Looking through the threads where file size is discussed, I am wondering how big is a big file?

I recently inherited an stp file for editing. It was about 500KB. Nothing complicated. I added some embossed 3D lettering (6 letters) some booleans and fillet ALL the edges of the letters. So a lot of fillets. The exported file was 3.6MB, which surprised me. Especially since it causes a few problems, emailing for one.

Is this a reasonable sized file, or would it be considered big in engineering circles?

I did an experiment: import stp to MoI, then immediately export. File size went from 471KB to 729B. (I did the same in Alibre and the output file was even larger than that)

Even so, to my mind 3.6MB is pretty big.

Hi Steve, I would not generally consider 3.6 MB to be a very big file. Maybe something like 15 years ago or so it might be more on the borderline of that possibly, but not these days.

I guess I would call that in a kind of "medium" sized range which I would put at a pretty broad range.

Are you having difficulties loading or dealing with the file?


It is likely that something around 95% of the data size of your file is contained in the surface data for the fillets. You can kind of get a picture of this if you select your object, run Edit/Separate on it to break it into individual surfaces, and then turn on control points with Edit/Show pts. You will likely see that the fillets are pretty dense with control points - that is a normal consequence of fillets as they have a curved shape to them and are not just simple flat surfaces. So it tends to take a fair number of control points for their shape to be described properly.


You asked:
> or would it be considered big in engineering circles?

Typically in engineering circles they are used to dealing with data of a size many times larger than that, so from that standpoint this would be small file definitely.


> I did an experiment: import stp to MoI, then immediately
> export. File size went from 471KB to 729B.

I'd consider that to be in the same ballpark, nothing particularly unusual or unexpected there...

If you would like to send me the files I could take a look at them and give you a more detailed technical explanation why the size grew a bit in that particular case. Were you exporting out as STP again?

It is likely to be something like MoI's STP export uses a larger number of digits in the point data that it writes out, like writing coordinates as numbers like -3.04520050437171 instead of truncating it to a smaller size like -3.0452

It is tough to say without examining your data but like I mentioned it is in the same ballpark and not at all an unusual or unexpected thing...

- Michael

I guess it's relative to what your machine can handle. Some of my work items get to be around 60-80mb for some pieces. I know what I am trying to achieve and usually work on large chunks in separate files rather an entire assembly in one file.

I have transferred files from Solidworks through MOI for meshing and they can run from 100-160mb sometimes and you really learn patience.

OK thanks Michael and Grendel, that makes me feel better.

>run Edit/Separate on it to break it into individual surfaces, and then turn on control points with Edit/Show pts. You will likely see that the fillets are pretty dense with control points

It is very educational to do this! I see what you mean when you say

>It is likely that something around 95% of the data size of your file is contained in the surface data for the fillets.

Also cool to see all the construction geometry.

Part of the problem could be that I work in mm...I read somewhere that that will increase file size, because of the bigger numerical values involved (or something)

The only problem I was having was with sending the files by email...i think some isp server timed out. But if I compress them, no problem.

Anyway, I feel a lot better about it now

Hi Steve,

> Part of the problem could be that I work in mm...I read somewhere
> that that will increase file size, because of the bigger numerical
> values involved (or something)

Yeah, that can be a problem. The deal is that regardless of what unit system you are working in (like whether it is inches, mm, whatever), many operations are carried out to a precision of 0.001 in that unit system.

If you have an object that is say 10,000 units in size, then 0.001 units is a very small fraction of its size, having operations fit to that degree of accuracy will tend to generate denser results, as things continue to refine calculations until they get to that degree of accuracy. Also it will take longer for calculations to complete as well.

So it tends to be best not to create things that are more than say 3,000 units in size or so. If you are creating a lot of objects like that you will probably need to switch to cm instead of mm.

I am working on improving this though, to move calculations to a more adaptive type of target accuracy that is based on a fraction of the size of the object instead of just a hard-coded 0.001 value. It will be a bit before this gets applied to all commands, it has been applied to a couple of them already so far though, like Join and Network.

When more commands are moved over to use the adaptive tolerance it should help to greatly reduce this problem.

- Michael

Steve,
if it makes you feel any better, I couldn't fit my 4th year capstone project on a CD, that was in 2005.

Adam.

Thanks, Adam ,that is comforting:)

PS that must have been quite a capstone project. What were you modeling? Can we see it