Then i restarted my computer (not sure if it's needed to make use of newly added environment variable, but at least it was done). i added OMP_NUM_THREADS as an environment variable and set it to 1.Hello, no idea if it's helpful, but here's my observations : I will test the new version when i will have more time tomorrow and report if what i wrote is still valid for the new one. So the following was only valid for MyPaint 1.2.0 beta 0.73 (Hope that understanding helps anyone trying different things out)Ĭutting to the chase, are things like the frame edit mode affected too? Just moving the frame edges doesn't use the brush engine, and it doesn't queue events like freehand does: it would only be crappified by crappy rendering performance.ĮDIT : i didn't noticed there was a new beta available a few hours ago We might be able to make a similar draw()s - per - second estimate as well, for debug mode. These don't happen concurrently given that "a" and "b" both process in the main thread, so it's safe! however it's possible that either one might starve the main thread. I need to understand the way the brush engine does its threadified tile access better, but it's clear that it has to write to the tiles that the rendering side needs to read. Both "a" and "b" process in the main GTK thread, although "a" forwards its work to the C rendering engine which is threaded (via OpenMP) by tile for sufficiently large sets of tiles (big brushes), and "b" uses partially OpenMPified C++ pixelwhacker code to render the layers stack for each updated tile before throwing it over to Cairo. Art apps need that figure as high as possible for accuracy's sake, so we just shove the events into a queue for later a) stroke rendering, and b) the screen updates you get as a result of the model data changing. Just a thought on making freehand lag more apparent: MyPaint already makes a rough estimate of events per second on the capture side, visible in debug mode. "one point two point zero!" - but don't sweat it, and thanks for the GIMP link ☺īut yeah. Sometimes the rendering of the entire stroke is delayed all together. There is a noticeable falter or stall, then a rush to complete the rendering of the stroke. Please note that the issue I'm highlighting is not related to this and should be obvious from observing the strokes as they form in the videos provided. A few brushes used are 'slow' brushes, brushes that, based on their individual settings, are set to lag behind the movement of the cursor. The first segment of the demo contains a few strokes that were inadvertently made outside of the camera's view. Anyway, this goes to show that the issue is not machine or Windows version specific. Oddly, the hiccup seems more prominent on the more powerful machine. The website rated the processor of the ATIV 700t at approx. Also experienced the same probleam with the first beta release (mypaint-w64-1.2.)Įxtra: Graphics chip is Intel HD 4000, Windows 8.1, 4 GB another video of two versions of MyPaint running on a much older machine, an LE 1700 made by Motion Computing. In case there is a problem reading the text in this compressed video, all relevant text is below.ġ.2.0-beta.0.73+gitexport.1932648 (64Bit) Again, Tumagonx's 1.2.0 release gives me a 'normal' experience. To answer achadwick's question, I did try the 32bit version and I suffered the same laggy/stuttery experience. Running on a Samsung ATIV PC Pro 700t running a core i5 CPU 1.7GHz (this machine runs the latest Krita and Gimp easily). All version numbers and specs are shown in the video and listed below for clarity. Google Drive and Dropbox links to the same video provided below.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |