May 2007 - Pro-apps-dev - lists.apple.com

Re: FxPlug versioning API
by Darrin Cardani 03 May '07

03 May '07

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: FxPlug versioning API
by Darrin Cardani 03 May '07

03 May '07

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: FCP6 TemporalImageAPI - returning an image different from main input
by Micah Sharp 03 May '07

03 May '07

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com Good point on the rowbytes, we'll give that a look. Thanks, Micah On May 2, 2007, at 8:39 PM, Paul Schneider wrote: 1) Get 720x240 field buffer at time n 2) Get 720x240 field buffer at time n + .5 3) Combine the two fields (field order correct of course) Yes, that should let you reconstitute a full interlaced frame. You can use the TemporalAPI to get the "other" field. - Paul On May 2, 2007, at 9:21 PM, Pierre Jasmin wrote: 1) Get 720x240 field buffer at time n 2) Get 720x240 field buffer at time n + .5 3) Combine the two fields (field order correct of course) Thanks, Micah On May 2, 2007, at 9:57 AM, Paul Schneider wrote: This email sent to jasmin(a)revisionfx.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… Yeah, that is what I meant. Not during render, but using the temporal API. Good to know, because that is what we do and it seems to work well. If we are getting fields back from the temporal API, but we want a full interlaced frame, is the following a valid way for us to re-create the interlaced frame: No unless you cache the first field result as a full frame internally as you will be called twice, the second time at frameNum.5 If you are working with FxBitmaps, and you are running in FCP, you might notice that the two fields point to the same memory, and we have done clever tricks with rowbytes to make it appear that you are only looking at a single field. You can probably use this technique to access the full frame more efficiently than using the temporalAPI and manually combining frames. I wouldn't expect FCP to change its behavior in this regard anytime soon. Micah Sharp wrote: If we are getting fields back from the temporal API, but we want a full interlaced frame, is the following a valid way for us to re-create the interlaced frame: No unless you cache the first field result as a full frame internally as you will be called twice, the second time at frameNum.5 We do need actual fields in our plugin and not something that is line doubled. And Paul is right that media does get scaled automatically, which is nasty, but I thought after much searching I found a way to turn that off. Yet another manual entry for us I think. Right now, FCP's implementation of the temporalAPI will always give you the native format of the media. In other words, if you use the temporalAPI on a progressive clip, you'll get a full frame back, even if you are in an interlaced timeline. This is arguably confusing, since it's not the same as the input image that we pass to filters (input images are in the timeline format, not the media format). I think Paul Miller has a good point here. _______________________________________________ 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/jasmin% 40revisionfx.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/pschneider% 40apple.com This email sent to pschneider(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/micah% 40redgiantsoftware.com This email sent to micah(a)redgiantsoftware.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: FCP6 TemporalImageAPI - returning an image different from main input
by Gary Oberbrunner 03 May '07

03 May '07

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com User-agent: Thunderbird 2.0.0.4pre (Windows/20070502) Peter Litwinowicz wrote: > I *rather* not worry about it, but frankly that gets me into trouble. > And burying your head into the sand doesn’t make things better. > > ... > Same with frame rates, resolution, coordinate spaces, etc. > > The more the app tries to make everything “nice”, the more “nice” > becomes a dirty word. +1 from me. Give us the real data. Just make sure what you're giving us is well described and documented so we can make sense of it. -- . . . . . . . . . . . . . . . . . . . . . . . . . Gary Oberbrunner garyo(a)genarts.com GenArts, Inc. Tel: 617-492-2888 955 Mass. Ave Fax: 617-492-2852 Cambridge, MA 02139 USA www.genarts.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: FCP6 TemporalImageAPI - returning an image different from main input
by Pierre Jasmin 03 May '07

03 May '07

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com Organization: RE:Vision Effects User-agent: Thunderbird 1.5.0.10 (Windows/20070221) Pierre _______________________________________________ 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… // did not go through as I was exceeding 12 bytes Here's a random question for all of you - AE/Premiere are doing a lot of resampling when they hand you full-size frames that were generated from fields, right? (Or are they just line-doubling the fields?) if the source is interlaced, Adobe Pr and Ae or Combustion deinterlace using field blending - to help speed up process if your source is footage it will tag lower or upper as an hint so you can skip scanlines (Premiere always set that on every frame height fields, AE only when the source is footage) - in Premiere the timeline advances by 0.5 increments when your project is in fields, in AE you have to set the comp to 59.94 to see both fields - in AE in RAM previews at 29.97... the other field is skipped if the source is not interlaced then it will pass you a progressive frame (fields happen only at the end via subsampling) if the output is 59.94 fields per second it renders 59.94 frames per second and reinterlaces if the output is 29.97 no fields and the source material is interlaced it will skip a field when rendering This to say it's ok to skip pixels when you scrub the timeline... for the sanity check test maybe render in AE a number counter movie with fields that advances (+1) one per field and color each field differently (eg red for lower and green for upper) - apply a blur: it should never turn yellow etc... (just to explain it's not necessarily a time only thing) - apply a blur on a progressive source in an interlaced output project or a progressive output project and you should not get a blur that is twice as high in one case vs the other make sure you use 29.97002997 when moving data to AE and from AE to FCP... (as OMF, Premiere, FCP... all are on the 29.97002997002997 NTSC period and not 29.97 - in fact the Premiere - AE link in prev cycle is sort of broken in fields because of that - you don't see it because by default AE shows you only one field but that difference will make you fall on the other side of the .5 on the second field, so in effect if you render a 29.97 movie in AE, import in Premiere and render that movie NTSC with the default fields based setting and import without changing the interp in AE, then look at it in 59.94, one field is gone (field 1 and 2 are the same). A long term solution could be to simply defer upscaling fields to full height frames until some tool requests a full frame, so then you could stay upsampled until the output. what you don't want to do in life is at each step scale and resample down, only scale once if needed, only resample down once at the end if needed. The same thing could be said about color space, you don't want to do YUV to RGB to YUV at each filter when you have many consecutive ones. You maybe want to do once YUV to RGB if the filter only works in RGB and then at the end convert back to YUV if the output codec is YUV... This email sent to site_archiver(a)lists.apple.com

1 0

Re: FCP6 TemporalImageAPI - returning an image different from main input
by Paul Schneider 03 May '07

03 May '07

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com 1) Get 720x240 field buffer at time n 2) Get 720x240 field buffer at time n + .5 3) Combine the two fields (field order correct of course) Yes, that should let you reconstitute a full interlaced frame. You can use the TemporalAPI to get the "other" field. - Paul On May 2, 2007, at 9:21 PM, Pierre Jasmin wrote: 1) Get 720x240 field buffer at time n 2) Get 720x240 field buffer at time n + .5 3) Combine the two fields (field order correct of course) Thanks, Micah On May 2, 2007, at 9:57 AM, Paul Schneider wrote: This email sent to jasmin(a)revisionfx.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… If we are getting fields back from the temporal API, but we want a full interlaced frame, is the following a valid way for us to re- create the interlaced frame: No unless you cache the first field result as a full frame internally as you will be called twice, the second time at frameNum.5 If you are working with FxBitmaps, and you are running in FCP, you might notice that the two fields point to the same memory, and we have done clever tricks with rowbytes to make it appear that you are only looking at a single field. You can probably use this technique to access the full frame more efficiently than using the temporalAPI and manually combining frames. I wouldn't expect FCP to change its behavior in this regard anytime soon. Micah Sharp wrote: If we are getting fields back from the temporal API, but we want a full interlaced frame, is the following a valid way for us to re- create the interlaced frame: No unless you cache the first field result as a full frame internally as you will be called twice, the second time at frameNum.5 We do need actual fields in our plugin and not something that is line doubled. And Paul is right that media does get scaled automatically, which is nasty, but I thought after much searching I found a way to turn that off. Yet another manual entry for us I think. Right now, FCP's implementation of the temporalAPI will always give you the native format of the media. In other words, if you use the temporalAPI on a progressive clip, you'll get a full frame back, even if you are in an interlaced timeline. This is arguably confusing, since it's not the same as the input image that we pass to filters (input images are in the timeline format, not the media format). I think Paul Miller has a good point here. _______________________________________________ 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/jasmin% 40revisionfx.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/pschneider% 40apple.com This email sent to pschneider(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: FCP6 TemporalImageAPI - returning an image different from main input
by Pierre Jasmin 03 May '07

03 May '07

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com Organization: RE:Vision Effects User-agent: Thunderbird 1.5.0.10 (Windows/20070221) 1) Get 720x240 field buffer at time n 2) Get 720x240 field buffer at time n + .5 3) Combine the two fields (field order correct of course) Thanks, Micah On May 2, 2007, at 9:57 AM, Paul Schneider wrote: This email sent to jasmin(a)revisionfx.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… Micah Sharp wrote: If we are getting fields back from the temporal API, but we want a full interlaced frame, is the following a valid way for us to re-create the interlaced frame: No unless you cache the first field result as a full frame internally as you will be called twice, the second time at frameNum.5 We do need actual fields in our plugin and not something that is line doubled. And Paul is right that media does get scaled automatically, which is nasty, but I thought after much searching I found a way to turn that off. Yet another manual entry for us I think. Right now, FCP's implementation of the temporalAPI will always give you the native format of the media. In other words, if you use the temporalAPI on a progressive clip, you'll get a full frame back, even if you are in an interlaced timeline. This is arguably confusing, since it's not the same as the input image that we pass to filters (input images are in the timeline format, not the media format). I think Paul Miller has a good point here. _______________________________________________ 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/jasmin%40revisionfx.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: FCP6 TemporalImageAPI - returning an image different from main input
by Micah Sharp 03 May '07

03 May '07

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com 1) Get 720x240 field buffer at time n 2) Get 720x240 field buffer at time n + .5 3) Combine the two fields (field order correct of course) Thanks, Micah On May 2, 2007, at 9:57 AM, Paul Schneider 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/site_archiver%40lists.a… If we are getting fields back from the temporal API, but we want a full interlaced frame, is the following a valid way for us to re- create the interlaced frame: We do need actual fields in our plugin and not something that is line doubled. And Paul is right that media does get scaled automatically, which is nasty, but I thought after much searching I found a way to turn that off. Yet another manual entry for us I think. Right now, FCP's implementation of the temporalAPI will always give you the native format of the media. In other words, if you use the temporalAPI on a progressive clip, you'll get a full frame back, even if you are in an interlaced timeline. This is arguably confusing, since it's not the same as the input image that we pass to filters (input images are in the timeline format, not the media format). I think Paul Miller has a good point here. This email sent to site_archiver(a)lists.apple.com

1 0

Re: FCP6 TemporalImageAPI - returning an image different from main input
by Peter Litwinowicz 02 May '07

02 May '07

site_archiver(a)lists.apple.com Delivered-To: pro-apps-dev(a)lists.apple.com Thread-index: AceNE/sZOd9HxvkHEdu4WwAKlbriPA== Thread-topic: FCP6 TemporalImageAPI - returning an image different from main input User-agent: Microsoft-Entourage/11.3.3.061214 _______________________________________________ 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: FCP6 TemporalImageAPI - returning an image different from main input
by Darrin Cardani 02 May '07

02 May '07

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