Jump to content

Reducing ARC (complexity), losslessly


Selc
 Share

You are about to reply to a thread that has been inactive for 2209 days.

Please take a moment to consider if this thread is worth bumping.

Recommended Posts

Might anyone know of a lossless method for reducing the Render Weight/Avatar Complexity of a mesh .dae file exponentially?

I've discovered a wildly unorthodox, but 100% lossless method for reducing any .dae file to 1/4th of its original complexity. But I can't get it down to 1/10th or 1/20th or 1/30th.

Again, I'm looking for a lossless method, so it can't require decimating, etc.

Link to comment
Share on other sites

Later edit: This response was in part to a post that was later deleted :D

If you have a fairly heavy mesh to start out with like many made with MD  then getting rid of any extra edgeloops is a good starting place -- even for the highest LOD setting.

Remember that not everyone has Advanced Lighting on so will not see normal maps (hence relying on them is not the best plan IMHO).

Be sure and test your lower LODs at a setting less than 4 (such as 2 or even 1). I have found some hairs lately that completely disappear at short range when viewed with LOD 2 -- not a good plan.

Let textures fill in some details.

 

And yes decimate is an awful function in my book. I have never used it. Keep things simple if you can and go from there.

 

 

Link to comment
Share on other sites


Selc wrote:

Might anyone know of a lossless method for reducing the Render Weight/Avatar Complexity of a mesh .dae file exponentially?

When it comes to ARC textures may add far more than the mesh itself so you want to be careful with how many and how high res textures you use. But I suppose you knew that already.

Link to comment
Share on other sites

Spent much of yesterday reinforcing my appreciation of Blender's Alt Merge function. It's well nifty.

Am curious to know what the OP needs to reduce that much and losslessly. dae files not intended for SL use? A creator who's found his or herself on the wrong side of the complexity numbers?

And of course this super ninja trick would be brilliant to learn about. Every little helps.

Link to comment
Share on other sites

  • 2 weeks later...


Bitsy Buccaneer wrote:

Spent much of yesterday reinforcing my appreciation of Blender's Alt Merge function. It's well nifty.

Am curious to know what the OP needs to reduce that much and losslessly. dae files not intended for SL use? A creator who's found his or herself on the wrong side of the complexity numbers?

And of course this super ninja trick would be brilliant to learn about. Every little helps.

Alt Merge is not lossless for planar hair.

I'm looking for the same secret technique that CATWA is using for their mesh heads. CATWA Jessica, for example, has 30+ mesh heads linked together during any given frame (most of them invisible other than the current frame) — yet the total complexity is only around 5,000 when worn.

How do they bring something that should be 300,000+ complexity down to 5,000 complexity?

Link to comment
Share on other sites

It's still a nifty trick, which is all I really said. Sorry if it isn't any help to you. What's so special about planar hair that Alt Merge doesn't work? Or is it just insufficiently lossless for your demands? At which point, it only makes sense to ask if any technique is.

I'm sure I'm not the only one who is curious to hear what you did to achieve a 100% lossless 3/4's reduction. If it is wildly unorthdox it may well contribute to the collective knowledge. Will you share that or just ask others to give you their prize techniques?

Link to comment
Share on other sites


Bitsy Buccaneer wrote:

I'm sure I'm not the only one who is curious to hear what you did to achieve a 100% lossless 3/4's reduction.

Judging by this post she didn't achieve it, she's just seen it somewhere else.

I may have found the answer and if I am right, it's not an actual reduction of render cost but a way to cheat the system into giving too low figures:

Calculated render cost depends on an object's size since the different LoD models are weighed according to an estimate how often they are displayed. A fitted mesh has two different sizes of course, one nominal - the one the item is uploaded at and you get if you rez it - and one actual - the size it has when streched over an avatar. The render cost is based on the nominal size while the LoD and switch points seem to be based on the actual size. That means you can cheat a lot simply by making the fitted mesh object very tiny and let the rigging scale it up to the size it needs to be.

If I'm right  - and the few quick tests I did seem to indicate that - we have a serious flaw in the whole ARC system, serious enough we have to ask if ARC means anything at all when fitted mesh is involved.

Link to comment
Share on other sites


Bitsy Buccaneer wrote:

So you're getting different render weights when you rez fit mesh items and when you wear them?

As far as I can see, you get exactly the same render weight regardless of actual size even though a smaller object is supposed to be less render heavy than a bigger one.

The only explanation to this is that render weight for fitted mesh is calculated from the size it had when it was uploaded and that size changes that is made in-world (by the rigging) is ignored.

Link to comment
Share on other sites

You are about to reply to a thread that has been inactive for 2209 days.

Please take a moment to consider if this thread is worth bumping.

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...