A Lab Gab ponderance ...

[Bagnu]

13 minutes ago, Love Zhaoying said:

There's a good idea, a multi-camera format that does close-ups on speakers as they talk. Like a real interview. Instead of one angle on all avatars, so you have to pay attention to which one is talking by looking for the voice indicator.

I like that idea!

I agree. A multicam recording really is the only way to get it looking good. As long as they don't break the 180 degree rule!!!

[Honey Puddles]

1 hour ago, Love Zhaoying said:

There's a good idea, a multi-camera format that does close-ups on speakers as they talk. Like a real interview. Instead of one angle on all avatars, so you have to pay attention to which one is talking by looking for the voice indicator.

I like that idea!

Even just taking that single camera view, recorded in a certain 'high rez' quality, and 'cutting' to zoomed in shots, back to the wide, back to the zoomed shot, would improve things. 

Just like something that isn't the SL equivalent of locking down a single camera on a tripod for half an hour straight.

[Love Zhaoying]

1 minute ago, Honey Puddles said:

Even just taking that single camera view, recorded in a certain 'high rez' quality, and 'cutting' to zoomed in shots, back to the wide, back to the zoomed shot, would improve things. 

Just like something that isn't the SL equivalent of locking down a single camera on a tripod for half an hour straight.

It could be done with a single "SL camera" that just switches angles, too.

[Bagnu]

It COULD be done, but LL doesn't have 2 people to record a video?

[Love Zhaoying]

10 minutes ago, Bagnu said:

It COULD be done, but LL doesn't have 2 people to record a video?

THEN, if we recorded it with 2 different logins and "edited" the two into one video, people would complain, "that's not possible in SL, they're cheating".

My same point from earlier: people are going to complain no matter what you do; just like the original post / topic in the thread.

[Ineffable Mote]

One person recording would seem more ideal, and they can run multiple avis better than most of their customers can. With a good camera hud, you probably wouldn't really need multiple avis either.

[Ayashe Ninetails]

If developers worried that much about complaints from their customers (and the many who aren't), nobody would ever do anything interesting or innovative.

Besides, most if not all of the ideas floated in this thread aren't even new and have been pretty standard forever.

Do these videos even need a new format - nope. Would it help them on YouTube - potentially. They could run a test or two and figure that out fairly quickly.

 

22 minutes ago, Bagnu said:

It COULD be done, but LL doesn't have 2 people to record a video?

They'd probably be better off contracting that out. They've already got several good video producers who they've featured on their main SL YouTube channel.

[Jaylinbridges]

6 minutes ago, Ineffable Mote said:

One person recording would seem more ideal, and they can run multiple avis better than most of their customers can. With a good camera hud, you probably wouldn't really need multiple avis either.

You mean Brad being interviewed by his Alt?  I'm confused.  And who thinks SL has peeps just standing around ready to man cameras for a lab chat?  SL brings in outside videographers if they want to make an entertainment video. (Oops cant say that, ready to they cameras?)

This Lab Chat was thrown together one day after the announcement.  Editing some action video can take days.

And finally I missed Strawberry's cats.  They were more amusing than Brad's fingers.

 

[Bubblesort Triskaidekaphobia]

1 hour ago, Love Zhaoying said:

There's a good idea, a multi-camera format that does close-ups on speakers as they talk. Like a real interview. Instead of one angle on all avatars, so you have to pay attention to which one is talking by looking for the voice indicator.

I like that idea!

I disagree.

A multicam format could work, but you would have to avoid having the RL person and the SL avatar on screen at the same time, I think.  I think it would be cool to show the SL avatar at the beginning, to show that Lab Gab is filmed in front of a live studio audience, in SL, but then cut out to the RL speakers, and stay with them from there on out.

The main reason for this is that the avatar looks silly when you put it on the screen next to a human being.  For examples of this, I don't think we have to look farther than Mark Zuckerberg, and his uncanny valley feeling interviews about Meta.  I mean, sure, the Zuckerberg avatar is ugly in any universe, and Mark is the human incarnation of the uncanny valley, but I think that even if that wasn't true, it would make a problem.  I think you could have the best looking avatar, on a screen doing a virtual interview behind the best looking actor or celebrity in the world, and it would still feel weird.  Once you adjust to seeing double, where one person has two bodies, that look different, you will notice SL looks more cartoonish when you put a RL person in it, or in front of it.  SL looks great!  SL isn't the problem.  The problem is the juxtaposition with RL.  You want to justapose it maybe at the beginning, and then cut into SL again if you are doing Q&A with the audience, to show that the question is coming from an in world person.  It needs to be done tastefully, though.  I don't see a reason to cut to the avatar being interviewed any more than you have to.

 

[Ineffable Mote]

58 minutes ago, Jaylinbridges said:

You mean Brad being interviewed by his Alt?  I'm confused.

 

I mean whoever's filming(running the screen capture software), I presume it isn't Strawberry or Oberwulf. 

Lab Gabs are a regular thing and I presume they have a bare bones team with some framework for producing informational/marketing materials, even on relatively short notice. Doesn't need large amounts of action or drama, where it's just basic corporate PR.

Little efforts are needed but they don't need to be made Herculean either, maybe that was more my point.

Edited April 29 by Ineffable Mote

[Jaylinbridges]

28 minutes ago, Ineffable Mote said:

I mean whoever's filming(running the screen capture software), I presume it isn't Strawberry or Oberwulf. 

I assumed it was Strawberry running the screen capture.  She was hired as a Linden after making hundreds of YouTube videos about Second Life.  (After LL first tried to shut down her YouTube channel for using the SL trademark.)

Brad using his Alt in an interview would be the ultimate LL cost savings. Senra Jamie from the Library with the cute free outfits, voice changer - think of the savings.  

Edited April 29 by Jaylinbridges

[Prokofy Neva]

On 4/27/2024 at 5:42 PM, Sammy Huntsman said:

Don't you mean you harassed her to get rid of it? Lol. Cmon now man, get it right. 

I was doing her a favour. I'm sure she thanked me and was unaware of how annoying it was. The end. 

[Luna Bliss]

The best solution to this problem we're discussing is for them to stand on stage using the Senra body with the Senra AO:

[Nika Talaj]

Given that the mouth movements are completely unsynchronized, there's no reason why a LabGab can't be done as just an audio interview, with Strawberry putting together the video in postproduction.  The result would be indistinguishable from a RL capture, so long as she doesn't make a bunch of cuts.  (Are you sure she isn't doing that now?) 

The "Brad" avatar could just be a clone of his actual avatar.  

 

 

[Love Zhaoying]

8 hours ago, Luna Bliss said:

The best solution to this problem we're discussing is for them to stand on stage using the Senra body with the Senra AO:

Complete with RBF? Wow!

[Jaylinbridges]

Be more interesting if they held the interviews on a SL bed.  Plenty of animations on mine (850).  I could loan him my Jamie Senra alt.  Bed Gab?

Edited April 29 by Jaylinbridges

[Paul Hexem]

Isn't that what Puppeteering was supposed to be before they realized they'd need to put in effort to get it working?

[Honey Puddles]

They're utilizing SL voice, so this could be done via some scripting.

Each person can wear an attachment that detects if certain animations are playing (voice-triggered lip movements for example). The script has a warm-up and cool-down period programmed in.. so for example.. checking 3 times every second, we check if the user is 'talking' now. If yes, we add 1 to a number. If no, we subtract 1. if the number is ever over 10, we reassign it's value as 10.

This means it will take approximately 1.5 seconds of active talking for the number to reach 5, and it will take 1.5 seconds for it to drop to 5.  Now it's just a matter of sending 'llSay(111, "talking"); and llSay(111, "stopped"); Person 2 may use 222, person 3, 333, and so on.

Second script on the recording person, has pre-positioned camera positions and angles for each person (set up before recording). The camera script listens for 'talking' messages, and keeps track then of who is talking, and whether it should cut to a specific person's camera, or to a wide shot. SL even has camera smoothing functions that could potentially make these transitions fairly smooth.

I'm not saying it doesn't take some competent scripting to make something like this work, but it's not exactly a networked vendor system or something along those lines.

 

Edited April 29 by Honey Puddles

[Love Zhaoying]

1 minute ago, Honey Puddles said:

They're utilizing SL voice, so this could be done via some scripting.

Each person can wear an attachment that detects if certain animations are playing (voice-triggered lip movements for example). The script has a warm-up and cool-down period programmed in.. so for example.. checking 3 times every second, we check if the user is 'talking' now. If yes, we add 1 to a number. If no, we subtract 1. if the number is ever over 10, we reassign it's value as 10.

This means it will take approximately 1.5 seconds of active talking for the number to reach 5, and it will take 1.5 seconds for it to drop to 5.  Now it's just a matter of sending 'llSay(111, "talking"); and llSay(111, "stopped"); Person 2 may use 222, person 3, 333, and so on.

Second script on the recording person, has pre-positioned camera positions and angles for each person (set up before recording). The camera script listens for 'talking' messages, and keeps track then of who is talking, and whether it should cut to a specific person's camera, or to a wide shot.

I'm not saying it doesn't take some competent scripting to make something like this work, but it's not exactly a networked vendor system or something along those lines.

 

Even without scripting, I was thinking about leveraging something like camera presets:

1) Camera Preset #1 - for wide shot of group

2) Camera Preset #2 - for closeup of speaker #1

3) Camera Preset #3 - for closeup of speaker #2

Whoever is creating the video would only need to switch between camera presets.

As far as "talking" goes, aren't there "good" talking animations that trigger when they see someone is talking? I didn't notice if the video had talking animations on "at all" (I assume not).

 

[Honey Puddles]

3 minutes ago, Love Zhaoying said:

Even without scripting, I was thinking about leveraging something like camera presets:

1) Camera Preset #1 - for wide shot of group

2) Camera Preset #2 - for closeup of speaker #1

3) Camera Preset #3 - for closeup of speaker #2

Whoever is creating the video would only need to switch between camera presets.

As far as "talking" goes, aren't there "good" talking animations that trigger when they see someone is talking? I didn't notice if the video had talking animations on "at all" (I assume not).

 

That's the one element I'm not sure on. As far as I recall, there are 'voice gestures' that exist, which means there are some 'is speaking' hooks to detect at least via gesture. I'm fishing around on the LSL portal right now to see if I can find anything better, if not, I'm making a feedback feature request.

 

[Love Zhaoying]

13 minutes ago, Honey Puddles said:

That's the one element I'm not sure on. As far as I recall, there are 'voice gestures' that exist, which means there are some 'is speaking' hooks to detect at least via gesture. I'm fishing around on the LSL portal right now to see if I can find anything better, if not, I'm making a feedback feature request.

 

I see now, llGetAgentInf() has nothing for "voicing":  https://wiki.secondlife.com/wiki/LlGetAgentInfo

I assumed there was SOME support but could be wrong, because of course the viewer has no idea if you are "talking" or if there is just background noise, music, etc. on your audio. (Could have sworn I remembered something.)

Edited April 29 by Love Zhaoying

[Love Zhaoying]

I start to wonder what the original inspiration for the thread was, if you will, the "preponderance".

 

[Honey Puddles]

4 minutes ago, Love Zhaoying said:

I see now, llGetAgentInf() has nothing for "voicing":  https://wiki.secondlife.com/wiki/LlGetAgentInfo

I assumed there was SOME support but could be wrong, because of course the viewer has no idea if you are "talking" or if there is just background noise, music, etc. on your audio. (Could have sworn I remembered something.)

There are existing gesture triggers for /voicelevel1, /voicelevel2, and /voicelevel3, but there don't seem to be corresponding triggers for LSL to detect. There are gestures in the library under "Gestures > Speech Gestures", but a casual search on the wiki suggests they may not be working, and of course these triggers/gesture/animations predate fitted mesh and/or bento.

A feedback feature request definitely seems to be warranted.

https://feedback.secondlife.com/scripting-features/p/add-agent-speaking-to-llgetagentinfo-to-detect-whether-avatar-is-is-not-speaking

[Rowan Amore]

Most mesh heads have the option for Talking when you use local chat.  All one would need to do is trigger that then delete it from local when done talking.  Might require some synchronization for the participants but it should be easily doable.

Edited April 29 by Rowan Amore

[Honey Puddles]

Added that to the feature request! I'd forgotten about animated mesh heads with non-bento bones, animating mouths for typing, not necessarily working for talking.