topic Re: Dimension ad element position accuracy in Revit Architecture Forum

Dimension ad element position accuracy

kh_jea — Mon, 30 Aug 2021 09:25:22 GMT

A colleague is setting out the grids for a new project. She set her dimension accuracy to a (in my opining ridiculous) small length. Centimetre with 12 zeros behind the decimal. This had brought to light an accuracy problem in Revit.

Even though the grids were placed with rounded length 810cm and 630cm, when placing the dimensions, the length is slightly off.

What's weirder not all dimensions have this, and using Global Parameters (blue ones) doesn't help either. (Sometimes deleting the dimension and recreating is does help.)

The project Units are set to millimetre with zero decimals. But changing this doesn't increase the accuracy.

Re: Dimension ad element position accuracy

RDAOU — Thu, 06 Apr 2023 20:12:25 GMT

@kh_jea

Attach the Revit file...you can copy the grids and the dimensions to a blank model if you do not or unable to share the model.

PS: was that value a manually input? if it is the case you will need to reset the precision and replace the dims...Revit has a 6 digits precision/accuracy so that might has something which is contributing to it ... trying to manually re-enter the rounded figure to reset the decimal 0 wont work because Revit will ignore such negligible inputs past the 6'th decimal (see below GIF)

Re: Dimension ad element position accuracy

kh_jea — Mon, 30 Aug 2021 10:11:39 GMT

"she tried but it wouldn't let her...right?" Exactly.

When I asked them why she insisted on using that level of accuracy, the reasoning was to get a perfect "levels-grids" model to the other project partners. Even though at best they use an accuracy of 1 millimetre.

I already told her to leave it as is, but I would like to know what causes this.

As these inaccuracies could add up in larger projects.

Re: Dimension ad element position accuracy

kh_jea — Mon, 30 Aug 2021 11:55:24 GMT

When I copy you're method it doesn't work on my end.

The behaviour seems unpredictable.

When I change the value to 1000mm, it changes to 999,9999... instead of 1000,0000..5 So it's not 800mm less like one would expect.

If I create a new dimension type and change the values using this one, it still doesn't pick up on it.

I already tried redrawing the grids, thinking an error was made copying the lines, but I get similar results.

Re: Dimension ad element position accuracy

kh_jea — Mon, 30 Aug 2021 12:09:46 GMT

I tried starting a new project and beginning from scratch. I uses the array tool to copy 10 grids with 8100mm in between.
First 4 are correct, next 6 are "inaccurate".

Re: Dimension ad element position accuracy

RDAOU — Thu, 06 Apr 2023 20:13:11 GMT

I just left, i will have a look at your file when im back.

If you have the grids locked to planes or associated to parameter or driving elements, that would make your life miserable reseting them. In such case deleting them and replacing is much easier

Is it everywhere or just this dimension string?

Re: Dimension ad element position accuracy

kh_jea — Mon, 30 Aug 2021 12:19:27 GMT

It gets even stranger.

Taking the arrayed lines and copying them to a different spot, changes the dimensions as well.

Creating a dimension between 2 parallel lines, they are simply copied, not redrawn, gives different values depending in where I place the dimension.

I have seen this happen with lines that weren't 100% parallel, but these are copied so they are definitely parallel. If they weren't parallel, the dimension values should become bigger to one side and smaller to the other, but that's not the case.

@lim.wendy @amanda.k Does Autodesk have an explanation for this behaviour?

Re: Dimension ad element position accuracy

lim.wendy — Mon, 30 Aug 2021 12:31:26 GMT

Hi @kh_jea,

I will check with the relevant team.

@RDAOU,

Great insights you shared, thank you. 🙂

Re: Dimension ad element position accuracy

RDAOU — Thu, 06 Apr 2023 20:14:45 GMT

@kh_jea

K ... I had a quick look at your file and there seems to be something very odd with the dimension style. I used the Dimension Style 2.0 mm - Transparent / Axis 0,0cm control and it seems to work fine...

The other Dimension Style names 0.0000 seems to have something wrong with it

@lim.wen

Re: Dimension ad element position accuracy

lim.wendy — Wed, 01 Sep 2021 09:14:26 GMT

@kh_jea,

I have reported this issue and it will be researched by our development team. Thanks for reporting this. I will update this threads when the development team update the problem report.

Re: Dimension ad element position accuracy

thomasarnoldsson5732 — Thu, 06 Apr 2023 08:13:33 GMT

@lim.wendy Any news regarding this? I still see the same issue and now I'm testing it in Revit 2024

Re: Dimension ad element position accuracy

kh_jea — Thu, 06 Apr 2023 08:18:12 GMT

Autodesk doesn't seem to care about providing sturdy software for architects, these past years they've been pre-occupied in adding niche tools nobody asked for.

I've given up all hope Autodesk will actually fix the issues users keep reporting.

Re: Dimension ad element position accuracy

lim.wendy — Thu, 06 Apr 2023 09:25:33 GMT

I am currently inquiring with the appropriate team. Regrettably, the issue remains unresolved. 😞

Re: Dimension ad element position accuracy

Mike.FORM — Thu, 06 Apr 2023 13:35:23 GMT

I also just noticed this when using the Autodesk developed Dynamo graph "Calculate Total Length of Selected Lines".

I had made the node which creates the list of all the lines after they are filtered to be an output so I can see all the line segment lengths in project and some of them have show a value other than 0 in the 11th decimal position. The value could be anything +/- a random amount in the 11th decimal position.

All lines were drawn where my snapping was only going to every 100mm at the shortest. You can also see the Calculated Total Length does not show the decimal error as it is only rounded to 6 decimal places.

Re: Dimension ad element position accuracy

RDAOU — Thu, 06 Apr 2023 22:41:38 GMT

@Mike.FORM

There is nothing wrong with what you see there. It is how numbers and computations are carried out in programing language (binary32 where what you refer to as a decimal point is in fact a floating point in the mantissa). In computers/programing world numbers and calculations are carried out/translated using IEEE single precision standard, and numbers are sets of 1s and 0s where the point is neither 1 not 0 but simply part of the format.

The values you see on the Dynamo end (or that of a temp Dimension in Revit) are not random but are in fact a variation and/or variations of the same number. In programing language 1 + 1 is not always 2.... Considering that in programing language 1 + 1 can for instance also be 2.00000000298023

Considering the 6 decimal point precision limit in Revit and looking at the values in the images you posted, it appears that Autodesk did not implement any form of rounding for the segment length in the dynamo graph; hence, the confusion between what you read for segment length in Dynamo Player as an output (raw floating values) and the values which would be displayed by Permanent Dimension strings in Revit (fixed/rounded). However, the total length seems to have been rounded down to 6 decimals to tally with what you might end up with is you manually calculate the total by place permanent dimension strings and adding them up.

Which is correct (Dynamo vs Revit)? Both
Is there a fix or will Autodesk do anything about this? in my opinion, there is nothing wrong to fix the first place and I do not think that this falls anywhere on the priority list

EDIT: It is an interplay of accuracy and precision. You need to keep in mind that by design Revit was not intended as micro chips manufacturing software where the decillionth of a number makes the measurement more precise or accurate. Hence, for what this software is intended for, anything beyond the 6th decimal becomes meaningless unless you want them to switch to double precision to enable us to carry out measurements down to the nanometer precision...ie: distinguish the coordinates of the adjacent mitochondria in a skin cell.