LoD is Level of Detail. It is a term used in 3D modeling to improve performance by reducing the amount of data that has to be rendered. In Second Life™ it seems to be poorly understood and ignored by many. The result is a poorly performing Second Life. This video shows the major problems.
I did not intend to belittle anyone with the video. But, having a ton of poorly rendering mash in a region gave me the extreme example I needed.
Some of Mad Pea’s builds are absolutely gorgeous. But, with a little bit more work they would not have had the LoD failings I was seeing. It may also be that they intended to have the various victims’ business cards be hard to find and they used LoD to make it even harder. I doubt it, but may be…
Fixing the Problem
There are those that will immediately conclude that the Lab should do something to make designers design better mesh. I suspect few of those people realize their reaction tends to reveal the fascist/dictatorial side of their personality. Plus there is the entitlement generation’s attitude of having someone do whatever needs fixing for them.
There is a free market solution. Just as we want labeling and information in RL, so we can make informed decisions, the same ‘informed solution’ is possible in Second Life. We just need some easy way to know how well a mesh is made and get such information in front of new users.
Of course the complication is in the ambiguous idea of what defines ‘well made’. The rule oriented immediately go to work making rules that are supposed to define ‘well made’. But, rules are the bureaucrats dream that stifle innovation, creativity, and our freedom to chose and decide for ourselves.
How does one balance beauty verses computer efficiency? How does one set a beauty verses efficiency ratio that everyone will like? Via rules and dictatorial imposition of standards there is no way to satisfy everyone. So, some minority gets screwed so others can have what they think they want. If we are lucky, that is where the damage stops.
When we simply post the facts about a product and allow people to decide, we get amazingly good results. Think Consumer Reports. People are willing to pay for objective information in RL. That supports someone doing the research needed to determine the quality, usability, and durability of goods in RL. We have that to some extent in Second Life. But, most new people are not going to know how to find that information. Many long term residents don’t know how to find that information and most probably don’t know it is available. Plus we have the challenge of figuring out the reviewer’s biases and agendas and learning where they set their balance points and how well they agree, or not, with ours.
Fortunately computers are very good at math. We already have our viewers calculating Avatar Render Cost. The computer can objectively and unemotionally report how much work it has to do to render something. Also, the viewer can be extremely consistent in its reporting. But, this reporting is hidden away in the menu structure of our viewers.
It shouldn’t be too hard to include some measure for deciding on whether an object has made reasonable use of LoD. There may be some easy way for the computer to decide if an item is keeping it shape or not. But, I suspect deciding shape consistency between LoD’s is rather complicated. It might be worthwhile. But, I suspect a simple comparison of polygon counts between LoD’s might suffice.
What It Might Look Like
If the viewer were to calculate Render Cost and LoD quality along with something on texture use and display the information whenever we try on a new mesh attachment, demo or purchased item, wouldn’t that get this issue in everyone’s face? I think then as users decided what level of quality they were willing to deal with for the appearance of a specific item we would see performance in Second Life improve.
Those designers that know how to use normal maps to get additional detail while reducing polygon count and use LoD to maintain the shape of things as distance increases, would have an advantage. Those that don’t would have an incentive from market pressure to learn. Plus those that do reviews would have an easy way to report the render quality of items in a consistent and objective way. Since the Lab would not be imposing restrictions we would be free to make or buy whatever we liked regardless of anyone else’s idea of beauty verses render cost.
If you think this sounds like a direction to go, click over to the JIRA and click WATCH on JIRA Feature Request: BUG-7928 – Information on mesh render cost. If you have other ideas, add them in the comments section below.