parameter to either Motion project, setting one to blue and the other to red. After having saved both Motion projects and you get back into FCP and render, you'll notice that some frames that belong to one effect show up in the rendered result for the other. This effectively precludes using effects with custom parameters in Motion effects in FCP. (Apple guys, the problem ID is #6079258.) The third problem is only in Motion: If an FxPlug effect generates pixels that are fully transparent, but have the R, G and B channels set to some non-zero value (e.g. 127 if we're talking UInt8), then Motion won't treat these pixels as fully transparent, but will instead show any underlying layer as "washed out", and the black default background will shine through as grey. This is with the normal blending modes. (Apple guys, the problem ID is #6079465.) FxBitmaps are pre-multiplied so shouldn't ever see an ARGB pixel value of (0, 127, 127, 127). If you'd started out with a non pre- multiplied pixel with those component values, then when the pixel was pre-multiplied you should end up with (0, 127 * 0, 127 * 0, 127 * 0) == (0, 0, 0, 0). A few weeks ago I was on the list about effect duration being reported incorrectly using the information that comes with renderOutput: requests -- sometimes I was getting 3000 frames. I've not been able to reproduce this lately, but I will of course report it if and when I do. This email sent to site_archiver(a)lists.apple.com

1 0