View Issue Details [ Jump to Notes ] | [ Print ] | ||||||||
ID | Project | Category | View Status | Date Submitted | Last Update | ||||
0015069 | ParaView | (No Category) | public | 2014-10-24 00:20 | 2015-01-02 10:43 | ||||
Reporter | Alan Scott | ||||||||
Assigned To | Ben Boeckel (Kitware) | ||||||||
Priority | normal | Severity | minor | Reproducibility | have not tried | ||||
Status | closed | Resolution | fixed | ||||||
Platform | OS | OS Version | |||||||
Product Version | 4.2 | ||||||||
Target Version | 4.3 | Fixed in Version | 4.3 | ||||||
Summary | 0015069: Color legend marks are crazy. | ||||||||
Description | The color legend marks calculator is crazy. I realize that this is a hard problem, but what we have isn't correct. Here is an example, with two screenshots. Linux, 4.2.0, local server. * Load can.exo. Apply. * Last time step. Color by ACCL. * The next part is dependent on the size of ParaView - thus, varies. I manually change the color range to be 4000 to 2e8th, and get a minimum and maximum tick mark, and one in the middle. Changing the range to be 4100 to 2e8, I see a minimum and maximum, and 4 in the middle. There is a huge jump here, with font rescaling and everything. It is very disruptive. Please make it transition better. | ||||||||
Tags | No tags attached. | ||||||||
Project | Sandia | ||||||||
Topic Name | finer-linear-scalar-bar-increments | ||||||||
Type | incorrect functionality | ||||||||
Attached Files | color-legend-dense.png [^] (24,080 bytes) 2014-10-24 00:23
color-legend-sparse.png [^] (23,358 bytes) 2014-10-24 00:23 2014-12-03-113918_1918x1063_scrot.png [^] (262,313 bytes) 2014-12-03 11:40 | ||||||||
Relationships | |
Relationships |
Notes | |
(0033663) Alan Scott (manager) 2014-10-24 15:01 |
OK, I think I figured this one out. The problem is in file vtkPVScalarBarActor.cxx, around line 342. If the original algorithm decides there are too many raw tick labels, in a loop, it tries to create fewer of them. It basically tries fewer and fewer of them, until it has a good scale count. This is done by working down a magScale array of constants. The constants are spread way too far apart (or the algorithm itself is lacking). What I did was increase the number of items in this array, and decreased the distance between items. Here is my code that fixed this bug: //const double magScale[] = { 1.0, 2.0, 2.5, 4.0, 10.0, // 20.0, 25.0, 40.0, 100.0 }; const double magScale[] = { 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 15.0, 20.0, 25.0, 30.0, 35.0, 40.0, 50.0, 60.0, 70.0, 80.0, 90.0, 100.0 }; By the way, an easy way to test this is to change the state setting/ General/ Color Map Range options to "Clamp and Update Every Timestep". |
(0033664) Alan Scott (manager) 2014-10-24 15:02 edited on: 2014-10-24 15:02 |
Targeting 4.3, since I provided the fix above. |
(0033666) Alan Scott (manager) 2014-10-24 15:45 |
Next, we have issues in the same code with log color. I will add to this bug, since it is very close to the same problem, and in the same code. To replicate: * Load can.exo. Apply. * Last time step. Color by ACCL. * Color Editor/ Use Log scale. * Rescale to Custom Range. Min of 1e7, max of 1e8. There are NO labels! This is wrong! Another example is timestep 0, which as data, but it is very, very small (e-5'ish in size). The reason this is occurring is in vtkPVScalarBarActor::LogTickMarks(), where it calls LinearTickMarks(), after moving into log space, it insists that only integers be returned. But, some color labels don't have integers only!! I changed the call to be: ticks = this->LinearTickMarks(logrange, maxTicks, minDigits, false); This looks much better, however, the tick labels are actually on the wrong ticks and need to be corrected. I will let you guys fix it up, but returning floats looks much better. As a suggestion, LinearTickMarks should take three options for the last variable - true (I want integers), false (float is OK), and recommended (try to return ints, but I will take anything). |
(0033888) Ben Boeckel (Kitware) (developer) 2014-12-03 11:39 |
Pushed into stage. It looks to be right with changing the true -> false in LogTickMarks. The labels look to be in the right spot here. Attached is a screenshot showing the labels I see for 1e0 -> 1e6 (to make it easier to see what's going on). |
(0033906) Utkarsh Ayachit (administrator) 2014-12-08 08:04 |
Topics merged into master (v4.2.0-231-g445a425): 0015171_camera_front_of_plane_widget 0015172_automatic_annotations (VTK) exodus-15127 fix-log-mark-labels fix-osx-fileinfo fix_pvscalarabar_warning genericio_reader_optimization include-missing-header linking-with-kits-enabled (VTK) map-scalars-14943 pvweb-fix-extract-block-bug pvweb-fix-incorrect-filter-name qt5-find-components |
(0033925) Alan Scott (manager) 2014-12-10 15:19 |
The log scale has been implemented, but the linear scale has not. Note that I didn't test the log scale. Pushing back to Todo. |
(0033949) Utkarsh Ayachit (administrator) 2014-12-13 08:45 |
Topics merged into master (v4.2.0-276-g7d070dc): cleaner_multislice_view finer-linear-scalar-bar-increments (VTK) python-calc-15125 python-view-15122 simple-streaming-source sorting-15130 |
(0033973) Alan Scott (manager) 2014-12-15 20:20 |
Perfect. Tested master, remote server, Linux. |
Notes |
Issue History | |||
Date Modified | Username | Field | Change |
2014-10-24 00:20 | Alan Scott | New Issue | |
2014-10-24 00:21 | Alan Scott | Product Version | => 4.2 |
2014-10-24 00:23 | Alan Scott | File Added: color-legend-dense.png | |
2014-10-24 00:23 | Alan Scott | File Added: color-legend-sparse.png | |
2014-10-24 15:01 | Alan Scott | Note Added: 0033663 | |
2014-10-24 15:02 | Alan Scott | Note Added: 0033664 | |
2014-10-24 15:02 | Alan Scott | Target Version | => 4.3 |
2014-10-24 15:02 | Alan Scott | Note Edited: 0033664 | |
2014-10-24 15:45 | Alan Scott | Note Added: 0033666 | |
2014-11-14 22:52 | Utkarsh Ayachit | Target Version | 4.3 => 4.4 |
2014-12-02 08:32 | Utkarsh Ayachit | Assigned To | => Ben Boeckel (Kitware) |
2014-12-02 08:32 | Utkarsh Ayachit | Target Version | 4.4 => 4.3 |
2014-12-03 11:39 | Ben Boeckel (Kitware) | Topic Name | => fix-log-mark-labels |
2014-12-03 11:39 | Ben Boeckel (Kitware) | Note Added: 0033888 | |
2014-12-03 11:39 | Ben Boeckel (Kitware) | Status | backlog => gatekeeper review |
2014-12-03 11:39 | Ben Boeckel (Kitware) | Resolution | open => fixed |
2014-12-03 11:40 | Ben Boeckel (Kitware) | File Added: 2014-12-03-113918_1918x1063_scrot.png | |
2014-12-08 08:03 | Utkarsh Ayachit | Fixed in Version | => git-master |
2014-12-08 08:04 | Utkarsh Ayachit | Status | gatekeeper review => customer review |
2014-12-08 08:04 | Utkarsh Ayachit | Note Added: 0033906 | |
2014-12-10 15:19 | Alan Scott | Note Added: 0033925 | |
2014-12-10 15:19 | Alan Scott | Status | customer review => todo |
2014-12-11 15:42 | Ben Boeckel (Kitware) | Topic Name | fix-log-mark-labels => finer-linear-scalar-bar-increments |
2014-12-11 15:42 | Ben Boeckel (Kitware) | Status | todo => gatekeeper review |
2014-12-13 08:45 | Utkarsh Ayachit | Status | gatekeeper review => customer review |
2014-12-13 08:45 | Utkarsh Ayachit | Note Added: 0033949 | |
2014-12-15 20:20 | Alan Scott | Note Added: 0033973 | |
2014-12-15 20:20 | Alan Scott | Status | customer review => closed |
2015-01-02 10:43 | Utkarsh Ayachit | Fixed in Version | git-master => 4.3 |
Issue History |
Copyright © 2000 - 2018 MantisBT Team |