# The Gradient filter produces ugly results
**URL:** https://discourse.paraview.org/t/the-gradient-filter-produces-ugly-results/15068
**Category:** ParaView Support
**Created:** [July 17, 2024, 3:35am UTC](https://discourse.paraview.org/t/the-gradient-filter-produces-ugly-results/15068 "2024-07-17T03:35:26Z")
**Posts on this page:** 8
**Page:** 1
### Author:  [@svetlanatkachenko](https://discourse.paraview.org/u/svetlanatkachenko)
#### Post date: [July 17, 2024, 3:35am UTC](https://discourse.paraview.org/t/the-gradient-filter-produces-ugly-results/15068/1 "2024-07-17T03:35:26Z")
Vorticity shows the following:
Y-velocity is much better:
Context: Flow near two parallel inlined heated plates
Same data zoomed out, Vorticity:
Y-Velocity:
Is there something that I can do to make the result look smoother and make sense?
Related topics
> [@QCriterion in Paraview](https://discourse.paraview.org/t/qcriterion-in-paraview/2355):
>
> Hi, I’m working on a turbulent duct flow. I used Gradient of Unstructured Data set filter on velocity vector then I checked the Compute QCriterion option and Compute Vorticity option. (as you can see below) [image] Finally, I used Contour to visualize the QCriterion. (as you can see) [image] The result can be seen in the following image ,but I think it’s not correct. Where is the problem? Did I follow wrong procedures?
> [@How to compute vorticity in Paraview](https://discourse.paraview.org/t/how-to-compute-vorticity-in-paraview/4474):
>
> I’ve opened the results of a simulation (Velocity field and Pressure) in Paraview. I know that the vorticity is by definition the curl of the velocity. How can I proceed to to compute the vorticity in Paraview? I can’t find a clear answer to this after searching. Could you please give a detailed answer for someone beginner? P.S: I am using a structured mesh for my domain, if that matters. I appreciate your help.
> [@Q-Criterion using the Gradient filter](https://discourse.paraview.org/t/q-criterion-using-the-gradient-filter/9149):
>
> Dear all, I am trying to visualize the Q-Criterion as one would normally do with the GradientOfUnstructuredGrid filter. Unfortunately it appears to be integrated in the Gradient filter in 5.10 and I can’t find the old options to calculate the Q-Criterion anymore… Is there a way to do it?
I’m using ParaView 5.11.2 on Windows 11.
Thanks in advance for your advice.
---
### Author:  [@nicolas.vuaille](https://discourse.paraview.org/u/nicolas.vuaille)
#### Post date: [July 18, 2024, 4:17pm UTC](https://discourse.paraview.org/t/the-gradient-filter-produces-ugly-results/15068/2 "2024-07-18T16:17:47Z")
Hi,
please share some data to reproduce.
Also you may want to try the 5.12 version or even the 5.13 release candidate ([https://www.paraview.org/download/](https://www.paraview.org/download/))
---
### Author:  [@svetlanatkachenko](https://discourse.paraview.org/u/svetlanatkachenko)
#### Post date: [July 18, 2024, 9:28pm UTC](https://discourse.paraview.org/t/the-gradient-filter-produces-ugly-results/15068/3 "2024-07-18T21:28:24Z")
Hi @nicolas.vuaille
Thank you for your note.
Attached example data.
[a1.vtm](https://discourse.paraview.org/uploads/short-url/vlPFtFbKL6pUVG1cqWpMCd8DL7V.vtm) (493 Bytes)
Vorticity:
Velocity:
I will check the other paraview versions soon.
How does vorticity work in your version? Could you please check?
---
### Author:  [@svetlanatkachenko](https://discourse.paraview.org/u/svetlanatkachenko)
#### Post date: [July 18, 2024, 11:55pm UTC](https://discourse.paraview.org/t/the-gradient-filter-produces-ugly-results/15068/4 "2024-07-18T23:55:43Z")
What does this mean? This pops up when calculating vorticity in the file attached above, in paraview 5.13 RC.
---
### Author:  [@nicolas.vuaille](https://discourse.paraview.org/u/nicolas.vuaille)
#### Post date: [July 19, 2024, 7:43am UTC](https://discourse.paraview.org/t/the-gradient-filter-produces-ugly-results/15068/5 "2024-07-19T07:43:16Z")
Hi @svetlanatkachenko ,
The `vtm` is a multi-file file format, you send only the root, so actual data is missing (`a1/a1_2_0.vtu`).
Concerning the error, it means that your data is a composite data, and some subpart does not have the requested array. Looks at the `Information` panel (and try the `Extract Block` filter).
Also please prefer copy-paste error message directly over screenshot: it is easier to read and helps other people find this post in the future 🙂
---
### Author:  [@Kenneth\_Moreland](https://discourse.paraview.org/u/Kenneth_Moreland)
#### Post date: [July 19, 2024, 1:00pm UTC](https://discourse.paraview.org/t/the-gradient-filter-produces-ugly-results/15068/6 "2024-07-19T13:00:41Z")
> [@nicolas.vuaille](#):
>
> Also please prefer copy-paste error message directly over screenshot: it is easier to read and helps other people find this post in the future 🙂
BTW, the `Output Messages` box has a `Copy to Clipboard` button to make this copy-paste easier. It’s easy to miss that.
---
### Author:  [@svetlanatkachenko](https://discourse.paraview.org/u/svetlanatkachenko)
#### Post date: [July 19, 2024, 11:43pm UTC](https://discourse.paraview.org/t/the-gradient-filter-produces-ugly-results/15068/7 "2024-07-19T23:43:39Z")
[a1\_2\_0.vtu](https://discourse.paraview.org/uploads/short-url/34Ruraf4hs4yaSfYriSSGHAzmZi.vtu) (66.7 KB)
---
### Author:  [@Kenichiro-Yoshimi](https://discourse.paraview.org/u/Kenichiro-Yoshimi)
#### Post date: [July 20, 2024, 3:30am UTC](https://discourse.paraview.org/t/the-gradient-filter-produces-ugly-results/15068/8 "2024-07-20T03:30:06Z")
The poor aspect ratio of the triangular elements in the cross-section may be causing inaccuracies in the numerical differentiation. Indeed, interpolating the velocity vector on a Cartesian grid using the `Resample To Image` filter and then calculating the vorticity with the `Gradient` filter improves the contours.
If the accuracy of the numerical differentiation for the polyhedral cells is not significantly compromised, calculating the vorticity for the polyhedral cells before cutting may improve the contour.