That may work for a simple media player scenario. What if you are working with a multitrack composition and need to determine the image format for frame N where there is potentially a video on each track? Then, you have to repeat the logic that MLT does to determine which producer of which track will be used.
A while ago, I worked on something for a client with a similar request. If you specify format mlt_image_none and set width and height to zero, then you will get back an image in some mlt_image_format that is closest to source format (as long as the producer correctly handles mlt_image_none request format in its get_image function). So, MLT can do the above logic for you. There should exist similar functionality for mlt_audio_none. To make this work while continuing to use the "loader" producer, you must specify to use the "abnormal" to suppress the normalization filters. But, like I said, once you start using MLT modules' filters and transitions with this capability, you can expect problems. Also, you must be aware that if the consumer "real_time" property is not zero, then there are worker/rendering threads that request an image format. They default to mlt_image_yuv422, but you can override that with the "mlt_image_format" property: https://github.com/mltframework/mlt/blob/master/src/framework/mlt_consumer.c#L475 On Wed, Jul 27, 2016 at 1:01 PM Brian Matherly <c...@brianmatherly.com> wrote: > Depending on the producer, you might be able to query the image metadata > and determine the native format. > > For example, the avformat_producer will often set the pix_fmt. Try to > query "meta.media.0.codec.pix_fmt" on a frame from the avformat producer > widget. > > Perhaps you could query the pix_fmt on the first frame and falllback to a > default format if it is not set. > > ~Brian > > > ------------------------------ > *From:* Paweł Goliński <golp...@gmail.com> > *To:* mlt-devel@lists.sourceforge.net > *Sent:* Wednesday, July 27, 2016 11:24 AM > *Subject:* [Mlt-devel] Avoiding implicit conversions when fetching > image/audio from frame > > Hi, > > I want to pass as much of rescaling/resampling/color conversion tasks > to my consumer as I can. So, when calling, e.g. mlt_frame_get_image() > for image, I want to use parameters that guarantee the least number of > implicit > conversions will take place, as I can take care of them in my consumer. > What’s > the best way to do that? Is there some way to ask a „loader" producer to > produce > „the most native” frame possible? > > Example - when I pass „mlt_image_rgb24a” to mlt_frame_get_image(), the MLT > loader producer does its best to produce what I want, so on some video > sources > there is an implicit y420->rgba conversion. But my consumer could handle > y420 > as well, so I could avoid this conversion. Question is, how to know which > format > to request, to avoid such conversion? > > > Best, > Paweł > > ------------------------------------------------------------------------------ > What NetFlow Analyzer can do for you? Monitors network bandwidth and > traffic > patterns at an interface-level. Reveals which users, apps, and protocols > are > consuming the most bandwidth. Provides multi-vendor support for NetFlow, > J-Flow, sFlow and other flows. Make informed decisions using capacity > planning > reports.http://sdm.link/zohodev2dev > _______________________________________________ > Mlt-devel mailing list > Mlt-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/mlt-devel > > > > ------------------------------------------------------------------------------ > _______________________________________________ > Mlt-devel mailing list > Mlt-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/mlt-devel >
------------------------------------------------------------------------------
_______________________________________________ Mlt-devel mailing list Mlt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mlt-devel
