Simplify Land Impact Calculation
planned
Signal Linden
Simplify the calculation of Land Impact (LI) by eliminating all relevance of object scale on LI. It is highly confusing that the size of an object impacts its LI, especially when an object switches from old to new accounting methods, suddenly becoming much more expensive LI-wise than it was previously.
Context
Second Life has two methods of calculating the LI of content:
- "Old" style
Basic prims with blinn phong materials are counted using the original LI method: 1 prim = 1 LI.
- "New" style (Mesh and Material Calculations)
Mesh content and objects with new-style PBR materials use a newer style of accounting that takes many things into account, including object scale, download size, etc.
- Download weight
- Physics weight
- Server weight
- Display weight
Solution
Let's err towards simplicity here:
- Eliminate impact of scale on LI
The goal should be that users should not be surprised when working with LI. Eliminating the scale from the calculation would mean that LI would not wildly vary based on the size of an object. This should make the accounting system much less variable and confusing.
Links
Log In
ZortyTheExplorer Resident
It was confusing for me as a new user to see 100,000 vertice objects be treated as low as 0.5 L (with 8 materials) because it was under 1m of size, but a 64m x 64m CUBE with 8 vertices could balloon so high (or a 128 vertice lo poly wall with optimized trim sheet textures woudl be 30Li).
Land impact should also consider how many materials are on an object (how many draw calls), and their texture size as well.
Efficient creators should be rewarded for their work, not punished.
Also LI "workarounds" where creators put 100,000 vertices on a lamp object for 0.5Li then enter 0 for ever other LOD should be considered a cheat, and the land impact put more on the quality of their LODs.
Frankk Parisi
This sounds an interesting idea Signal Linden AS LONG AS the change does not increase the LI of ANY currently rezzed objects. I have a 30K region which is almost full (about 1.5K li left) and if the 30K suddenly gets exceeded by a recalc of existing assets then im in trouble? Not sure what would happen - will assets disappear? be returned? which ones? That would be a total disaster - so please let us know asap what the plan is in that respect? thank you. (What i would REALLY LOVE of course is a new option to up the sim limit to 40 or 50K - and would be happy to pay a bit more for this - but thats on another Feedback request)
Beatrice Voxel
I don't think simply eliminating the scale impact would really fix things. As others have pointed out, there's other factors at play here, some of which aren't immediately obvious.
What I do think would be useful, is a prim weight "breakdown" tool, basically a dialog box or even a chat dump, that tells you what a given object's cost is, and which factors account for that. Knowing that an object's physics weight is abnormally high (esp. if it's not really used as a physics-enabled object in the first place) would be really helpful, for example.
Nexii Malthus
Beatrice Voxel you can already see the breakdown of costs via clicking More info in the build menu.
ZortyTheExplorer Resident
Beatrice Voxel Sometimes it will say the physics are high when all it is is a box, so the calculation has always been confusing.
Rohacan Hirons
Hi, I'm against this. Allowing people to create large, poorly optimized objects with a high LOD of 100k triangles and a lowest LOD of just 1 triangle is not a good idea. This kind of content is already one of the reasons why people complain about Second Life’s poor performance.
At one point, you had a good idea — using animesh LI calculation without the skeleton penalty. Why did you give up on this nice approach?
Nelson Jenkins
Rohacan Hirons Literally nothing in this proposal has anything to do with LODs. It neither solves nor exacerbates the unrelated problem you've presented.
Nexii Malthus
Rohacan Hirons the scale penalty on LI does bias people somewhat towards bad LODs, because the penalty is so aggressive and makes no sense that it leads people to try to minimise LI in the worst ways, one of which is to heavily decimate LOD. But other than that is unrelated to your problem (which is something that needs to be looked at separately)
ZortyTheExplorer Resident
Nelson Jenkins Nah it is correct to factor this in as well, as it is abused all the time to keep the LI factor low, at great expense to the inworld experience
Rosa Hexem
my hope for this has increased 10 fold now i finally am seeing the out of pocket contrarian takes from people who don't get it, which is like the hallmark of a good feature, or from people who are arguing against it but by complaining about some other thing which is, like, ok
Lucifera Morningstar
Gonna say I'm against it. Especially given that more often than not, LI changes with scale due to poorly-implemented physics, in my observation. Watched a friend resize a mesh boulder, LI jumped up, but when I rightclicked > edit > more info, it was the PHYSICS weight that jumped up by orders of magnitude.
Daniele Tatham
I'm absolutely against it.
Reason: I've studied the current model, and it has a lot of advantages!
Once you combine the right parts, you can drastically reduce the LI!
I was able to reduce the over 500 regular LI on my Linden Home, through clever combinations, to well below the maximum permitted 351 LI. Even so much that I was able to rez my sailboat if I wanted to.
Nevertheless, I agree with you: the current system is far too complicated and should be simplified considerably.
ZortyTheExplorer Resident
Daniele Tatham "Clever combinations" huh.
You mean instead of being able to upload a mesh in it's entirety, one must break it down into 40 smaller parts, each that will hopefully register as 0.5Li to 3 Li each (wall vs doorway), which is VERY TEDIOUS, and opens the product up to be used as parts when it's de-linked.
One woudl think uploading a big mesh (like a warehouse I made super optimized, and with ONE collision mesh AND trim sheet and texture atlas, should be rewarded, but no... because it was a 40m x 20m x 20m structure it bloated up to 300LI or something crazy , vs 50 Li with 'clever combinations' of separate pieces.
Big should not be penalized, especially if it uses less verts, textures and draws vs someone's 100,000 vertice lamp at 1m with 8 x 1k textures on it.
Also, it would encourage creators like me to build more things like that. Right now I dont' want to bother. it is a LOT of work to upload each piece, LOD each piece and put it together, only just to get lower Li. It never made sense to me
Tech Robonaught
ZortyTheExplorer Resident I set my Blender scale to meters, I found physics calculations most forgiving when it's a 1:1 ratio and not enlarged or reduced once in world. (yep, I've made a physics enabled mesh smaller and seen the LI go UP).
Mari Moonbeam
So will items currently rezzed in world change impact UPWARD and then we'll lose the latest rezzed item on parcel to autoreturn?
Many No copy gacha items are detailed and we've sized them down - is there a risk they get bounced if impact rises and then are lost?
I shop, not build ,so keeping my places happy are my current and future focus.
Mari Moonbeam
Please answer- if an item was 10 LI and the new owner scales it down to 7 - after this change will the item pop back to 10 LI- as the creator made it? Or stay at 7LI? What about copies in inventory that weren't adjusted?
Tech Robonaught
As a counterpoint ... cap avatar height to 6.5 feet. 8 foot avatars need bigger EVERYTHING and that uses more Land impact. Sometimes I think by design.
ZortyTheExplorer Resident
Tech Robonaught this is an argument long made by those of us who like to work with standardized and at least real world sizes as trying to calculate a 120+% rescaling to suit SL gets to be a pain.
You're right. When everything is to scale, less land use is needed. I can fit an entire nightclub in the 'living room' of a classic SL home no problem. Many clubs and coffee shops, even with live bands are actually quite small in RL.
Allegory Malaprop
Signal, I'm sure you know and this was an oversight and sorry to be pedantic but it shouldn't be spread erroneously, Blinn-Phong materials-
materials
- do in fact switch an object to the new mesh accounting system. Specular and/or custom Normals, pre PBR, do change traditional prims and sculpted prims to use mesh calculations, for good or ill. (Plan ahead accordingly on your linkset, it can even be for good, though there are of course other ways to flip accounting systems as well.)IF, and it's a huge IF, this doesn't blow up the LI on anything on the smaller side that's rezzed, yes. If it does, well, you've just wrecked a bunch of regions. It may not be great practice (I have Opinions, but let's just not), but small scale detailed rezzables have absolutely been incentivized over large scale builds when it comes to mesh, from day 1. You're also not
entirely
throwing out Download as a calculation entirely if you remove scale and respect small costs...but you pretty much are.But also, I have to laugh- "The goal should be that users should not be surprised when working with LI"- LI by scale is SO MUCH MORE PREDICTABLE than PHYSICS LI. I can make educated guessed on LI by scale and I'm usually in the ballpark. Physics LI doesn't report itself correctly in the uploader, is a complete and total mystery with Analyze, and has the wonderful??? effect of often reducing physics LI when you make an object
larger
...and I'm not even going to get into the .5m issue or DOOR PHYSICS /cry. (Now, Physics working the scale to LI way it does actually makes logical sense except
the uploader gives zero clues what is going to happen, you have to actually uploaded an object and set it to the correct Physics, at the correct scale, to know what it actually will do. And that goes double for knowing when to use Analyze. I always figured the LI scale thing was a tradeoff because things would be visible, and therefore load, at higher LODs from father away, as LOD is more or less based on the size of the object on the users screen.)Tech Robonaught
Allegory Malaprop I've not once seen PBR increase a builds land impact.
Allegory Malaprop
Tech Robonaught Link in sculpted prims, especially scaled large, for Download bloat, or tortured prims for Physics (cuts, twists, hollows, etc. - some in a box you can get away with, but try a torus. Don't do the latter especially unless you have a lot of free LI in your parcel or risk returning things). Check the More Info popup with Advanced information to see what the Physics will cost you, it'll give the cost there even before it is set to the new accounting system to use it.
Tech Robonaught
Allegory Malaprop I uploaded a bridge tipped on its side just to get around the .5 rule lol - it worked.
Load More
→