maya 2019

[Anonymous]

Hi,
 Any idea when Maya 2019 release ?
thanks alot

[pmakal]

Yeah ok, I guess saying that's impossible is a little bit of a stretch but you got the idea 😛

[jasoncbraatz]

I do think it's impossible; and here's why ( I used to work @ Autodesk ).  A likely scenario is an activist investor like the hedge fund Sachem Head, who Autodesk has already had to break up the board for, would want to divide up the product lines into separate companies to create shareholder value.  So therefore, Maya (and everything related to the Entertainment / gaming sector, for example) could be spun off into a completely new company... so could any one of a number of software products when you have a very diverse portfolio of products (like Autodesk does).  Day 1, investors in Autodesk profit from this because the coverage of a "pure play" software company is generally more valuable, or commands a higher multiple, than those with a diverse portfolio.  What this would then do for customers of Maya is either a terrific thing in the long term (better management of a smaller company makes it more nimble and able to incorporate newer technology quicker)..  or it dies on the vine as a separate company (remember Quicken?  Most don't.)  It boils down to the same thing whether it's a Autodesk company or it's not: who is managing the product.  Product managers, like Daryl Obert (for Maya) are actually easy to contact, you can find his info on Youtube if you search for "Maya Mondays."

There's a negative value in the near-term for any company to kill Maya ; so in the end, it's impossible [at least in the present-day] that it could ever be "killed" for some time to come.  That said, I think it's in the interest for every 3D artist/TD/generalist/freelancer/studio to have at least a good understanding of the differences with each new version of Cinema4D, Houdini, Blender, Modo (or anything on the horizon) as they come out.  Watch the blogs, even if it's a skim.  For one, while Maya may not go anywhere - the artist may want to.  They were still selling VHS players when DVDs came out.. so sticking with a platform or technology isn't ever a proven strategy.  Neither is sticking with a company.  I highly recommend, even at the super-small studio level, setting aside around 1 working day a month of just absorbing new changes in the industry, new software, etc (through Youtube, Lester Banks, FXPHD, etc), downloading demo's or watermarked versions of new things which come about and just playing around is also useful.  I go back to my Ryobi vs. Craftsman analogy.. software is a tool - the artist is the one who has to get something to look like anything.  But the artist may find that another tool does something for them quicker (or in rarer cases, better).  I've been in CG industry since 1993 (!) rendering on an Amiga.  Tools ALWAYS change.  I think I've heard about 20,000 times in that period how mocap would replace keyframing, how AI would transform the entertainment industry (dating back to the 1990s), how XSI was the "standard" and thus "would never, ever go away."  So this means mileage varies... mocap replaced some things but not all, and XSI got quite a slow death.  But it didn't happen overnight, and those who continued to just be on top of the tool marketplace weren't disrupted, they just, over a period of time, "migrated."  The more you are abreast of the advancements and new products from all of the tool companies, the lesser impact it takes on you or your studio when a migration would be needed... (and you'd even be the one welcoming it when it does happen).

[Anonymous]

Any updates on 2019 Maya release? 

[regname5832]

I am Maya user and I like this soft and I used to use it. But if Maya 2019 be the same to Maya 2018+some small features and bugfix... so, I think, I'll switch to Blender with their features and evolution speed. Stagnation of Maya software development is so sad for me.

[damaggio]

Yes please, go to Blender today.

[jasoncbraatz]

Better yet.. migrate directly to MS Paint 3D, it comes free with Windows 10.

[abercaine]

yeah sure a revamped UI and eevee, really crazy progress.
Also, this was because of the code quest now everything will slow down as hell. 

[Anonymous]

I believe this is sarcasm.

I think you should however realize that blender 2.8 is a massive leap from where blender used to be. For instance blender now has the best viewport of any 3d dcc on the market. Makes viewport 2.0 look bad in comparison.

[jasoncbraatz]

Your guess that it was sarcasm is correct; but how it derived isn't.  

Your viewport argument is DOA.  Maya supports DX11 shaders, making the viewport WYSIWYG if you want to.  Since Maya DAGs can have multiple shader groups per node, puggybacking a DX11 shader with an Arnold Shader [of the same texture] for example, is near-WYSIWYG, sans GI and AO, etc.  But it depends on the project.  You wouldn't want this LOD with a large production scene, it'll slow down a brand new 2080ti to it's knees, so there's a reason why it's done this way and not pretty out of the box.

As mentioned, as artists we should be staying on top of the overall technology (render engine trends/ai-denoising, studying new proceduralism techniques, photogrammetry modelling, etc) to expand our "toolbelts" when needed.  But in the end, it's always the artist, and not the tools. Thus, having "tool envy," is quite useless.

[abercaine]

@Anonymous

So ridiculous seriously, typical answer of the blender guy that really don't know much on how things are working.
 They just having to put in front of you blingbling stuff and a new skin/UI and you think now that blender is on the top of the sky.
Have you tried to go in edit mode in blender 2.8? it's even slower than the 2.7. For sure things are evolving but this biased vision of things is really ridiculous.

As jason said maya vp2 is fully capable also any renderer could decide to make their shader fully compatible with the viewport like what Arnold is doing now.
Have you seen this video? Man look at the date this video is from 2014! And look how good it's already looking.
https://www.youtube.com/watch?v=8qeOFibRmoo   

[jasoncbraatz]

Good call on turd rendering, since Maya would be the best tool to use for proper human turd rendering.  Let us examine the process of modelling and rendering the human turd:

- photogrammetry of a dense mesh of the turd, using Autodesk's ReCap and exporting it as an FBX to Maya
- since the average human turd is approximately 83.82mm by 30.48mm in circumference, generally with poles, Maya is indeed the best tool to use for such density.
- Using Maya (2018's) new UV tools, a simple automatic UV unwrap should suffice since it uses a best-guess methodology on a single object, and it'd be a perfect scenerio using a human turd.
- Assuming that rendering is done with Arnold, we'd have to sample the SSS at three points using Arnold's built-in skin shader (quickest for good SSS results)
 * sample the specular color (in Maya's HSV color picker) of a 2D picture of the turd and input that as the initial color
* the single surface scatter should be near the areas of the turd where the blacks meet the browns intensely
* the double scatter should be sampled with HSV of the most intense color
- Adding an additional SSS shader and plugging it into the shading group with the file of the UV of the turd picture.  A 32-bit EXR could be attained using a ReCap + Photoshop
- A good lighting model of the turd would be to use a single Arnold dome light node with a normalized dome map for the inside of a toilet bowl.  I'm not suggesting that'd be the only lighting model, however.  It's certainly known that human turds, using a lighting model from an outdoor sky, would be equally relevant if the final turd was to be placed amoungst foliage outside [for example].
- Upon examining the details of a human turd, there are varying amounts of random indentations which actually manifest as almost little hairs on the turd.  Thus, I suggest that we create an Xgen description and creating an input map to properly get this achieved.
- The proper camera settings in Maya would likely require a DoF (using an Arnold bokeh) at ISO 400 - OR - one could render these out as AOV passes and re-comp them in Nuke, Fusion, Flame/Smoke, Photoshop or AE
- Given that toilet bowls are small, the need for atmospherics is not there, so there'd be no need to reference in - say - a VDB or create a maya fluid to recreate an atmospheric effect.  However, the water in the bowl could be effectively be recreated using a Bifrost procedural process.

The physics of the turd aren't discussed here since I believe you were referring to rendering just a still image of a turd.  However, if you'd like to add some dynamics to it, it'd add about a dozen more steps.  You are correct in that Maya would be the best choice as a DCC to render a turd effectively, as the limitations in rendering in Blender wouldn't provide enough of a palette of PBR consistencies required to produce it.  Cinema 4D could do it as well as Houdini but with different methods then mentioned above, of course.

[Anonymous]

.

[JabbaTheNut]

@jasoncbraatz,

That is awesome! I am definitely packing that process into my knowledge box for future use 😁

[Anonymous]

I only made a simple statement about blender 2.8 having the best viewport. This is because it was rather suprising that a comment was made that in effect, compared blender with paint 3d.

That being said, when I referred to the viewport, I meant quality not performance. It's unfair to judge a software in beta for performance(referring to your comment on edit mode), performance optimisation is usually done at the last stage of software development when all major features are in place.

If we are all honest with our selves, out of the box blender eevee-powered viewport is superior to viewport 2.0. This is not based on biased opinions. My reasons are below:

Viewport 2.0 is far more limited in range of effects it could do by default (if compared to eevee).

There is no such thing as Irradiance cache,

SSS(it could be done with a fake looking method only in shaderFX),

No screenspace reflection/refraction, no planar reflections, no lightprobes,

No proper AO (it is there but a relatively primitive one),

No proper DOF (when compared to eevee),

No bloom,

No volume lights, no proper volumes,

No contact shadows.

Yes you could write shaders and even use shaderFX editor to mimic some of the pbr stuff (only on material level - you couldn’t add SSR or contact shadows for example), but out of box the only thing it could do is a rough approximate representation of the arnold shaders, without proper shadows and other effects listed above.

Overall the visual quality of eevee is simply just superior.

[animation]

Eevee is far superior to VP 2.0 out the box, no doubt about it. Great post! Blender 2.8 is kicking butt. I love Maya and Blender, and both can be used for different projects (or both). Since Maya killed Maya live, I have been using Blenders tracker which is quite capable. Blender's compositor (and Natron) is also badass, and it is pretty seamless to go from maya where we do the heavy lifting - and for FREE. No proper deep support, but it will come sin e the same academic papers and algorithms are available to all integrators (which is really all autodesk is- they don't really invent much beyond putting a nice interface on algorithms publicly available and put some sugar on top)

Blender really does seem to be the future in many cases

[abercaine]

nowadays you just have to through some glow and bloom to kids and they think that they will rule the world.
What i want from a viewport is performance not some useless stuff, i mean it's cool but not ground breaking.

We have two cases you either work with a final frame thing or Real-time thing.
-For final frame engine you have GPU raytraced renderer like Redshift, which already blazing fast in ipr mode, more than enough for look dev and at least i have a real thing going on.

-Now the second case, you work for the Real Time industry, i will choose unity or Unreal because they are Real game egine that are faaaaar better than eevee even on quality and performance.

Because eevee is not even a game engine it's just a RealTime engine and basically you can't do sh*t with it besides just rendering in blender which is useless, you can't even export the shader to unreal or something; no you are just in blender doing rendering with an Real Time engine (not even a game engine), and a Real Time engine that is actually much inferior to Unreal on every aspect.

So what's the deal here tell me? Yes it's cool and funny and seriously i don't give a sh*t about it.

Now i think it would be fairly easy to have a maya plugin with the Unity or Unreal engine working in the viewport and be able to send directly to Unity or Unreal, that's would be really useful.

So thinking that blender is the future because of eevee it's really ridiculous.

[animation]

@abercaine You clearly didn't understand the points here. 

1. NO ONE said they are going to ditch maya because of ONE feature (eevee), "Bloom and glow being thrown at kids" is a ridiculous statement, there are plenty more features that are amazing and it takes more than that to impress "kids" these days. 

2. Maya already has a send to Unity function, as well as Blender since about 2013

3. NO ONE said anything about Eevee being a game engine (which btw, Blender has seperately already) 

4. You can render directly out of the view port to any AOV you want in Blender or Maya, as if it were sent to an offline renderer - all with high bit depth and alpha etc. (Only realtime features supported render to AOV's), saving time when sub-frame GI re-caching is not needed 

5. Blender Stack exchange has myriad tutorials for exporting shaders to Unreal or Unity.

6. Performance in the viewport is great, so you got what you want.

7. The gorgeous viewport is not useless - do you know how much more money you get when a client sees 90% of the final result (or can make changes) live on screen without submitting to a farm? Clearly you have no idea what you are talking about.

8. Redshift/Octane/Cycles is great for offline rendering, but you are not showing clients real time animations with full-fidelity using those without hearing complaints about noise, I don't care how fast the GPU's are. The viewports have their places and most people love having them both as options.

[abercaine]

he -sending things from blender to unreal or unity is the worst workflow ever, it sucks as hell.

- Are you following a little? The game engine in blender was ditched with the 2.8, so there is no game engine at all now, even the logic editor was removed. 

- AOV in Cycles are fairly limited and with eevee it's even more limited.

- And no viewport performance are not quite there compared to Maya or Houdini, really. 

- "do you know how much more money you get when a client sees 90% of the final result (or can make changes) live on screen without submitting to a farm? Clearly you have no idea what you are talking about. "

Yeah, when your client is your little brother LOL.
Sure you have your client and right after seeing eevee  wow he will magically triple the quotation or the bill you give it to him just because of eevee. what a joke 

[Anonymous]

I understand you care about Maya and you are passionate about it.

However, being emotionally attached to a software does not do anyone good.

If progress in development of the software is to be made the weak points should be admitted and improved upon.

For the record, I would like to state your statement on unreal and unity visual quality being 'faaaaar' better than eevee is inaccurate. I will explain.

Before eevee was created, blender foundation had some sort of agreement with unreal where they were permitted to study unreal's code to help implement eevee. Now, it wasn't a copy and paste kind of thing but they studied the code for perhaps a better understanding.

The visual quality of eevee in so many cases is as good as unreal  (for reasons previously stated) in some cases it is better, in other cases it is worse. However, what makes people excited is that it's still improving. 

Now, you may think that me stating eevee has a better visual quality in some cases is an overstatement, I felt the same way too, but I read that the developers at blender were moving towards visual quality at the cost of render times. In my opinion eevee is slower in those cases.

Where the real limitations of eevee  (compared to unreal) exist are things like ability to bake lighting.

Finally, Maya is still a great software, we don't always have to make things software A vs software B. Progress made in any major 3d software affects the entire 3d community and industry. It could stir up positive competition and development in competitors. This is what we should all hope for. For instance I don't think Nvidia graphics cards  (especially their new real-time ray tracing ones) would be this good without competition from AMD. I hope blender 2.8 has this positive effect on maya 2019 (and others as well). It will be for the good of the 3d community.

Thanks.

[aerys.bat]

This thread is off the hook! Hedge funds breaking up Autodesk, software wars, and some highly interesting rendering tutorials. Wow!

Here is a mundane technical guess about what's holding up Maya. The VFX Reference Platform spec specifies Qt 5.12 for CY2019. This is a massive change after 3 years of using a heavily modified version of Qt 5.6. But the first version of Qt 5.12 was only released this past Thursday! On top of that, there were 500+ more fixes that didn't make it in time for 5.12.0 release. This large number of known bugs means it's likely ADSK will want to wait another month for the 5.12.1 point release.

(If you don't know what Qt is, it's is a big software library for writing cross-platform applications. It's used by many CG applications that need to run on Windows, Mac, and Linux. A common first encounter with Qt is when scripting Python dialogs in Maya with the PySide and PyQt modules.)

My guess is that all the major features for Maya 2019 are finished, but QA testing and final release are being held up by Qt. Waiting for a somewhat obscure but important software library is not the most exciting theory (Autodesk killing Maya?!?!) but I'm pretty convinced that's what going on. This theory also comes with a prediction: If they wait for 5.12.1, then take an additional 4 weeks for final QA & release prep, that puts the release of Maya 2019 about 8 weeks from now, late January/early February.