3D in Natron

[Micah Pendleton]

     So, the biggest thing keeping me from switching to Natron instead of Blender for compositing is that Natron has no form of 3D. I am not talking about stereo-scopic 3D, I mean real 3D for animation and stuff. It doesn't have to be modeling... infact, I would prefer it if Natron didn't try to enter the 3D modeling side. I would just like to be able to import a .3ds, obj, or .blend into the scene and render it in within Natron. This is an awesome feature that Nuke has. And the thing is, YOU DON'T HAVE TO MAKE A WHOLE 3D SYSTEM, you should be able to port from Blender like OpenShot does. 

     At the moment I'm using Blender for all my compositing,  can definitely see my self using Natron completely with 3D capability added.

Thanks, 

Micah

Dream Big, Pay Small

[johan.s...@gmail.com]

Op maandag 3 november 2014 01:13:17 UTC+1 schreef Micah Pendleton:

I see a lot of potential for this too, typical "2.5D" tricks that's already used in the VFX industry.

If you had 3D inside Natron you could for example re-light a model in compositing when it just needs some tweaks, for example some missing highlights or a little extra fill/bounce light. That avoids doing a potentially long render all over.

Further, you could add missing reflections on CG objects, again to avoid re-doing a time consuming render. Let's say you are comping in some fire and smoke using a video of fire. (Fire, smoke and explosion are typically hard to simulate and make believable) Now you realize there is a window that doesn't look right because it doesn't reflect the smoke when it should. Then you could just track the 4 corners of the real-world window, drop in a 3D card to project the fire footage on it and adjust the projected angle as needed. Now imagine the thing that needs the reflection is a car, curved surface, much harder case. But if you can import an obj and project the video on that it is not so different from projecting it on a flat pane of glass. The 3D projection does the hard work for you.

If you can import your model, keyframe its transforms, add some basic lights and do projections, that goes a long way without implementing any 3D modeling tools in the comp software.

[soz...@gmail.com]

I guess the devs are well aware of the benefits of 3D features in a compositing application...
But for now there is a lot to be done before getting to this.
Maybe before getting to 3D compositing, having more nodes to manage 3D renders can be a good step forward (even if for that it's not the good time to speak about it),
I'm thinking about a vector blur node (IIRC this is not far from beeing included) , and maybe an UVMap node could be a great addition : you can export an UV pass of your 3D render and use it to map some textures or videos to your beauty pass.
Or also do some masks for free that can be tedious to do only by roto.

You can already do some basic relighting with a normal pass. So for me these are some of the basics node that you can use a lot when dealing with 3D renders.

Also over-using 3D features in a compositing application kills a bit the benefits of compositing. You can easily get to a point where your compositing takes the same amount of time to render than your 3D render... For me it's better to export as much renderlayers as possible, (note that 3D render inside compositing application may be recalculated several times on the fly if you don"t cache it, where a renderlayer straight from your 3D application is only rendered once...)

I prefer a lot to have well polished tools that can be used to do a lot of different things, than to have many features poorly designed while lacking of fundamental tools.

Just my two cents...

[Johan Schreurs]

Good point!

I agree, better to have features that work well than to have a bunch that are half-finished. And I didn't think of the render times.

Looks like there are plenty of options already to do the things I was talking about. I might try that out if I can make a bit of time.

Its true that render layers and passes give you a ton of control for the lighting. In fact when I want maximum control I render layers for each light and group of objects. What I mean is, if I have 3 lights, some foreground objects and some background objects then I would want 6 render layers (with passes). That way I can still adjust the intensity of each light in the comp. I would also render a matte for the foreground and one for the background.

Op dinsdag 27 januari 2015 01:55:38 UTC+1 schreef soz...@gmail.com:

[Frédéric Devernay]

I second Paul’s opinion.

There is still lots of things to be done on the 2D side.

What we need now is developers.

There is urgent work to do on the following effects:

- Retiming / Interpolation / Morphing from vectors (aka optical flow)

- vectorblur

- all the transform nodes would benefit from state-of-the-art samplers (see https://github.com/devernay/openfx-misc/blob/master/README.md#transform-and-cornerpin-produce-aliasing-when-scaling-down-images )

Please contact the Natron team if you are willing to help (C++ skills required)

(BTW, the Card3D node in Nuke is actually a 2D effect, so it may be included soon in Natron)

fred

--
You received this message because you are subscribed to the Google Groups "Natron" group.
To unsubscribe from this group and stop receiving emails from it, send an email to Natron-VFX+...@googlegroups.com.
To post to this group, send email to Natro...@googlegroups.com.
Visit this group at http://groups.google.com/group/Natron-VFX.
To view this discussion on the web visit https://groups.google.com/d/msgid/Natron-VFX/982e442c-562f-4fe8-8722-e67709f4b412%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[mr.doct...@gmail.com]

I have been using nuke since version 4.6 not long after the 3D environment was introduced.

As far as I understand this was a rushed implementation of the feature set and hasnt got much better to date, dont get me wrong it does a great job but it could work alot better.

So i ask you guy not to rush the implementation of the 3D aspect of natron.

Sure get some thing out there but please dont lock your self into a dead end development road map for the sac of getting it out there, there are some key things that should come first.

obviously a camera, primitives (card first) and a "project 3D" node would be a great start.

80% of the work I do in 3D in nuke is simple to extremely complicated camera projection work for clean up, set extension and environment creation.

any future expansion on the 3d feature set I think would need to be planned from the out set.

yes that is correct, the 3D card is 2.5D basically the same as the AE 3D environment, sure add the 3D card but lets look at a true 3D tool set in natron

[mr.doct...@gmail.com]

relighting in comp isnt really a 3D tool set job, its pretty much a 2D exercise using normal passes and UV passes and mattes, yes it is nice to have objects in 3D space such as lights and axis to make the positioning of these "fake lights" more intuitive but essentially its a 2D task