Dear Harley, You and Phil have both touched on an important issue (albeit in different ways). Just as in the TV case, there are different points in the chain where a user may wish to make a quality measurement. Depending on where the measurement is made, the reference may or may not be available. Many users are very interested in full-reference measurement methods for compression and transcoding applications. Therefore, I would assert that the multimedia effort must address both full-reference and reduced reference/no reference methods just as in the TV case. The fundamental issue is the same, namely, different measurement methods are needed for different parts of the chain (i.e., different applications). This begs the eternal question, which application area do we address first? Contrary to Michele's proposal, my experience with users indicates that there is little need for quality measurement for live streaming. As Phil has pointed out, if there is a quality issue with a stream, there is little that can be done to correct the problem. In addition, the majority of multimedia material available today is not streamed live, it is downloaded or is available on demand. I encourage others to offer their perspectives on user requirements so we may reach consensus on the appropriate application areas and measurement methodologies to address. Regards, Ann Marie > -----Original Message----- > From: H. Myler [mailto:hrm@engr.ucf.edu] > Sent: Tuesday, July 17, 2001 11:21 AM > To: VQEG > Subject: Re: [VQEG] Comments on MM test plan > > > At 9:53 AM -0700 7/17/01, Corriveau, Philip J wrote: > >Dear friends and colleagues, > > > >With the upcoming MM meeting there are a few issues that > >are still wide open. > > > >We do not know what the application area is ! > > Indeed! This is the initial item for discussion on the agenda. > > >We also must include Audio if this test is to be > >a true MM test, otherwise we will not be taken > >seriously by the user community. > > An item for debate. We are looking at video quality of a > multimedia stream--some would argue that audio is not > important for that reason. You make a valid point though and > we need to discuss it. > > >We do not decide on a test methodology until we > >have flushed out the application area and exact > >goals and then we fit the method to the area > >being tested. Not the other way around.. > > Ann Marie pointed this out some time ago, hence the agenda > change. However, my goal was to start with a straw man test > plan of sorts to stimulate discussion. > > >Encoding, Transmission, Decoding and Display seem > >to be the individual components. > > Agreed, but does this include original source? Do we want to > examine video that has undergone some preprocess prior to the > encoding? In other words, do we fix the original source > format or allow for permutations across a set of input HRCs? > Are we concerned with the transmission channel? Should we > assume a fixed bandwidth and go from there? Should our focus > be solely on the final displayed video? What are realistic > formats for this video? > > >Are we putting ourselves into the life streaming hole > >since this is not reality.. There is no business > >model for streaming video/audio since there is no QoS controls > >on the internet. > >What type of measurement is required by the customers? > > > >Have we done any research into what the industry really > >needs in this application area ? > > > Michele posted some material that addresses these issues, but > further comment is welcome. > > > >Just a few thought provoking comments and questions. > > > >Philip > > > > > Thanks Philip, your comments are both appropriate and > helpful. Let's get some discussion going on these issues > before everyone leaves for Orlando next week. > > --Harley > > > _______________________________________________ > ituvidq mailing list > ituvidq@its.bldrdoc.gov > http://www.its.bldrdoc.gov/mailman/listinfo/ituvidq > _______________________________________________ ituvidq mailing list ituvidq@its.bldrdoc.gov http://www.its.bldrdoc.gov/mailman/listinfo/ituvidq