3D-Coat 3.7 updates thread
#61
Posted 22 November 2011 - 10:09 AM
Macbook Pro4,1 - 10.6.8 - 2gb Ram - 512mb VRAM
#62
Posted 22 November 2011 - 04:41 PM
Please check if just saving scene restes performance. If so the lag may be related to multithreaded files saving.Andrew, it happens after a while of working on an object, doesn't matter what you are doing and what brush you are using, I have on numerous occasions, been working on a project which has no lag after brushing a stroke, then gone away from my computer for maybe 1 hour, then when I get back to 3DC the lag is suddenly there after just performing 1 brush stroke. The question I would like answered is "why does toggling proxy mode on then off fix the problem"? Even when I have imported an obj mesh and I am working with it in the Paint room the delay after each stroke occurs, then a quick trip to the Voxel room to toggle proxy mode twice and its back to painting without any delay. I mean, what has the "Toggle proxy mode" got to do with an imported obj file in the Paint room, there is no voxel sculpt in the Voxel room at all!
T.
#63
Posted 22 November 2011 - 06:02 PM
Please check if just saving scene restes performance. If so the lag may be related to multithreaded files saving.
I'm not quite sure what you mean by the above statement. I do not have automatic saving enabled. I manually save all my files on a multithreaded computer. Are you saying the possible root cause of the lag is by saving a multithreaded file. I will test...
#64
Posted 22 November 2011 - 06:22 PM
I mean - if you experience lags what will happen if you will just save scene and continue to work? Will lags stop for sometime or nothing changes?I'm not quite sure what you mean by the above statement. I do not have automatic saving enabled. I manually save all my files on a multithreaded computer. Are you saying the possible root cause of the lag is by saving a multithreaded file. I will test...
#65
Posted 22 November 2011 - 06:29 PM
Version 3.7.01A Linux 64 bit non-cuda
beta tools enabled and remove stretching selected
I opened 3DCoat (not just a new scene)
Selected the default voxel sphere (appox 250,000 one)
Swtiched to surface mode, then used the move tool for a short while.
Next used the rapid2 brush and on the area shown it produce the small slice hole and black hole in the mesh.
Look in the upper right part of the mesh to see the small slice looking hole. This is similar to the ones I got on the file I sent you.
I was testing for lag but will start again on a fresh opening of 3DCoat, since I got the error stated above...
Question. When errors are random and not reproducible all the time should these be reported to Mantis?
#66
Posted 22 November 2011 - 06:37 PM
After a save most of the time the lag is gone, but as the time pass it will become sluggish and saving will not solve the issue. Multithreaded save may be a possible cause but it's not the only one imho. With an intense use (lot of changes to model) over a few hours the reactivity slowly decrease. Also not sure if related but when save doesn't help much: cleanup memory and then after undo, sometimes bring back a bit of reactivity (not always though).
I can confirm that. Sometimes it seems random but the program will become sluggish after extended use. Closing and then reopening 3DCoat fixes the problem..
By the way, thanks for all your hard work on 3DCoat...
#67
Posted 22 November 2011 - 06:48 PM
- fixed crash mentioned by Michalis (export after reduce all)
- sphere tool with cubic interpolation fixed on transformed volumes
- fixed "Clear all" issue in surface mode
- multiselection in free form pose/primitives (but still without extra gizmo)
#68
Posted 22 November 2011 - 07:01 PM
About saving,on linux there is a problem(another kind of delay),that happens ONLY if you save the file on the host sub directories(windows) or on USB external drive.
In this case after the save you suddenly have a delay and slow down in performance,but if you save in your home sub directories no problem at all(apart the previous delay we are talking about)
#69
Posted 22 November 2011 - 07:13 PM
I'm talking about the Linux delay bug,it's a year I live with it,I think I know very well how and when it happens.
#70
Posted 22 November 2011 - 07:22 PM
#71
Posted 22 November 2011 - 07:49 PM
The picture here shows one cpu when lagging starts, going up to 83%. It hit 100% before being released.
The yellow cursor is left behind. The "crosshair" even though not captured in the screen cap is at the end of the stroke that is not produce yet because of the lag.
Linux version 3.7.01A 64bit non cuda
#72
Posted 22 November 2011 - 09:08 PM
- fixed crash mentioned by Michalis (export after reduce all)
Confirmed. It has been fixed!
Thank you.
#73
Posted 23 November 2011 - 02:26 AM
T.
#74
Posted 23 November 2011 - 09:24 AM
Surface mode testing
So far I failed to find a solid reproducible way of causing the lag. It still happens but causing it to happen in a way that I can write down the steps to make it happen, no such luck so far.
I have found though that I can break apart the mesh by certain regular brush actions and cause what appear to be tiny overlaps in the polygon mesh, create tears in the mesh and sometimes black holes. What appears to be happening in the tiny overlaps is a vertex in the triangular mesh edge is not connecting to the edge it needs to or an extra edge has been created. The tiny possible overlaps cause artifacts in the voxel mesh when you press the enter key or go back to voxel mode. Some of the apparent overlaps are so small you would not see them unless you went hunting for them.
Check the wirefile picture (left side) to see the tiny overlap. The surface mesh picture (right side) shows what the surface artifact looks like before I hit the enter key. This is a really zoomed in view of the problem. It's late so I will make a bug report tommorrow...
#75
Posted 23 November 2011 - 02:11 PM
Before clicking -
After clicking with the artifacts -
Maybe it is due to it being the default resolution but it seems broken to me the way it has those contour which look like artifacts. It should probably also cut only certain parts but it seems to cut then draw a slight extrusion with the mask so i think this is probably a bug.
Edit - Yes this is definitely a bug, i just checked with a much higher res and when not masked it gives a regular high detail cut but it give the low quality contour style artifacts when masked .
#76
Posted 24 November 2011 - 01:23 AM
By the way, I also had this exact same problem with my Intel 8core Xeon system, 16gb ram, Geforce 9800GT, also had FC14 and before that I had Centos 5.4 on it.
I hope you can sort this out soon as it is a MAJOR inconvenience when you are concentrating on the job at hand and you have to break concentration to Toggle the proxy mode.
T.
#77
Posted 24 November 2011 - 01:46 AM
T
#78
Posted 24 November 2011 - 01:51 AM
T.
#79
Posted 24 November 2011 - 07:56 AM
Tser,I'm not sure it happens at 200 strokes(I mean in general).Andrew, the delay 2-3 second delay (after each brush stroke), seems to occur after about the 200th brush stroke, I did a test with the default hi res sphere and mainly used the move tool and smooth, I counted every brush stroke I made. Using the latest 3.7.01B (CUDA)GL64) on Fedora core 14 (with all the updates), my system is 16core@2.3ghz AMD Opti, 32gb ram, mirrored 2TB drives, Quadro 4000, ASUS MoBo, system is rock solid and I also run Mudbox 2012 and ZBrush4 without any problems or delays.
By the way, I also had this exact same problem with my Intel 8core Xeon system, 16gb ram, Geforce 9800GT, also had FC14 and before that I had Centos 5.4 on it.
I hope you can sort this out soon as it is a MAJOR inconvenience when you are concentrating on the job at hand and you have to break concentration to Toggle the proxy mode.
T.
In my experience it depends on different things.
First of all,how high is resolution(for example with more resolution the delay happens before)
Second,how big is the radius of the brushes(big radius brush makes the delay happens faster)
For example,as I work it happens frequently that the bug starts on the move tool,which is a tool that in general I use with a radius bigger than other brushes.
Anyway,this bug isn't a surface mode/liveclay bug, it happens even on voxel mode.
Imo it must have something to do with the data structure used to store the points(to find the closest to the mouse pointer),maybe after the stroke the structure needs to be updated/recreated for next strokes, and here 3dcoat start to have problem.
#80
Posted 24 November 2011 - 08:09 AM
T.
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users




This topic is locked






