If I was ADSK and have $100 to spend for SI

[satyajit]

Here's what I'd do...

1. ADSK has positioned SI as a high end tool specialized for FX work. While we know, SI is not the king of VFX. Max, Houdini, Maya holds the place. ICE is particularly strong with some types of FX and not really cutting edge on others. Still now it misses a volume shader better than Fume FX. Fume FX single handedly kept users stuck in Max. So, really, if you want SI to be the king I'd possibly use 50$ from my dev budget on developing that shader or would buy up Sitni Sati.

2. The Destruction part of the FX mechanism of SI is still weird. Look at Volume breaker and Thinking Particles. Not only they can break up meshes into good looking pieces and fast, but the pieces carry the material infos nicely and also they are themselves deform-able. This really matters a lot. There's still no way of doing that in SI. Lagoa is not really breaking up meshes. It's creating particles from their volume, operate on them and then you can mesh them up again. Do you think the mesh is going to look the same after this amount of transformation? So, where there's a need to break up mesh into dust, ICE is chosen. But that's a smaller chunk of the grand total of VFX job. So, I'd spend 25$ from my dev budget into breaking things dynamically and beautifully.

3. The remaining 25$ of my budget would go into replace the existing system by new rigging system. I cutting edge one, modular and everything. With flash based wizards and really jazzy look. It's not that hard, we can pick the developers who've already progressed in doing something like that. I can go for Scripted OPs or ICE, or can give an option of that in the wizard. But primarily it should generate a rig that would include flash based interface, squash stretch, face rig, multiple rigged characters, shadow rig and also the rigs should be scalable. All the third party rigs are still missing one thing or the other. It should be better than GEAR and CAT at least. Just let the Animation Mentor Rigs be our inspiration. So, when there's a need for a quickie and stable rig for all purposes, SI should be the answer.

[Eugen]

Isn't most of what you say is available as a plugin?

Shattering: Exocortex ImplosiaFX.
http://exocortex.com/products/implosia

Rigging: you mentioned GEAR. It's open source, so including it into Softimage by AD would be a problem license-wise, but I'd say the SI devs should concentrate on other things than re-invent that wheel anyway. GEAR can be developed further by 3rd parties.

Flash based interface? PyQt would be the better choice.
https://github.com/caron/PyQtForSoftimage

Volume shader: what renderer are you talking about?
mental ray/Arnold? See Holger Schönberger's page
http://www.binaryalchemy.de/
(For VRay, 3Delight, Redshift, Octane etc. I'm not sure.)


The matter of fact is: Softimage works, really good, and just because of that it should be continued.

My 'clever strategic plan' for Softimage would be:
I'd put those 100$ into: streamlining what's there, fixing bugs, making it as open a platform as possible, so 3rd parties get even better options to extend it.

Most of the really interesting and innovative stuff for Softimage comes from outside of AD anyway these days. What's wrong with that?
BUT - this should be consciously supported, and more publicly, too. It would help if AD dropped this attitude of secrecy, played with open cards and got into a stronger dialog with the community about the next features.
I'm talking about some more 'democratic' process here. It's happing already to some degree, mostly in cooperation with bigger FX houses I believe, but should be
intensified.
We are talking in places like this already, too, and in the beta, but it's still all too vague, and no guarantees are made from AD side.

They have the advantage of an enthusiastic user base - why not calculate with that?

As has been said before - if no 3rd party dev knows for sure if his investment will pay off, why take the risk? A clear decision and statement that Soft will be continued for the next, say, 5 years by AD, would help tremendously.

[satyajit]

Those were really the suggestions occurred to me. Of course there can be much better suggestions. But then again I'll validify my points.

1. I'm sorry to say, BE volume shaders while are cool is nowhere comparable to FumeFX, the tendril based look, speed, shading etc. simple way ahead and industry standard. BE and Emfluid combined are still not there.

2. I'd stress the way TP, volumebreaker works other than Implosia or other fracturing methods. For example, I repeat, the fractured elements of TP interacts with each other not only with rigid body hull but they are deformable themselves. This adds to the extra punch. Also, it simply needs better algorithms of stable collision of active and fractured meshes sitting in close proximity. If you compare these two using the same scene you will know what is the difference and it really matters.

3. There's a difference between company based rig and third party open sourced rig. Users can really run after ADSK support during a rig problem and if it breaks down. It also gives value to the gold support. Using PyQT is a better solution, agreed. But there's a dear need to replace the existing wizard and system.

I didn't suggested the work on opening up the API and wait for third party devs, though a nice suggestion, it looked to me much more costly as it needs to get down to core and change things there. Though it'd be wonderful, I'm thinking for short run success first. XSI is already a sufferer of long term dev from its birth. So, I don't want to get into 3 years cycle. Thinking short term achievable success first. Being realistic.

[Hirazi Blue]

I really don't think you have any idea where $100 will actually get you in this day and age, but anyway.
If Autodesk were to spend the smallest amount of money on Softimage for maximum effect, they should simply hire a (better) writer to rewrite the whole marketing/promotional blurb. Better marketing/promotion potentially means more sold licenses. More sold licenses makes it more attractive for Autodesk itself to spend some more development time on it. Win-win?
;)

[Eugen]

We're running in circles... Maya is the chosen flagship.

We don't know the numbers, so we can't actually talk much. I just hope AD does not think Softimage only 'cost' them now...

If this 35mio investment for this live-cell treatment for Maya was a good investment is a hard question to answer even for themselves I guess.
Yet if they had that kind of pocket money in the first place, chances are good they are thinking big anyway... I hope.

[Eugen]

Those were really the suggestions occurred to me. Of course there can be much better suggestions. But then again I'll validify my points.

1. I'm sorry to say, BE volume shaders while are cool is nowhere comparable to FumeFX, the tendril based look, speed, shading etc. simple way ahead and industry standard. BE and Emfluid combined are still not there.

2. I'd stress the way TP, volumebreaker works other than Implosia or other fracturing methods. For example, I repeat, the fractured elements of TP interacts with each other not only with rigid body hull but they are deformable themselves. This adds to the extra punch. Also, it simply needs better algorithms of stable collision of active and fractured meshes sitting in close proximity. If you compare these two using the same scene you will know what is the difference and it really matters.

3. There's a difference between company based rig and third party open sourced rig. Users can really run after ADSK support during a rig problem and if it breaks down. It also gives value to the gold support. Using PyQT is a better solution, agreed. But there's a dear need to replace the existing wizard and system.

I didn't suggested the work on opening up the API and wait for third party devs, though a nice suggestion, it looked to me much more costly as it needs to get down to core and change things there. Though it'd be wonderful, I'm thinking for short run success first. XSI is already a sufferer of long term dev from its birth. So, I don't want to get into 3 years cycle. Thinking short term achievable success first. Being realistic.

1. Which renderer?

2. Depends on the dynamics engine. Does anyone know which one they are using?

3. Rigging might not be a good USP, because Maya has a very good rigging toolset already.

4. "Get down to the core and change things there" - to be more precise, it means exposing more internals to the SDK. That's a different thing.
I can't estimate how costly that would be, of course.

Thinking of short term success - sounds good, but what would give Softimage a better chance for ANY kind of success than the commitment to continue it for the years to come, not just for another year, then let's see, then maybe another year, then let's see again... it's that kind of attitude that creates uncertainty.

[Mathaeus]

well, both Volume breaker and Fume FX are plugins for Max, you have to buy them separately. Prices are around 500 and 900 bucks, for members of huge Max community. For small SI community, possible price could go to about ten times more, I'm afraid. I really don't know how would you get all this only for extra 100.

Getting the tool to do believable look, that's a lot of time for testing, and possible re writing from scratch. There is story, somehow frequent in SI or Maya field, "this or that Max plugin has only good defaults" - but that's only story, perhaps good for TDs to find job somewhere, and that's all... As it mentioned, Lagoa does not work in the same way as Volume breaker.

Rig, and a lot of horrible defaults in SI... imho there is a philosophy of SI or Maya people, both developers and important users, that work on all these "small things" does not belong to developers. Even for price of total disasters when something actually is not "small thing" like well known "it's all about shaders". Maya seems to be champion in that "arbitrarily half done" approach, but SI is not far behind.

Long story short, if you want the mentioned level of quality, stay away of anything that's even close to philosophy of SI or Maya creators. They just can't do that.

[Eugen]

Long story short, if you want the mentioned level of quality, stay away of anything that's even close to philosophy of SI or Maya creators. They just can't do that.

The old problem - developers easily drift too far off in their own world and have no idea about production needs any more.
Again -> As they say 'A cobbler should stick to his last', Softimage devs should concentrate on 'core' features and the 'big picture'.
There might be central parts of the architecture that are for their eyes only, for legal reasons, or because they simply need not concern anyone else. Similar to Fabric Engine - it's open source, except the multi-threading core.

If the start creating production tools like CrowdFX, they should do it either in close cooperation with production houses, or just leave it be.


Presets - good point. Should be brushed up and opened up more, also. Should be possible to create user-defined presets that override, NOT destroy, the factory ones, and can easily be switched on or off, and migrated to other installations easily.

[satyajit]

Those were really the suggestions occurred to me. Of course there can be much better suggestions. But then again I'll validify my points.

1. I'm sorry to say, BE volume shaders while are cool is nowhere comparable to FumeFX, the tendril based look, speed, shading etc. simple way ahead and industry standard. BE and Emfluid combined are still not there.

2. I'd stress the way TP, volumebreaker works other than Implosia or other fracturing methods. For example, I repeat, the fractured elements of TP interacts with each other not only with rigid body hull but they are deformable themselves. This adds to the extra punch. Also, it simply needs better algorithms of stable collision of active and fractured meshes sitting in close proximity. If you compare these two using the same scene you will know what is the difference and it really matters.

3. There's a difference between company based rig and third party open sourced rig. Users can really run after ADSK support during a rig problem and if it breaks down. It also gives value to the gold support. Using PyQT is a better solution, agreed. But there's a dear need to replace the existing wizard and system.

I didn't suggested the work on opening up the API and wait for third party devs, though a nice suggestion, it looked to me much more costly as it needs to get down to core and change things there. Though it'd be wonderful, I'm thinking for short run success first. XSI is already a sufferer of long term dev from its birth. So, I don't want to get into 3 years cycle. Thinking short term achievable success first. Being realistic.

1. Which renderer?

2. Depends on the dynamics engine. Does anyone know which one they are using?

3. Rigging might not be a good USP, because Maya has a very good rigging toolset already.

4. "Get down to the core and change things there" - to be more precise, it means exposing more internals to the SDK. That's a different thing.
I can't estimate how costly that would be, of course.

Thinking of short term success - sounds good, but what would give Softimage a better chance for ANY kind of success than the commitment to continue it for the years to come, not just for another year, then let's see, then maybe another year, then let's see again... it's that kind of attitude that creates uncertainty.

1. Well, FumeFX uses it's own renderer (Fusionworks) and I believe Mental Ray and VRay works with it too upto some degree. So, MR can be an option.

2. TP has it's own dynamics engine alongside Bullet Physics. You can use both of them alongside, each has it's strength areas.

3. Rigging in Maya is cool, but SI handles rigging very differently and hence may be a good choice in many cases. Especially with the stack based workflow of SI, it will have it's place when there's a need for going back and forth. Granted, people are not going to jump ship on that, but if presented in a cool way it will definitely have it's place. Quickie, stable, extendable through ICE and iterative production friendly. Sounds like a proposition.

Many still prefers CAT over Maya Rigs.

4. Agreed again. Opening up more internals would definitely attract more 3rd party devs.

I'm saying, if it's a goal to position SI in the market, a flexible VFX platform can be viable option. But it has to be complete, not with loose ends. It seems that SI devs are detached from the industry and how it's competition is working. So it's going half the way...

Just another note, look at development scope of the FX tree, another reason why it can secure the VFX spot.

[Eugen]

To sum up, after so much has been said... the single most important thing that will secure continuity of Softimage is... *drum roll*...
Autodesk's deliberate decision to do so!

This might be a gut decision (which are often the best), even a risk, but it has to be a CLEAR decision! It's up to AD to make up their mind, and make another OFFICIAL statement, including the words bright and future, and a number of years. That ALONE will set things in motion!
There are more than enough options and suggestions on the table now.

[Mr.Core]

this thread is another proof that only thing SI is really needing is flexible and bugless SDK - the rest of work will do third parties, since ADSK dont care on major innovations

[Eugen]

+1000!

Did we consider the case that Softimage is simply 'mature'?
That's not a bad thing. It delivers what it was laid out in the first place, and does it good (mostly) - which other apps can say this of themselves?
Now that production became more complex and workflows (have to) include various other tools, maybe it's even not necessary any more for AD to be 'innovative', in a sense that many new tools should be introduced, like sculpting etc.

Try to cram in as much as you can all the time, and you will suffer the same fate as all others who do so - run into the burden of support an ever growing code base.
There's enough stuff in Soft already that needs grooming.
The days of the big über-applications might be over... which does not mean they have to be cancelled - just consolidated.

-> focus on 'core' (in the truest sense of the word) competences... the central one being a scene assemblage stage and a framework for any tools needed, for users, by users.

[luceric]

the "100 dollar test" is a product management thing that autodesk does on client visits. Discussions are opened and suggestions are put on the board. then the client is asked how much they would actually pay, out of 100$, for these. The point is that there are limited dollars, just like there is limited time. What usually happens is that there are lots of ideas, but the client would only actually fund 3 or 4 of them. For example, for softimage you could have a list that includes feature requests like "add 3D paint" and "retopology". Then in the second part of the exercise, the client actually put 40$ on "fixing reference model stability" and these two drop get zero dollar. Eventually the guy that suggested retopo might say something like "ha well, we use topogun already" while his colleagues nod. I'm not saying I'm a fan of this method, just explaining for the people who might not know why "100$".

[Bellsey]

I've done alot of these $100 type meetings and I think they're very productive, but there's pluses and minuses though.
Ultimately you ask anyone what they want, and people want everything, and then some, plus the moon on a stick.

So for me, it gets the user to really think about what they really want, when they're faced with stark choices. It can often produce some interesting results, when forced to choose. You can have this or this, but not both. This is what I was kinda suggesting (maybe badly) in the other thread about choosing a possible future direction.

[Eugen]

For Softimage, is it still 100$ or 50$ in the meantime?

[luceric]

nice one.
anyway the 100$ doesn't work in the format of this thread. it works in two steps: first, everyone lists want they think should be done. then, everyone agrees how to spend the 100$, therefore forcing people to make choices.
if everyone has their own 100$, it's meaningless.