Skip to content
This repository has been archived by the owner on Mar 28, 2024. It is now read-only.

[BUG-139203] [Animesh] Feature Request: New land impact formula that references tri counts. An alternative to the tri count cap. Also, a reference for tri counts with Animesh attachments. #2416

Open
sl-service-account opened this issue Oct 11, 2017 · 4 comments

Comments

@sl-service-account
Copy link

Please do not triage until October 25th or later. Please read in full.

Information

Recently it was decided a tri count cap would be the primary limiting factor that decides whether or not a skeleton can be applied to a linkset.
The problem with a sole tri count restriction approach is that it focuses on content creators optimizing their content or creating entirely new content explicitly for the feature and deters users from using existing third party content to assemble their own Animesh characters.
The problem with existing content is that most of it isn't optimized because it was never designed to be rezzed, but instead, attached where the only penalty was high avatar complexity, something many tolerate as far as their own looks or the looks of their friends are concerned.
Still, nothing was done to actually block people from wearing high tri count content, only block rendering of the wearer if the complexity value exceeded a threshold. Even with low avatar complexity, content can still have high LOD tri counts.
The majority of users in Secondlife are not 3D modelers nor programmers. They have to rely on third party content to assemble their avatars, regions and experiences.
The allure and interest in Animesh from the beginning has been that users would be able to turn their already purchased mod, rigged mesh into animated content, not wait for creators to offer explicit Animesh products.
Animesh was supposed to be a boon for rigged mesh avatar, hair, clothing and accessory creators who already have products available. Users would be purchasing content they never would have worn them self just to assemble Animesh characters of their own.
As long as low tri counts are the primary limiter, Animesh will unnecessarily suffer the same fate as features such as Pathfinding & Experiences where only a limited few will actually be able to make use of it.

At the last Content Creators meeting, Medhue Simoni asked if there was any way land impact could be tied to triangle count. It was determined that this could be a possibility. https://www.youtube.com/watch?v=P9j5w-5nnZg&t=47m20s
Taking this suggestion into account, I looked at many popular avatars of varying tri counts and land impact and compiled a simple, yet effective formula that aims to create a land impact for Animesh characters that rewards optimization and heavily penalizes "exotic" tri counts.
First though, the primary tri count cap is still necessary, but needs to be raised to 500k. This value will allow all existing avatars in Secondlife a chance to be used as Animesh characters as well as allow them to be accessorized with existing content.

The new land impact formula for Animesh is as follows:

Animesh Land Impact = Max(base land impact value, actual object land impact, Ktri count * modifier value)

  • base land impact value - It was deemed that adding a skeleton to the region should have some kind of attributed cost. Recommended values have been anywhere from 16 to 128.

  • actual land impact - This is the natural land impact of the linkset which is primarily influenced by LOD level tri counts and can fluctuate on a wide spectrum depending on the scale, materials, prim torture, physics shape/status, etc.

  • Ktri count * modifier value - This is where we multiply tri counts against a modifier value. This is a final gatekeeper if natural land impact is less than base land impact such as by using minimal tri counts for Med, Low, Lowest LOD levels or by scaling rigged links to 0.01 for all axii.

    Below is a list of popular mod, rigged mesh avatars, their land impacts, tri counts and what their Animesh land impact values would be with this formula applied with tri count modifiers ranging from 1 to 5 and base land impact of 32.

    |Avatar|Natural Land Impact|Ktri Count|Mod A (Ktri's x1)|Mod B (Ktri's x2)|Mod C (Ktri's x3)|Mod D (Ktri's x4)|Mod E (Ktri's x5)|
    |
    |-|-|-|-|-|-|-|-|-|
    |Niramyth Wolfkin (large body w/ fur) v1.5|257|381|381|762|1143|1524|1905|
    |
    |All pieces scaled to <0.01,0.01,0.01>|105|381|381|762|1143|1524|1905|
    |
    |Niramyth Wolfkin (large body only) v1.5|223|50|223|223|223|223|250|
    |
    |Niramyth Aesthetic (w/ bento head & hands) v1.9|242|78|242|242|242|312|390|
    |
    |All pieces scaled to <0.01,0.01,0.01>|17|78|78|156|234|312|390|
    |
    |DreamCrawler Feral Dragon v1.3|1691|51.4|1691|1691|1691|1691|1691|
    |
    |Feral Dragon Addons (Fins) v1.3|485|21.4|485|485|485|485|485|
    |
    |DreamCrawler Next Gen Fox (large body)(v3.4)|364|41.9|364|364|364|364|364|
    |
    |DreamCrawler Next Gen Fox (med body)(v3.4)|278|41.9|278|278|278|278|278|
    |
    |Red Remora - Fennec Fox Avatar(body, ears, tail, wings)|127|17.5|127|127|127|127|127|
    |
    |Red Remora - Fox Avatar(body, tattoo layer body, ears, tail, whiskers)|544|68.9|544|544|544|544|544|
    |
    |Breeder's Choice Quarter Horse(body only) v2.1|141|61.8|141|141|185.4|247.2|309|
    |
    |Teegle Horse body|?|39|39*|78*|117*|156*|195*|
    |
    |Bakemonoya Large Tiger|201|26|201|201|201|201|201|
    |
    |TWI Red Fox Body (head, body, legs, tail)|?|113|113*|226*|339*|452*|565*|
    |
    |TWI Timber Wolf (heads, body, tattoo layer body, legs, tails) v1.11 (non-bento)|454|76|454|454|454|454|454|
    |
    |TWI Lion (head, body, legs, tail)|?|130|130*|260*|390*|520*|650*|
    |
    |Bird-E-Mart Eagle (one size & style)|28|14|32|32|42|56|70|
    |
    |Utilizator Avatar 2.0 body|40|26|40|52|78|104|130|
    |
    |Utilizator Kemono body|54|14.4|54|54|54|57.6|72|
    |
    |Blue Galaxy Solarian Female|12|19|32|38|57|76|95|
    |
    |Orange Nova Dvalinn Deer body|47|25|47|50|75|100|125|
    |
    |Orange Nova Slugger Shark|23|16|32|32|48|64|80|
    |
    |piggu Jonathon (head & body)|41|34.4|41|68.8|103.2|137.6|172|
    |
    |Lucia Nightfire's catgirl avatar(AV 2.0 body, tail, hair, clothing)|103|100|103|200|300|400|500|
    |
    |All pieces scaled to <0.01,0.01,0.01>|40|100|100|200|300|400|500|



    Values w/ asterisk = Natural land impact values not available. Tri count values used instead.



    What we see is that low quality or optimized avatars still show ideal land impacts while land impacts of high LOD tri count avatars will yield much higher values even if their natural land impact is near or below base impact values.

    If other factors such as scaling, materials, prim torture, physics cause natural land impact to exceed the tri count modified value, it will be used instead.



    Making land impact the driving factor instead of limiting creation solely based on tri counts will allow any kind of Animesh characters to be created or to be assembled with existing content as long as the owner is willing to bear the cost of land impact consumed.

    With the recent 50% increase in region prims, Animesh land impacts are still manageable.



    # Other Information



    But what about high tri counts in regards to Animesh attachments?

    There should be a tri count limiter with attaching as well, not just a sole Animesh attachment count limit. A 100k tri count limit seems ideal but only if all of the wearer's attachment's tri counts count toward the cap.

    If a user attempts to attach an Animesh character whose tri count value plus the total tri counts of all the wearer's attachments exceeds 100k, the Animesh character cannot be attached. This means a user may not be even able to attach a single Animesh character depending what they're already wearing.

    So with Animesh attachment count as the primary limiter and a overall wear limit of 100k tri's as the secondary limiter, we have a better means of restricting what can be worn not just how many.



    # Closing



    Low tri count caps will ultimately discard most existing mod/rigged mesh avatars, hair, clothing and accessories from qualifying for use with Animesh, rendering an already huge market of content useless and destroying the multiple commerce type potential the feature initially had.

    Even if a different formula/calculation is used, we should favor potential high land impacts versus definite low tri count caps. We should block attaching Animesh not just by count, but also based on total tri counts of what is already worn.



    Thanks in advance for any consideration.



    # References



    https://marketplace.secondlife.com/p/The-Wolfkin-Full-Body-Mesh-Wolfman-Avatar/4817944

    https://marketplace.secondlife.com/p/NIRAMYTH-AESTHETIC-SMITH-PACKAGE-10/7037604

    https://marketplace.secondlife.com/p/DreamCrawler-Feral-Dragon-Starter-Pack-Voluptas/6574592

    https://marketplace.secondlife.com/p/DC-NG-WereWolf-Series-WereFox-Starter-Pack-Rubeus/12364100

    https://marketplace.secondlife.com/p/Red-Remora-Fennec-Fox-Avatar/10382652

    https://marketplace.secondlife.com/p/Red-Remora-Fox-Avatar/7562739

    https://marketplace.secondlife.com/p/BC-Quarter-Horse-Avatar-21/4138999

    https://marketplace.secondlife.com/p/Tiger-Avatar-White-Lsize/3213370

    https://marketplace.secondlife.com/p/TWI-Timber-Wolf-Avatar/5527314

    https://marketplace.secondlife.com/p/BRDMRT-Eagle-boxed/7574669

    https://marketplace.secondlife.com/p/UTILIZATOR-Avatar-20-beta/2861916

    https://marketplace.secondlife.com/p/UTILIZATOR-Kemono/6070005

    https://marketplace.secondlife.com/p/Solarian-Feline-Avatar/7456366

    https://marketplace.secondlife.com/p/piggu-Jonathan-Avatar-20/11447858

    https://marketplace.secondlife.com/p/Orange-Nova-Dvalinn-Deer/9685779

    https://marketplace.secondlife.com/p/Orange-Nova-Slugger-Shark-V20/8268189|

Original Jira Fields
Field Value
Issue BUG-139203
Summary [Animesh] Feature Request: New land impact formula that references tri counts. An alternative to the tri count cap. Also, a reference for tri counts with Animesh attachments.
Type New Feature Request
Priority Unset
Status Needs More Info
Resolution Unresolved
Reporter Lucia Nightfire (lucia.nightfire)
Created at 2017-10-11T05:03:44Z
Updated at 2017-12-01T20:48:22Z
{
  'Business Unit': ['Platform'],
  'Date of First Response': '2017-10-11T07:13:35.465-0500',
  'How would you like the feature to work?': 'Please do not triage until October 25th or later. Please read in full.\r\n\r\nh1. Information\r\n\r\nRecently it was decided a tri count cap would be the primary limiting factor that decides whether or not a skeleton can be applied to a linkset.\r\nThe problem with a sole tri count restriction approach is that it focuses on content creators optimizing their content or creating entirely new content explicitly for the feature and deters users from using existing third party content to assemble their own Animesh characters.\r\nThe problem with existing content is that most of it isn\'t optimized because it was never designed to be rezzed, but instead, attached where the only penalty was high avatar complexity, something many tolerate as far as their own looks or the looks of their friends are concerned.\r\nStill, nothing was done to actually block people from wearing high tri count content, only block rendering of the wearer if the complexity value exceeded a threshold. Even with low avatar complexity, content can still have high LOD tri counts.\r\nThe majority of users in Secondlife are not 3D modelers nor programmers. They have to rely on third party content to assemble their avatars, regions and experiences.\r\nThe allure and interest in Animesh from the beginning has been that users would be able to turn their already purchased mod, rigged mesh into animated content, not wait for creators to offer explicit Animesh products.\r\nAnimesh was supposed to be a boon for rigged mesh avatar, hair, clothing and accessory creators who already have products available. Users would be purchasing content they never would have worn them self just to assemble Animesh characters of their own.\r\nAs long as low tri counts are the primary limiter, Animesh will unnecessarily suffer the same fate as features such as Pathfinding & Experiences where only a limited few will actually be able to make use of it.\r\n\r\nAt the last Content Creators meeting, Medhue Simoni asked if there was any way land impact could be tied to triangle count. It was determined that this could be a possibility. https://www.youtube.com/watch?v=P9j5w-5nnZg&t=47m20s\r\nTaking this suggestion into account, I looked at many popular avatars of varying tri counts and land impact and compiled a simple, yet effective formula that aims to create a land impact for Animesh characters that rewards optimization and heavily penalizes "exotic" tri counts.\r\nFirst though, the primary tri count cap is still necessary, but needs to be raised to 500k. This value will allow all existing avatars in Secondlife a chance to be used as Animesh characters as well as allow them to be accessorized with existing content.\r\n\r\nThe new land impact formula for Animesh is as follows: \r\n\r\nAnimesh Land Impact = Max(base land impact value, actual object land impact, Ktri count * modifier value)\r\n\r\n- base land impact value - It was deemed that adding a skeleton to the region should have some kind of attributed cost. Recommended values have been anywhere from 16 to 128.\r\n- actual land impact - This is the natural land impact of the linkset which is primarily influenced by LOD level tri counts and can fluctuate on a wide spectrum depending on the scale, materials, prim torture, physics shape/status, etc.\r\n- Ktri count * modifier value - This is where we multiply tri counts against a modifier value. This is a final gatekeeper if natural land impact is less than base land impact such as by using minimal tri counts for Med, Low, Lowest LOD levels or by scaling rigged links to 0.01 for all axii.\r\n\r\nBelow is a list of popular mod, rigged mesh avatars, their land impacts, tri counts and what their Animesh land impact values would be with this formula applied with tri count modifiers ranging from 1 to 5 and base land impact of 32.\r\n\r\n||Avatar||Natural Land Impact||Ktri Count||Mod A (Ktri\'s x1)||Mod B (Ktri\'s x2)||Mod C (Ktri\'s x3)||Mod D (Ktri\'s x4)||Mod E (Ktri\'s x5)||\r\n|Niramyth Wolfkin (body w/ fur) v1.5|257|381|381|762|1143|1524|1905|\r\n|Niramyth Wolfkin (body only) v1.5|223|50|223|223|223|223|250|\r\n|All pieces scaled to <0.01,0.01,0.01>|105|381|381|762|1143|1524|1905|\r\n|Niramyth Aesthetic (w/ bento head & hands) v1.9|242|78|242|242|242|312|390|\r\n|All pieces scaled to <0.01,0.01,0.01>|17|78|78|156|234|312|390|\r\n|DreamCrawler Feral Dragon v1.3|1691|51.4|1691|1691|1691|1691|1691|\r\n|Feral Dragon Addons (Fins) v1.3|485|21.4|485|485|485|485|485|\r\n|DreamCrawler Next Gen Fox (large body)(v3.4)|364|41.9|364|364|364|364|364|\r\n|DreamCrawler Next Gen Fox (med body)(v3.4)|278|41.9|278|278|278|278|278|\r\n|Red Remora - Fennec Fox Avatar(body, ears, tail, wings)|127|17.5|127|127|127|127|127|\r\n|Red Remora - Fox Avatar(body, tattoo layer body, ears, tail, whiskers)|544|68.9|544|544|544|544|544|\r\n|Breeder\'s Choice Quarter Horse(body only) v2.1|141|61.8|141|141|185.4|247.2|309|\r\n|Teegle Horse body|?|39|39*|78*|117*|156*|195*|\r\n|Bakemonoya Large Tiger|201|26|201|201|201|201|201|\r\n|TWI Red Fox Body (head, body, legs, tail)|?|113|113*|226*|339*|452*|565*|\r\n|TWI Timber Wolf (heads, body, tattoo layer body, legs, tails) v1.11 (non-bento)|454|76|454|454|454|454|454|\r\n|TWI Lion (head, body, legs, tail)|?|130|130*|260*|390*|520*|650*|\r\n|Bird-E-Mart Eagle (one size & style)|28|14|32|32|42|56|70|\r\n|Utilizator Avatar 2.0 body|40|26|40|52|78|104|130|\r\n|Utilizator Kemono body|54|14.4|54|54|54|57.6|72|\r\n|Blue Galaxy Solarian Female|12|19|32|38|57|76|95|\r\n|Orange Nova Dvalinn Deer body|47|25|47|50|75|100|125|\r\n|Orange Nova Slugger Shark|23|16|32|32|48|64|80|\r\n|piggu Jonathon (head & body)|41|34.4|41|68.8|103.2|137.6|172|\r\n|Lucia Nightfire\'s catgirl avatar(AV 2.0 body, tail, hair, clothing)|103|100|103|200|300|400|500|\r\n|All pieces scaled to <0.01,0.01,0.01>|40|100|100|200|300|400|500|\r\n\r\n* = Natural land impact values not available. Tri count values used instead.\r\n\r\nWhat we see is that low quality or optimized avatars still show ideal land impacts while land impacts of high LOD tri count avatars will yield much higher values even if their natural land impact is near or below base impact values.\r\nIf other factors such as scaling, materials, prim torture, physics cause natural land impact to exceed the tri count modified value, it will be used instead.\r\n\r\nMaking land impact the driving factor instead of limiting creation solely based on tri counts will allow any kind of Animesh characters to be created or to be assembled with existing content as long as the owner is willing to bear the cost of land impact consumed.\r\nWith the recent 50% increase in region prims, Animesh land impacts are still manageable.\r\n\r\nh1. Other Information\r\n\r\nBut what about high tri counts in regards to Animesh attachments?\r\nThere should be a tri count limiter with attaching as well, not just a sole Animesh attachment count limit. The 500k tri count limit comes into play here too.\r\nIf a user attempts to attach an Animesh character whose tri count value plus the total tri counts of all the wearer\'s attachments exceeds 500k, the Animesh character cannot be attached. This means a user may not be even able to attach a single Animesh character depending what they\'re already wearing.\r\nSo with Animesh attachment count as the primary limiter and a overall wear limit of 500k tri\'s as the secondary limiter, we have a better means of restricting what can be worn not just how many.\r\n\r\nh1. Closing\r\n\r\nLow tri count caps will ultimately discard most existing mod/rigged mesh avatars, hair, clothing and accessories from qualifying for use with Animesh, rendering an already huge market of content useless and destroying the multiple commerce type potential the feature initially had.\r\nEven if a different formula/calculation is used, we should favor potential high land impacts versus definite low tri count caps. We should block attaching Animesh not just by count, but also based on total tri counts of what is already worn.\r\n\r\nThanks in advance for any consideration.\r\n\r\nh1. References\r\n\r\nhttps://marketplace.secondlife.com/p/The-Wolfkin-Full-Body-Mesh-Wolfman-Avatar/4817944\r\nhttps://marketplace.secondlife.com/p/NIRAMYTH-AESTHETIC-SMITH-PACKAGE-10/7037604\r\nhttps://marketplace.secondlife.com/p/DreamCrawler-Feral-Dragon-Starter-Pack-Voluptas/6574592\r\nhttps://marketplace.secondlife.com/p/DC-NG-WereWolf-Series-WereFox-Starter-Pack-Rubeus/12364100\r\nhttps://marketplace.secondlife.com/p/Red-Remora-Fennec-Fox-Avatar/10382652\r\nhttps://marketplace.secondlife.com/p/Red-Remora-Fox-Avatar/7562739\r\nhttps://marketplace.secondlife.com/p/BC-Quarter-Horse-Avatar-21/4138999\r\nhttps://marketplace.secondlife.com/p/Tiger-Avatar-White-Lsize/3213370\r\nhttps://marketplace.secondlife.com/p/TWI-Timber-Wolf-Avatar/5527314\r\nhttps://marketplace.secondlife.com/p/BRDMRT-Eagle-boxed/7574669\r\nhttps://marketplace.secondlife.com/p/UTILIZATOR-Avatar-20-beta/2861916\r\nhttps://marketplace.secondlife.com/p/UTILIZATOR-Kemono/6070005\r\nhttps://marketplace.secondlife.com/p/Solarian-Feline-Avatar/7456366\r\nhttps://marketplace.secondlife.com/p/piggu-Jonathan-Avatar-20/11447858\r\nhttps://marketplace.secondlife.com/p/Orange-Nova-Dvalinn-Deer/9685779\r\nhttps://marketplace.secondlife.com/p/Orange-Nova-Slugger-Shark-V20/8268189',
  'ReOpened Count': 0.0,
  'Severity': 'Unset',
  'Target Viewer Version': 'viewer-development',
  'Why is this feature important to you? How would it benefit the community?': '?',
}
@sl-service-account
Copy link
Author

Beq Janus commented at 2017-10-11T12:13:35Z

Is this a penalty LI calculation for Animesh or a suggestion for the replacement of the current LI calculation across all content?

The existing LI formula already has an element of Tri count included in the streaming cost (albeit a proportional cost based on the gzipped vertex data); it also takes into account the scale to cover the various visibility scenarios relating to the displayed LOD which is also potentially important when considering the animation.

@sl-service-account
Copy link
Author

Vir Linden commented at 2017-10-11T18:24:29Z

Flagged as need more info to allow for commenting.

@sl-service-account
Copy link
Author

polysail commented at 2017-10-26T21:00:01Z

The use case of an avatar coming along and wearing an entire second avatar needs to be accounted for. Adding complexity + (multiplier) would be needed.

@sl-service-account
Copy link
Author

Lucia Nightfire commented at 2017-12-01T20:48:22Z

Reduced the initially proposed attachment tri count cap(all attachments plus the animesh character attempted to be worn) from 500k to 100k.

As far as rezzed content is concerned, as long as tri counts yield land impact, optimized content will still yield ideal land impact while unoptimized content will have bloated land impact.

As far as a complexity plus mulitplier, I'm not sure one is needed as complexity is already updated with animesh mesh + skeleton values.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Projects
None yet
Development

No branches or pull requests

1 participant