LightWave 3D Group Offers $695 LightWave 11.6 Crossgrade to 2D and 3D Artists

Started by cyphyr, September 15, 2014, 08:53:54 AM

Previous topic - Next topic

Oshyan

Pablo, you definitely have some legitimate complaints, the "update-while-typing" issue is one that certainly annoys me as well and I agree it should be fixed. I'll make sure there's a bug filed for it (I think there is but that fixing it is for some reason not as trivial as you might think). However I don't encounter the UI speed issues you have and with a machine as fast as your AMD FX machine (which I assume is your primary), I wouldn't expect such issues... However you *are* dealing with some very large populations and I wonder if any other software would handle them any better really. You mention you've never tried populating 50 million instances in e.g. Lightwave (or Modo, presumably), so why not try for comparison and to set your expectations realistically?

- Oshyan

PabloMack

Oshyan, I have painful experiences implementing edit buffers where new entries do not take effect until the whole entry is complete. I don't know how the system is implemented on a Mac OS, but in Windows, the message-oriented heart of a windowed application doesn't permit the main thread to hang around in the context until the edit is complete. This would cause the UI to freeze up and I know all agree that is not acceptable. So maintaining the context while allowing all other (possibly unrelated) events to proceed in parallel fashion is a challenge but can be done. I don't see any way around having some sort of global state machine that has "transitions" whereupon events such as key strokes and mouse events that affect the field being edited can show progress in the UI while permitting everything else to take place in an interleaved way. I suppose it is the nature of the beast. With TG having so many fields to support in so many detachable panels/windows, a change such as this certainly would be a daunting task to say the least.

I can't say that I am an expert in the way Windows manages threads. But guaranteeing that the UI interface has priority over time-consuming activities like render preview is probably the key to having a responsive UI. Nothing the main thread does should permit it to be tied up for very long. But I have witnessed that a core can be tied up on processing that has been assigned low priority and it can happen that the higher-priority process that has become unblocked just sits there waiting for the OS to assign the core to it. I am sure this is not a TG issue as this behavior I am sure can be attributed to the OS's scheduler.

Please take my criticisms in a positive light. If I used terms that sounded too harsh, I was merely trying to answer Kadri's question. I am far from the point of having to go looking for another application to do what TG does. Thanks so much for your support and perspective.

Oshyan

Not to worry Pablo, your input is appreciated and we just want to make sure that we're doing what we can to ensure TG performs as well as possible (given limitations of OS, hardware, coding time, etc, etc.). I'm still curious if other apps can populate the same number of instances and remain more responsive. ;)

- Oshyan

Kadri

Quote from: Oshyan on September 22, 2014, 04:54:41 PM
... I'm still curious if other apps can populate the same number of instances and remain more responsive. ;)
...

Actually i am afraid to test it in Lightwave :)
I am not sure but just of the way i have seen Terragen handling huge image files and
objects i doubt that Lightwave can do it but who knows.
Would be nice to seeing others testing this in different programs.

PabloMack

Quote from: Kadri on September 22, 2014, 05:12:03 PMActually i am afraid to test it in Lightwave :)

Afraid indeed! I just tried cloning 1M instances of an object and my whole system locked up. I don't think it got past the instantiation before both screens went black. I did get the Windows 7 ring cursor back eventually but it wouldn't move. I had to just force a shut down. What I expected was that it would be catastrophic failure rather than performance slowly degrading as TG seems to do. Since 1M instances is only a tiny fraction of what TG will handle its says a lot for TG to be able to do as well as it does with huge populations. Since LW has lots of animation capabilities I expect it requires a lot more memory to store one instance so it doesn't take very many instances (on a logarithmic scale) to bring it to it knees. But locking up the whole system is pretty bad. Looks like I was talking apples and oranges when it came to comparing the two software packages.

Kadri


LOL! By the way which version do you use Pablo ? Mine is 11.61 (i haven't installed 11.63).

PabloMack

The About menu of Modeler says "Ver 11.6.1 build 2727" so I probably have the same build as you. I'm still waiting on support for Kinect 2 so I haven't been using LW very much lately. But I'm about to get another LW project so it will be good to brush up on my skills.