Hi, the Blender developer Brecht mention in meeting it will be published today.
Cheers, mib
3 Likes
Development
- Developers start updating the manual this week. (organization task )
New Features and Changes
The focus continues to be on stablizing for the 2.80 release. About 105 bugs were fixed last week.
- COLLADA: reworked material import & export. (details ) ( Gaia Clary )
- Gizmo’s can now activate on drag or on press, with a preference in the keymap to choose. With drag objects behind the gizmo can be selected, but it introduces a few pixels delay when starting to drag. ( Campbell Barton )
- Constraints now have icons in the outliner and properties editor. ( Alessio Monti, Andrzej Ambroż )
- Keymap: drag to box select now works in all animation editors. ( Jacques Lucke )
- Python API: allow external render engines to use Eevee for LookDev shading. (details ) ( Brecht Van Lommel )
Weekly Reports
Summer of Code
Follow progress and weekly reports in the Summer of Code forum
3 Likes
Today’s Meeting.
Development
- We will aim for a 2.80 release candidate around July 11, 3 weeks from now. After the release candidate only important bug fixes would go in, with every commit carefully checked. This will also give us the opportunity to test making the release build with all platform maintainers.
- Much of the Blender manual was updated last week, but still about another week of work remains. For anyone interested in helping out, there are still a few chapters without anyone assigned, and many of chapters where screenshots need to be updated. (organization task )
New Features and Changes
No new features now that we are getting close to the release. About 90 bugs were fixed last week.
Weekly Reports
Summer of Code Reports
Here’s an overview of Summer of Code progress so far. Some of the projects have features available for testing (builds available here ):
- Outliner : improved selection and navigation tools, start of 3D viewport synced selection, and many other improvements. ( Nathan Craddock )
- Procedural texture nodes : improved math, vector math and mapping nodes. New map range, clamp and white noise nodes. ( Omar Ahmad )
- Custom bevel profile : initial implementation of custom profile in bevel modifier. ( Hans Goudey )
- LANPR : integration with grease pencil, and many other fixes and tweaks. ( Yiming Wu )
- Fast Import/Export : STL export is working, OBJ is in progress. ( Hugo Sales )
For other projects work is happening under the hood still:
- Cloth simulation : adaptive mesh subdivision implementation is in place, to be integrated with the simulation. ( Ish Bosamiya )
- Embree : initial implementation of conversion from Embree to Cycles BVH, performance benefits will come once this is more complete. ( Quentin Matillat )
- VR: lots of low-level work to integrate OpenXR into Blender’s windowing system. ( Julian Eisel )
The final countdown with less than a month before 2.8 RC. We should also only be a week away from the dev team’s attention turning to the blocking issues in areas such as performance.
12 Likes
New nodes and node features? Interesting!
Also, I’m happy that bevel is getting more work!
I wonder what we can expect in the future for bevel - is the ability for the bevel to “eat away” existing geometry going to come, maybe? I saw it mentioned somewhere.
1 Like
These look hella interesting. Some suggestions:
-
Map range: Should have a toggle for preview, so that values below 0 show as black, values above 1 show as white, and everything in between show as a solid grey. This to ensure operations or musgrave output stay within intended range. When done tweaking, turn off preview mode and you know you have a good 0-1 range which can be hard to depict otherwise (especially with filmic). Secondly, maybe it should have a mask outputs for everything below 0 or above 1?
-
Vector Math: Should have a mask output, which would be specially useful for the cameraman example using minimum mode; the three remaining quadrants produce garbage and are not useful for anything. For image you could set it to clip, but for procedurals it is much harder to setup a node system for manually masking the area.
-
Voronoi: I know it hasn’t been started yet, but a blur function when using Cells mode would be very useful. Secrop showed me how this can be kinda done with nodes alone; it’s not perfect, but a whole lot better than I expected it to be.
Edit: Some other math suggestions, should show or hide sockets dynamically. Functions for:
- fLerp - linear interpolation. Same as color mix but with numeric inputs instead. Possibly vLerp?
- fSign - returns -1 for negative, 1 for positive, and 0 for 0.
- fStepFunctions - 3 functions; linstep (linear ramp), smoothstep, and smootherstep. Input value and two points exposed. Reasoning is that, while less controllable and configurable than the color ramp node, the two points can be driven per pixel which color ramp cannot.
- fBias - Log based function to raise or lower the 0.5 point of a linear ramp.
- fGain (fContrast) - Two fBias functions to increase or decrease the contrast of a linear ramp.
I believe fBias and fGain exists in optimized versions avoiding the log. fGain is not interchangeable with a stepfunction.
Omar’s second big feature post is dripping with new procedural texturing features, including a massive Voronoi update that includes this smoothing feature.
The student has actually just managed to get custom profiles working for common cases.
9 Likes
That looks awesome. But does that also apply to colored cells mode? I use it to procedurally modify normals to get big chunks of same angle rather than continuous change, but I’d like the transition between those normals to be smooth.
1 Like
I do not know as no example was shown.
The big feature post is here.
2 Likes
Oh. My. God. I’m literally crying here. Can’t wait for it.
GSOC Branches are not able to do what students are showing, yet.
We have to wait for them to update those branches to be able to test these features.
what chances it will be in 2.8 or only later on In 2.9 ?
thanks
happy cl
2.80rc is scheduled for middle of July before the end of GSOC.
Its development was frozen to focus only on bugfixes.
So, there is zero chance to have that in 2.80.
Half of GSOC that happened on last 3 previous years were integrated to blender 2.79 or 2.80.
Others were about continuing branches like Mantaflow or LANPR and will be merged into master after several years.
Because polishing 2.8 is a process that will concentrate mentors attention, there is a risk to delay a merge at a date when student could have lost interest about it.
If nobody is there to provide necessary work to make it end up in Blender before code is too much outdated, that does not happen.
But basic stuff that benefits from a large consensus about their usefulness, that does not seem to go against Blender workflow, that seems to be simple to adopt by users and are popular among testers : this kind of stuff is generally adopted during the 2 following years.
Fast Import/Export and Outliner branches are probably projects that will be merged quickly. Because they are clearly covering the subject of polishing 2.8.
Cycles Improvements branch is also about basic nodes additions. As a Cycles project, it has a huge probability to quickly end up in an official release as hair principled shader did. Target is a less ambitious than denoiser, it should take less time to add some of theses nodes to an official release (maybe there will be some in 2.81).
Except, if there is an issue for an unexpected reason, devs are failing : those 3 ones should end up in first half of 2.8 series.
LANPR could end up into second half. It seems to be popular and it should take same kind of path that Freestyle took but quicker.
About bevel GSOC, this tool/modifier have literally a developer dedicated to it since years. If Howard Trickey does not validate this GSOC, it would be because work done would not be satisfying for a meaningful reason.
About Embree for GPU, if student does not make it, there is also a big probability that a confirmed Cycles developer like Stefan or Sergey or Brecht does a tentative to achieve that.
In fact, the only real project that could be abandoned this year and not have a crucial interest for devs to try to reach that target would probably be Cloth Simulator Improvement.
But currently, its developer is active. So, wait and see.
4 Likes
forgot to mention the UDIM project
hope it make it too !
hope we get these news nodes for cycles and EEVEE
these gives a lot more options and very interesting
thanks
happy bl
4 Likes
Those procedural additions are just beautiful. I could see myself using them immediately in multiple procedural materials.
I find it a little distressing to hear that the most likely one to die is the cloth simulator improvements. From the description, this would easily cover from some of the worst difficulties the current clothsim has, especially stability and topology dependence.
Hopefully the project makes it to completion.
3 Likes
if the plan is to release 2.80 by mid july then what’s the status of optimization/ performance? shouldn’t this week be the start or will they be tackled last two weeks or so?
2 Likes
Is Bechman anisotropic completely broken in the current version (2019-06-21 21:37)? I’m using UV driven tangents if that matters.
Development
- We have about two more weeks of bug fixing before we do the release candidate.
- The user manual is mostly updated for 2.80 now, with the few remaining chapters to be done in the coming weeks before the release.
- We plan to update OpenSubdiv to 3.4 before the release still, to fix some important bugs.
- New committers were added for the add-ons repository: Cédric Brandin and Shrinivas Kulkarni.
- There is a blog post on the status of the sculpt mode features branch , and how and when to merge it.
- The first evaluations for Summer of Code have to be done this week, by Friday, June 28th 18:00 UTC.
New Features and Changes
The focus continues to be on bug fixing, with about 95 bugs fixed last week.
- Sequencer sidebar panels were reorganized since they were still unclear. ( William Reynish, Peter Fog )
- Deselect all and invert selection now work in the file browser. ( Campbell Barton )
- macOS bundling was updated in preparation for the release, where there will be a single Blender.app to install, that is code signed and notarized. ( Arto Kitula, Brecht Van Lommel )
Weekly Reports
3 Likes
Interesting. I remember Ton stating somewhere that getting a code-signed / notarized status for a macOS app required payment to Apple. Does that still apply, or does the single Blender.app form suffice?
UDIM will be ready on 2.81 that was said by Pablo directly on the live stream.