Project Bento User Group update 3 with audio

Project Bento extends the avatar skeleton, adding a significant set of bones (e.g. 30 for the face, 30 for the hands (both of which can be seen inaction in the image above of an avatar by Matrice Laville), plus bones for wings, tails, additional limbs and ears / antennae
Project Bento extends the avatar skeleton, adding a significant set of bones (e.g. 30 for the face, 30 for the hands (both of which can be seen inaction in the image above of an avatar by Matrice Laville), plus bones for wings, tails, additional limbs and ears / antennae

The following notes and audio were taken from the weekly Bento User Group meeting, held on Aditi every Thursday at 13:00 SLT.

For details on each meeting and the location, please refer to the Bento User Group wiki page.

Bone Survey

The  Bento Bone Survey, intended to gauge the level of interest in various bone proposals for the Bento skeleton is now available. The Lab is requesting animators and content creators with a vested interest in Project Bento to rank the bone proposals to help with prioritising which additional bones people would like to see added to the Second Life avatar skeleton as a part of the Bento work.

If you are an animator / content creator with such an interest, and you haven’t already completed the survey, please ensure you do so as soon as possible, as it is likely to be closed on or shortly after Monday, February 1st.

The Lab notes they do not currently know how many additional bones we can reasonably add, or whether some of the associated bugs can be fixed.

Drax Files Filming

The final session of Bento filming for The Drax Files World Makers will take place on Monday,  February 1st at 11:00 SLT at the Bento Island region on Aditi. All those with an avatar developed using the Bento avatar skeleton extensions are invited to attend.

Deformed Avatars

Distortions with Altitude

Cathy Foil has been examining issues with deformations at altitude occurring in avatars using the Bento  skeleton. She first noticed the issue whilst Bento was in closed session and reported details to the Lab. However, efforts at pinning down he precise cause have been hampered up until recently by not being able to gather sufficient data from the skeleton and a lack of detailed comparisons between the effects as seen on various viewers utilising the Bento code.

Cathy has been addressing this using the most recent release of the Bento project viewer, and carrying out further detailed tests at various altitudes to document the severity of the issues more fully (they tend become noticeable at altitudes over 1000 metres). She has now passed her latest findings over to the Lab for their investigations.

Distortions and deformations in avatars using the Bento skeleton were noticed in the closed session Bento development cycle. Since then, Cathy Foil has been digging more deeply into the problems
Distortions and deformations in avatars using the Bento skeleton when at altitude (the greater the altitude, the more noticeable the issues, as seen left-to-right, above) were noticed in the closed session Bento development cycle. Since then, Cathy Foil has been digging more deeply into the problems

“Squished” Avatars”

Cathy has also been investigating the issue of avatars deforming as a result of using a mesh uploaded with different joint positions. This tend to result in avatars looking “squished” or otherwise deformed.

As she uses Maya, part of her investigations point towards possible issues with versions of the plug-ins used by Maya to convert .FBX format mesh files to .Collada .DAE (the format used by SL). Whether or not the plug-ins are the cause isn’t definitive as a cause, but may be a contributing factor. As Vir noted in the meeting, if this is the case, then it is unlikely the Lab can do much about it, but those using Maya affected by the issue may need to check the version of the Collada plug-in they are using, and make sure they have the most recent to see if this clears the problem.

If it is more of an issue with the mesh uploader, however (because the Lab have missed something, for example), then obviously, there is more of a chance of the Lab taking remedial action.

Cathy Foil discusses her investigations into avatar deformations at altitude

Joint Reset Issues

A long-term problem with Bento has been it is easy for joints to get into the wrong state and / or position when animating an avatar using the Bento skeleton extensions. It’s often not clear why this happens when it occurs, which can be somewhat random. Vir has started collecting data on the issue. Nothing definitive has been identified in terms of a cause or causes, but Vir will be continuing to work on the issue.

“First Frame” Animation / Default Avatar Pose Issues

Another issue which has been noted is that of a first frame of an animation begin played three times as the animation is run, and which may be connected to default avatar poses within Second Life. This is a fairly “old” bug, dating back as far as 2007 (see BUG-11194), and which seems to be more prevalent with Bento as a result of bones being used differently (e.g. arm bones being used as forelegs, etc).

Medhue demonstrates the forleg crossing using his wolf avatar. The issue is a lot more extensive than this, with the avatar sometimes looking as if it is standing with forelegs folded in the same way a human folds their arms
Medhue demonstrates the forleg crossing using his wolf avatar. The issue is a lot more extensive than this, with the avatar sometimes looking as if it is standing with forelegs folded in the same way a human folds their arms

Medhue Simoni demonstrated the issue using two of has non-human avatars, an elephant and a wolf, which saw both attempt to cross their front legs (as a human might try to fold their arms) when transitioning between animations.

A lengthy discussion followed on the problem, which might also involve a number of other issues, including poses “baked” into the default avatar skeleton, possible race conditions between the viewer and the server, and issues with animation prioritisation.

It may also be related to how Second Life handles transitions between animations (referred to as animations “easing out” and “easing in” when moving from one to another), which has had other problems reported against it as well – see also BUG-11122, for example.

As indicated in Vir’s comment on BUG-11194, and noted in the discussion, there is some initial concern from the Lab that if the playing in triplicate was an intentional change – and one made well before most developers currently at the Lab having joined the company – or if it requires symmetrical changes on both the viewer and the server, it might result in some content breakage without detailed analysis, although Vir has now started digging into the code to have a look.

The discussion on the subject took up a good part of the meeting, and is reproduced below as a series of extracts for ease of reference, which are also presented in the order in which they were discussed so that a complete picture of the the discussion might be obtained.

Medhue Simoni explains the first frame animation issues as it is now affecting Bento

OptimoMaximo offers an explanation of the possible conflict with animations baked into the default avatar skeleton

Coyot Linden offers-up a theory on animation interleaving which might be occurring

General discussion on animation systems, including the server-side mechanism introduced a couple of years ago and problems in trying to identify actual causes for issues

Oz outlines what he needs in order to assign an engineer to look more deeply into te problem and possible causes

OptimoMaximo describes the issue of animation transitions (see also BUG-11122), and Oz discusses server / viewer interactions in handling animations

Cathy Foil offers her observations on the issue, and Oz raises concerns over the historical perspective of the changes which resulted in Bug-11194

Further discussion on the historical perspective of Bug-11194 and on the issue of first and last animation frame transitions / blending

In terms of further investigations. and given the triple-play issue with animations has been around for several years, Vir requested at the meeting that if people could send the Lab the basic .DAE file for an affected avatar model, plus the associated animation (.ANIM or .BVH) files associated with it, or send details of older products on the Marketplace known to exhibit symptoms, this would help the Lab dig deeper into things. He emphasised that where Bento content exhibiting the problem is concerned, complete avatars model – fully textured, etc., – are not required. All that is needed is the basic model and the appropriate animations.

Bento: Opportunity to Resolve Issues

During the meeting, Oz and Vir both re-emphasised the opportunity Bento offers to potentially re-visit and fix long-standing issues with avatar animations and joint positions, with Oz particularly repeating his view that even if an issue has been around for a long time doesn’t mean it is not going to be looked at, or no attempt will be made to fix it.

However, because staff do ebb and flow with the Lab, both he and Vir indicated that what is really required, is good, solid test cases which define an issue, which offer examples of what should happen / what actually happens, where the issue(s) occur, etc., rather than anecdotal evidence or descriptions. The more practical information which can be provided on JIRAs, the better.

Oz, Vir and Coyot discuss issue reporting and resolution