To 2.79 or not, bringing a good end to the 2.7x series.

As the title says, especially since it might be a while yet before the first of the 2.8x releases are ready.

My proposal, have a really long release cycle for 2.79 and make it a goal to get all existing projects (those that preserve backwards compatibility and most workflows) into master. The idea here is that since the BF is poised to end 2.7x (and a chapter of Blender development starting with 2.59), both of them can end with a bang and be large enough to carry users over until 2.8 is ready.

Another reason for that is that there’s a number of projects which I would consider to become a major catalyst for productivity, usability, and production; Cycles shadow catcher, Cycles denoiser, most of the GSoC projects, wiggly widgets, and the list goes on.

In a way, the need for a 2.79 may vary depending on whether or not there’s only going to be one 2.8 or it’s done in phases (ie. 2.80 > 2.89). If it’s done in phases for instance (as in 2.80 contains the new viewport code mainly), then there might be less of a need than if there’s one 2.8 and the one after that is 2.9).

Thoughts and ideas?

To somebody that don’t know nothing about 2.8

Will we lost a lot of features with 2.8? will we need a lot of times to see the new blender 2.8?

I am really interested in the Viewport rendering and how the Blender Internal is going to be replaced. I do not see much discussion of this, yet it is very important to all of the animation, illustration, anime, arch-viz where non-photo realistic renders in faster time frames make a world of difference in production. I also want to hear of any developments on Freestyle as there were NO! changes in the 2.78 release. This needs to be addressed, particularly in regards to the Cycles work, (non-Freestyle friendly) so that it can work more interactively with Viewport rendering. Perhaps cache the Map from the Viewport render and allow us to then tweak the parameters of the line qualities, in Real Time, using the Cache. This would put Blender in the foreground for these related 2D/3D visualization fields. I hope someone it listening and can give a response, as I have not been able to get much clarity on the full implications of the demise of the Blender Internal and what will fully replace it?? Anybody??

I would propose that Blender Render Internal stays in 2.79 and 2.8 series. It could go under various improvements.
For example, I was presenting some part of interactive geometry to the students. Notebook doesn’t have dedicated GPU but with Blender Render I could show them in stylized way 3D space really fast.

Also, Blender Render materials are very unique to use.

I would propose that BR stays for future developments.

I don’t see how Viewport rendering improvements are related to “Blender Internal going to be replaced”. And, personally, the day Blender Internal would be dismissed would be the day I’ll stop to use Blender and start to use different 3D softwares, btw.

2.8 is going to take a long time, if new features aren´t merged they risk either bitrot or loss of interest of developers. There will be a 2.79, even if not with a fixed release date. There just won´t be any dedicated focus on it. It´ll come when it comes (looks like no earlier than next year, though).

In the future release, there will be so fast and advanced viewport mode, that it is faster and has more features than Blender internal. OpenGL immediate mode and other legacy stuff will be removed, and the minimun OGL shall be at least 2.1 and the full feature viewport has 3.2.

So, based on this “logic”, I imagine that also Cycles will be removed as well…

Cycles is a physically solution at the level of Arnold/maxwell. BI instead is a old render engine that made the same that the actual game engines. The GI solutions of BI are old and really limited.

I don’t see reason to use blender internal right know when the real time graphics are the same level with a instant render time. I understand that to lost the BI will be hard for some people. BUt I think that if that give to the developers the opportunity to make new viewport that will give to the users a similar solution… I don’t see the real problem. Also no other software give us similar solutions to BI.

real time graphics are definitiely not on the same level as blender internal. They are a good approximation, sure, but let´s not exaggerate…

Give the features of BI that you cannot see in real time render engine.

I want remember that actually BI don’t have a GI solution for enviroment light because the solution don’t work.

Features are irrelevant, but you seem to claim they can all be done in real-time. Show us real-time SSS, on the same level as BI.

Look at what marmoset can do.

And maybe if the devs include USD and hydra.

the 3dview render has antialisasing etc, so I think it’s possible to replace BI.


never mind.

The only difference between using Freestyle stuff with BI or Cycles is that you have to use nodes to set-up textures modifying your linestyle.
If you are not used to add textures to your linestyle, there is absolutely no difference in the way to create your freestyle strokes.
Freestyle panels are exactly the same in Cycles UI.
So, Cycles is not more non-Freestyle friendly than BI.

You can prefer to create your NPR shaders in BI rather than in Cycles.
But it does not mean that it is impossible to create NPR shaders in Cycles or NPR shaders are Freestyle stuff.
Freestyle is just about strokes. And Freestyle is actually a part of Blender Internal renderer.

But if Blender Internal is abandonned, we will still need a renderer minimal part for outputting OpenGL renders image sequences.
The same way that Grease Pencil layers can be rendered and stored in layers of an .exr file.
Viewport renderer could include Freestyle stuff and allow same thing with freestyle strokes.

Currently, Grease Pencil allow to draw on surfaces but it is far from what Freestyle do.
Freestyle is creating 3D curves objects based on scene geometry and a bunch of parameters of length of strokes, chaining, clipping…
Maybe, devs will drop Freestyle for a faster OpenGL way to do exact same thing.
Maybe 2.8 object nodes will include nodegroups to do the same thing that Freestyle does and allow GPU rendering of theses NPR strokes through Cycles.

They will have to experiment before having an idea of what can be done.
It may be an improvement to drop Freestyle. It may be a mistake.
Replacing Freestyle by something else will probably mean a huge amount of work and a long period without a solution at same level.

And based on your so called logic, Blender Internal has more features and is more advanced than Cycles.

Not at all, BI is simply quite capable and a LOT faster than Cycles.

hmm why not do a partially upgrade, upgrade as most as possible close to the end goal, and dont be afraid of braking things for a while.
maybe a modular aprouch in which for the future sub parts can be improved (instead of major releases)
So blender version x, with viewport beta 1.0 and cycles 5.x and currently no sequencer, and no wiggly widgets, no 3ds support etc.
gradually add more features. 2.79a could be the last version with all current options still inside (+ some gsco).
though ad the same time 2.79b might be a test platform for 2.8 modules.
untill blender is completly upgraded.

The user could simple use the GE in blender too. Since the new realtime render will give PBS, it could also give some cell shading or even the old blin-phong. Its just a GLSL shader code.

If you don’t use GI maybe. If you want make something more that a flat rener BI needs more time that cycles only to make the proprocesing for the render.