Meters and their myths: Introduction

[metatropic]

There are lots of questions on meters floating around forums (and facebook...). I am going to condense what I understand of each meter, and share with you all the results of some testing I have conducted to test out a number of popular beliefs about the different meters and what impacts them.

To begin, why do we have these bloody meters? They are the bane of of our lives as designers....

The meters are there to ensure that the courses we publish will play with acceptable performance across all the platforms; they reflect the limits of the lowest quality hardware that the game will run on. The meters also serve to prevent users publishing bloated courses that lock down servers etc.

The three meters, an overview:

File size: The course file saved when you design a course is 5-10Mb in size. Pretty tiny. Every time you place on object or press a sculpt button it generates a line of code in the course file; that line details the object selected, its size, (X,Y) position, elevation data (if raised or lowered above ground level) and rotation. Its about 100 bytes of data per object I guess. The impact of each entry in the course file is the same regardless of the item used; a massive skyscraper and a single grass object have the same impact. I suspect each sculpting brush stroke has a similar impact, but hard to quantify. File size only gets seriously impacted by small objects planted very many times, ie grasses and small plants. Anything larger planted in similar quantities will max out the object meter long before the file size becomes critical. 

Object meter: In most cases the rate limiting meter for course designers. The object meter is essentially there to limit the load on the GPU of the system, by keeping the total polygon count to a manageable level. The object meter reflects the product of the number of an object and that object's complexity. Grasses mostly are simple with a low polygon count, a skyscraper has a huge number of polygons.

Memory meter: This meter is there to limit the  impact of loading the course on system RAM. To render each object the system needs to hold the instructions for it in RAM. Each different object used (regardless of number of times used) takes up a bit of RAM. The different themes come with a number of trees, plants, grasses and rocks "pre-loaded" in RAM, which will minimise the impact of using those objects on this meter (doesn't appear to eliminate the impact though....see below). So memory meter is impacted only the first time you use a non-theme object. To keep memory meter under control you need to limit the variety of objects use. There is plenty of room though, you have to be really promiscuous with objects to max out, eg using lots of different city objects.

Stay tuned for chapter 1: Object meters and their myths....

[metatropic]

Chapter 1: Object meter myths

To test out the object meter I've conducted some simple experiments within the designer. Basically I get the meter to a baseline of exactly 1% and then see what it takes to get to 2% or  another number. Results below:

[metatropic]

Chapter 2: Impact of theme on object meter

[metatropic]

Chapter 3: File size meter and myths

[metatropic]

Chapter 4: Discussion

The findings of my testing seem to tally well with my understanding of how the meters work. Important findings include:

Copy and paste is not inefficient on object or file meters: you get exactly the same number of objects if you copy and paste as if you single plant.

So should we just use copy and paste with impunity? I don't think so. Obviously it lacks precision. And it's fast. Anything fast will see you eating up your meter without thinking. Slower methods make you ask if something is really worth the time and effort. Having said that I like sticking a load of trees down as I prefer picking the trees sup from the course than having to navigate the menus each time I want to plant. 

Cluster planting is not inefficient on object or file meters: you get exactly the same number of objects with cluster planting as with single planting for both meters.

So should we abandon single planting? No. Horses for courses. Cluster planting has some advantages; it adds variety in size and rotation that is more natural than with single spam planting. It's obviously quick (see above for the downside to that). But again it lacks precision, and it limits you in terms of the size of the objects you are planting. If you want really small or big grasses you have to single plant. And you can't sink clusters. Plus you lose the ability to "delete all" of a particular object, useful when you are working out your planting scheme still. If you use big dense cluster brushes and spam them, the clusters will overlap, leading to significant inefficiencies. Overlapped areas will be overplanted to no visual benefit. So use a mix of cluster planting when appropriate (if sizes right, for covering large areas where you want some variety in sizes) and single planting (for particular sizes, details and where you need to raise or sink objects). I'm only really talking about grass planting here, who uses clusters for trees?

Single planting is not meter efficient: it has the same impact on object meter and is massively inefficient on file size compared to cluster planting.

A restating of the above, but an interesting way of thinking about it.

Using theme objects saves on memory meter only, not object meter.

I haven't tested this for file size but I bet the same is true. The memory meter has lots of headroom really, unless you are using lots of complex objects.

Sculpting affects file and object meters

This is most noticeable in LiDAR projects, where interestingly the file size and object meter always match; LiDAR is very inefficient, it think it basically uses thousands of very small brushes to brute force the elevation data onto the course. Some sort of Fourier transform could make this better I expect.
About 100 sculpts (regardless of brush size, height, type) will lose you 1 trees worth of object meter, and about 20 grasses of file meter. It pays to use larger sculpting brushes when you can. It's interesting that flattening to remove previous sculpting will ADD to your meters. Best to undo or reload...
I can understand the need for sculpting to impact the file size meter, but can't see why it should count towards object meter. Yes it will add to polygons, but the way it is scored means you could sculpt for hours, max out meters, then completely flatten and still have maxed out meters. 

Hope you find this interesting and useful. Interested to hear if you have found anything different. I haven't yet tested the effect of splining surfaces or placing surface brushes. My suspicion is that the game doesn't include them in the meters as they are unlikely to impact performance; I've certainly never tipped to 100% because of a surface being placed.

[metatropic]

And just tested surfaces; they do impact file size meter at least.

3 surface brushes=1 object's worth of file size
6 pointed surface spline=1object's worth of file size

Holes don't affect file meter.

Crowds affect file size the same way surface brushes do, ie 3 brushes regardless of size is the same as 1 object in terms of file size. So for crowds it definitely pays to use a bigger brush if file size is the concern. 
Water panes and OB seem to be the same too.

[fergdawg]

This is great! To follow up on your last point about surfaces and splines, they definitely do impact the meters. From my experience, making huge splines with a large number of points adds a lot to memory. I was working on a course where I wanted to connect all the fairways and make one humongous spline. Pretty quickly, joining all the splines together I had 25% memory (can't recall if it impacted objects or not, but I'm sure it did) and everything became much slower to work with. I tried breaking the splines up but that didn't seem to effect the meter at all. It seems to follow your note about sculpting and how adding brush strokes only adds to the meter, and that only undo will reset the balance.

Cheers for taking the time to do this experiment. It's very worthwhile to at least dispel some of the myths we all want to believe are true.

[DragonAggie95]

Regarding clusters:

Plus you lose the ability to "delete all" of a particular object

If you break the cluster, do you regain the ability to "delete all"?

I'm certainly going to try this and it should be easy to test, but I wanted to post the question to future readers.

[metatropic]

Jul 6, 2024 16:51:15 GMT -5 DragonAggie95 said:

Regarding clusters:

Plus you lose the ability to "delete all" of a particular object

If you break the cluster, do you regain the ability to "delete all"?

I'm certainly going to try this and it should be easy to test, but I wanted to post the question to future readers.

Yes you do. They become single plants in every respect.