Yay! We have a new MetaReality podcast out today: Aaaaaaand We’re Back!
My timeline of, summary of, and comments on the podcast follow. As usual this is not a transcript. I provide time marks for the things I see as subject change points or interesting entry points, so you can check out what interests you.
If you decide you should yell about something, be sure you check out the audio first. The subject content I hear and you read from my writing may have no relationship to what was actually said.
Metareality Podcasts
01:15 MetaReality podcasts are loved by a good number of SL users. As evidenced by Gianna getting all sorts of questions via various media about whether she was going to be doing more podcasts. Gianna is committed to continuing the podcasts. This recent break was due to RL busyness. So, we will see more podcasts coming. Yay!
2:00 Material System – Discussion about this new feature starts.
The addition of Normal and Specular maps has the possibility of improving the appearance and performance of Second Life™. Gianna is excited that creators can add more apparent detail with fewer polygons.
Karl explains that Specular Maps control how shiny an object is at various places. How light reflects leaving a shiny spot gives us visual clues to the texture of an object. So glass is shiny with sharp bright highlights and tree bark isn’t and has no highlights.
Karl sees the big win in materials as being in the Normal Maps, which are a technique that makes things look like they have more shape with fewer polygons. So a simple cube could be made to look like a stone wall. Shadows on the surface of a flat cube surface would show the effects of a 3D surface of stones. Another example is flat polygons in clothes could be made to have buttons that stand out from the surface without additional polygons.
The question that many are asking or making statements about is, “How does it affect performance?”
Gianna is frustrated with those claiming it will cause more lag. Like anything in SL or 3D virtual worlds it can help or hurt. But, if done right it has the possibility to reduce lag as fewer polygons have to be rendered for the additional detail shown.
There is lots of confusion as to what is and is not affecting performance. In many cases poorly made mesh clothes are placing a heavy graphics load on the viewer. This is why a feature is built into the viewer to allow users to reduce the render load by limiting the complexity of mesh objects being rendered. If an object is too complex, the viewer won’t render it. You can change that limit to suit your computer and personal preferences. But, most are unaware of that and simply blame mesh as the problem.
Karl thinks Specular Maps do not have much of a possibility to help or hurt performance, while Normal Maps do. If one considers the affect of adding detail with or without Normal Maps, Normal Maps are the more efficient choice. The Normal Maps have a greater possibility to improve performance than hurt it.
Karl also pointed out that Normal and Specular Maps are 10+ year old technology. The Wikipedia mentions the first use of ‘Displacement Maps’, another name for Bump Maps, was first introduced at SIGGRAPH in 1996. Normal Maps are a bit more complex than Bump Maps and came out a couple of years later. Look it up if you are that interested in the difference. Karl’s and my point is that this is old technology just arriving in SL.
So… basically, unless you are running a Commador-64 your graphics card is designed to use these maps very efficiently. The render of the Normal and Specular Maps is something that happens almost exclusively in the graphics card. My GPU (GTX560Ti) is only using about 20% of its capability now. My older 8800 only used a small part of its ability too. So, there is lots of room to add this new processing and see little if any change in performance.
Other factors are currently affecting our OpenGL render speeds. OpenGL has been proven to be as fast or faster than DirectX. But, the SL Viewer does not use all the newest features of OpenGL so it can be compatible with older computer systems.
09:00 – Karl points to the clothing meshes we are seeing now as insane.
To get wrinkles and other effects in clothes some are upping the poly count to ridiculous levels. Normal Maps can solve this problem and reduce poly counts. But, these high poly count clothes are giving a bad rep to mesh with some people; I think usually those that don’t understand mesh.