Hey everyone, so I was just wondering how many parts will cause an average user’s computer to start lagging to the point where its noticeable if anyone has done research on that issue or just knows the answer. Thanks!
Depends on what group of people you’re marketing towards. There is no “average user”, as there are phones, consoles, laptops, PCs, etc. If I were you, I’d playtest it until I get a good estimate.
2 Likes
For me its like 1000 parts 
3 Likes
I would playtest it myself but i’m not finished my game so I need to know how many parts in advance sort of thing… appreciate it if someone can give me a good estimate. Thanks.
depends on properties. are they anchored? are they can collide? transparent?
Most of them, like 99% of them are anchored yes. There are some transparent parts but those are negligible.
I did read somewhere on google that transparent part, even 0.1, needs less performances.
makes sense because of rendering as long as you be mindful to do the most possible rendering on the client and leave room for the server.
Yes, and while that’s true I barely used any transparent parts in my game- they are usually just one part like windows, etc. I hope someone can still give me a rough estimate though knowing now that I used mostly anchored, plastic, non-transparent parts. Thanks for the info though.
A number of part before the lag is real depend on some things. It depend on their size, transparency, Position (sometimes). Also, not only building can create lags, do not forget about the scripting part. Actually, i doubt that someone would be able to give you a number
The impact of aggressive part counts will be slightly nullified by working with StreamingEnabled, if that’s something you’re willing to do.
The answer that anyone here gives you will be somewhat inaccurate, and this kind of question can only be tested on a per-experience basis.
The more active instances that exist, the more rendering that will be required per client basis (most times, depending on their properties).
What you can do is learn to actively take advantage of the tools that Roblox provides you and maximize performance by watching statistics. Do not attempt to guess at the average user’s performance, actually put the time into testing it.
1 Like
Thanks, and I would test these things like I said but I think a more general account from actual devs who have made huge worlds that hit their limit would be very beneficial to me for one single reason: foresight- so I can prepare earlier.
This is fair, but will have wild variety from place-to-place, even if all of those places experience heavy rendering. What I linked gives a lot of really good practices and content to help you get the most out of your individual place.
Unanchored parts process heavier. Partial transparency processes heavier. Collision parts process heavier (provided they contact the player). Etc…
It’s very difficult to guess at. I still recommend game content streaming for larger places.
Not sure what you mean by ‘needs less performances’, but parts that are Transparency = 1 are not rendered at all, so they have less of an impact on performance.
However, parts which are slightly transparent (anything above 0 and below 1) have a bigger impact on performance than a normal part because you are both rendering the part itself and other parts behind it (probably a better technical explanation elsewhere but that’s essentially it I think)
Going back to the original question - the fact they’re unanchored is very important. I believe there are some test places people have made that you can find on the DevForums here.
1 Like
Thanks for the help- I will definitely check those out. But I still have my fingers crossed that a dev that has worked with a lot of parts rendered can tell me a good estimate at least with their machine (which they can also specify some of its specs). Using this information I can gauge and use calculus to find out the average amount of parts that it will take for the game to significantly lag on the average machine.
If not, then I will simply have to do what you guys already told me to do and do case-by-case real time testing but this honestly renders (no pun intended) the test as useless because then the test would be making the place itself.
it has been said by one of roblox developpers that part with transparency between 0 and 1 is better as render question than a part without transparency
I think you might have misread something because that is definitely not true.
zeuxcg says this himself