- Summer of Code coding started this week. Follow development in the summer of code forum. Daily Windows builds of all the branches are available, though it's still very early.
- Work started on updating the manual, but we're not quite ready to get all developers involved in this yet. So it's postponed to next week, when we should have a clear tasks list and guidelines, and the major reorganizations of chapters are in place.
- COLLADA support has been improved in 2.80 but could use more testing. Leave any feedback on this forum topic.
New Features and Changes
There were a few user interface improvements
- Gizmos now activate on drag after a small distance, which allows clicking to select items behind gizmos. There are now separate drag thresholds for mouses and pen tablets. (Campbell Barton)
- Navigation gizmos are now available in the image, clip and sequencer editors. (Campbell Barton)
- Continuous grab can now work only in the X or Y direction for operators that don't use the other direction. This makes dragging the playhead in the timeline more predictable. (George Vogiatzis)
- Stats info shows only unique evaluated geometry now, to give a more accurate indication of memory and processing cost on the CPU and GPU. (Dalai Felinto).
- Walk mode now uses the status bar keymap display for shortcuts. (Dalai Felinto)
The focus continues to be on bugfixing. Some of the bigger fixes were:
- macOS viewport performance is much improved on certain graphics card that had lagging issues. To solve this we had to make macOS 10.11 the minimum version that Blender can run on, though we already only officially supported 10.12 and up. (Tomoaki Kawada, Brecht Van Lommel)
- Cycles memory usage for instances is now back to what it was in 2.79. This optimization was moved into the core dependency graph, which means Eevee benefits as well. (Sergey Sharybin)
- Cycles denoising has been improved to avoid halos around bright edges. (Lukas Stockner)
- Snapping now works on curve objects with modifiers. (Germano Cavalcante)
- Texture painting performance on Windows with many cores is much improved, fixing an old threading bottleneck. (Bastien Montagne)