rant and fan boy-ing

[kleer001]

So, I've been tasked with reverse engineering an effect from a sister office. That effect was done in 3D Max with Krakatoa. I kinda got it, I think. Not to slag off a compatriot's work, but the thing is the original setup is kinda jenky. So jenky that the comping had to be so over the top that basically none of the dynamics are visible under the distortion necessary to cover up the weird jittery effects and random blobs. The element is originally a beautiful multicolored smoky thing. In the final production comp? It looks like a grey smeary mess.

 

I figure what I did will work just fine, but I'm preparing for some notes, sigh...

 

Here's the fun bit though. I had to open up their original scenes in Max. And 3D Max looks like garbage. Nothing is on the surface, everything is under a menu or three layers down in a single undifferentiated scrolling window. Now, I'm probably just too used to Houdini and its layout. But here's the objective bit. Houdini is more transparent, the interface is shallower and hence easier to navigate. You can open up a Houdini scene and using the default desktop (and barring obfuscation from the artist) everything (or at least waaay more than Max) is immediately visible. Maybe you have to dive down a little to get at the guts of an effect. Maybe some scripting is hidden somewhere. But the thing is in Max nothing is immediately visible. Want to know the transforms of an object? That's down in the bottom, next to the timeline. Want to change a light's settings? That's over in a tab (that's grouped with object creation for some reason) and under a box you have to open. Want that information in Houdini? It's all in one place. No interface scattering, no extra clicking, object attributes/parameters are all in one place. And I'm sure that's just scratching the surface.

 

 

/drunken-fan-boy-rant off

[Guest mantragora]

Any Autodesk software have this garbage problem. Any non Autodesk software doesn't have this garbage problem. Simple as that.

It's even bigger problem for someone that started 3D journey from non-Autodesk software. At some point you jump into it and all you want to is kill somebody when you have to use this shit with menus under menus under another men...

And it's even bigger problem for people that start they journey with 3D today because there is 50% chance (or even more I say) that they start from Modo, Houdini, Blender etc. and you will have really hard time explaining them why they should use those piece of shit programs that requires ton of plugins to have the same functionality.

I think it's something in the water they drink at Autodesk.

Edited January 14, 2014 by mantragora

[pclaes]

I feel where you are coming from .

 

I remember matching some of Scanlines proprietary point/volume rendering effects for the fluid highways in Cloud Atlas. That had some interesting layers and comp effects to it. I actually had a great time reverse engineering it. At the time the lookdev was between Scanline, ILM and Method, eventually they ended up really liking the stuff Method did so we started leading the lookdev and Scanline and ILM had to match our stuff . But it was pretty tricky to match Scanlines stuff on a pure reverse engineering type basis in the first place - it was great that they provided the nuke script and the raw elements. I found it a very creative and technical challenge - it ended up being a combination of shader work and volumetrics.

 

Good luck with it!

[Guest tar]

Horses for courses as they say... whilst one of things Houdini concentrated on is a clean workflow other software concentrated on features such as modelling, faster MoCap/bones in the viewport etc.  

[3__]

Bottom line is weather you can illustrate something on-time, on-spec. I've found that for most things max (& plugins) will deliver, even in the hands of a relatively green artist. It's those special cases that really burn you though - where something you're trying to do simply isn't supported by the software.  I suspect you haven't run into such a case as there are really nice looking smoky krakatoa renders all over the net, so you can relax: its probably just a people-problem.

[jordibares]

 

Any Autodesk software have this garbage problem. Any non Autodesk software doesn't have this garbage problem. Simple as that.

It's even bigger problem for someone that started 3D journey from non-Autodesk software. At some point you jump into it and all you want to is kill somebody when you have to use this shit with menus under menus under another men...
And it's even bigger problem for people that start they journey with 3D today because there is 50% chance (or even more I say) that they start from Modo, Houdini, Blender etc. and you will have really hard time explaining them why they should use those piece of shit programs that requires ton of plugins to have the same functionality.


I think it's something in the water they drink at Autodesk.

 

 

Regarding the interface comment (workflow really).

 

If I remember well none of the 3D applications Autodesk sell are Autodesk born products… Maya comes from Alias, Max from the old Yost group, XSI (now rebranded Softimage) come from Softimage, Motion builder from Kaydara, Mudbox from Weta internal dev… Even Flame from Discreet Logic...

 

The generalisation is pretty flawed and it is specially flawed in the sense that Softimage UI and workflow is imho the best there is, unfortunately Autodesk PR and Marketing don't seem to recognise it but if you open it you will see information very well exposed and the menu drilling that you see in both Max and Maya is not present.

 

The same happens with Motion Builder, it is a very good interface and workflow...

 

my 2 cents.

Edited January 18, 2014 by jordibares

[jordibares]

By the way, it is true houdini interface and workflow is very elegant and clean, the reason I mention Softimage is the best it is because it is viewport centric and all the behaviour is very consistent (not perfect though), from selection modes to how tools behave it is something you have to run a bit with it to get what I mean.

 

Hope this clarifies my generalisation

 

;-)

 

jb

Edited January 18, 2014 by jordibares

[Fabian]

I have to agree with Jordi. Having used both xsi and houdini for a long time I would say that for scene manipulation and construction (modelling, placing, moving, rotating, aligning(!), etc.) houdini does not hold a candle to xsi. The tools, layout, short-cuts and key combos make for a very elegant and incredibly fast workflow in the viewport. The same goes for animating a character even though I haven't done that in a while now.

 

Other areas of xsi fall short in comparison but I wish sidefx would take the time to understand what makes something like xsi so damn nice to use around the viewport and take it to heart for their future. Heaven knows they have gotten enough RFE's over the years on that.

 

I've seen training/demo videos for most major apps by great people over the years and pretty much every time end up comparing the workflow to xsi and I'm thinking "how can they put up with it?", max being the worst of the lot and modo probably the best. Although I think you need some old 'waver DNA in you to like modo. "Drop the tool" *shudders*

 

½

 

-F

 

Edit: I actually wrote a custom scene exporter once from xsi to houdini that would carry object placement/basic textures&shading/etc. and rebuild the scene in houdini so I didn't have to do that sort of work in houdini. I'm pretty sure even with the scripting it was faster that way.

Edited January 19, 2014 by Fabian

[Guest tar]

Do you have a link that show the speedy bits of Xsi so we can compare to Houdini. Thanks!

[Fabian]

(sticky keys and super-modes, very useful and imho one of the key ingredients that make moving around to different tasks feel so smooth)

 

(selecting)

 

(submenues with memory, this is surprisingly useful)

http://www.youtube.com/watch?v=Ita31Ju4Yes

 

(selections, nothing to impress a modo user but I wish some of this was in houdini, the poly brush selection tool is atrocious (the video only covers some basics))

http://www.youtube.com/watch?v=EbEAvXjU8y4

 

(snapping, temporary pivots, similar to houdini's ' mode but much smoother to use)

 

<rant>

 

One of the things is also that a lot of the modelling tools like bevelling, poly sewing, splitting, etc. are a lot faster to use and more robust than the equivalent sops. Having to fight sops like the cookie, poly bevel or font sop really undermine the whole procedural workflow. For example the font sop, yeah it's there but it requires black magic and manual fixing if you feed it anything slightly more complex and open letters set in OCR-A. And getting a simple eps logo in for a pack-shot type of stuff will have people refer you to the trace sop (!!!).

 

In xsi booleans also mess up but then there are the tools to quickly patch the problem areas up.

 

Another thing is that using Space+1/2/3/4/5 resets the camera each and every time and navigating back to where you were is a gigantic waste of time. This then leaves you with the quad view and the three step process of minimize current/select next view/maximize next view. This is never an issue when looking through the ready tracked and locked shot cam and making the explosion look good by working on your network and parameters but when building stuff in the viewport it gets really tedious really quickly.

 

There are rfe's for all this stuff so +1 at your leisure.

 

I'm sure touching gnarly old code to polish these inglorious things up is not a lot fun for the AlphaGeek coders but that's what you have interns for :-p

 

I love houdini, the toolset it offers and especially the people behind it. And I would like nothing better than to stop paying autodesk my protection of my damn near abandoned investment money but unfortunately I think these basic, boring everyday things need to be addressed comprehensively before it's viable for me to do so. Given SESI's finite resources though my hopes for this happening are slim as long as the bulk of the userbase seems to be asking for more efficient ways to blow the architectural wonders of the galaxy to smithereens :-p (and yes I've tried and just can't get myself to like Modo, C4D or LW, no offence meant to any fans)

 

</rant>

 

-Fabian

[lisux]

Thanks for the links Fabian.

As more studios been to use Houdini as a general lighting solutions these "simple" but common practices needs to

be improved.

SESI needs to think how to make Houdini easier and faster from the viewport, otherwise it is pain for people coming from other platforms.

[kleer001]

Horses for courses as they say... whilst one of things Houdini concentrated on is a clean workflow other software concentrated on features such as modelling, faster MoCap/bones in the viewport etc.  

 

Yea man, I'd only rig/animate/model/track/matchmove anymore if you put a gun to my head.

Leave me in my procedural particles/fields/volumes/maths closet please.

[Guest tar]

If Sesi implemented Xsi selection and viewport handling how much more productive would we be do you think?  Would it help only modelling of would it also help fx work?

[jordibares]

 

IMHO I would certainly be way more productive, probably you will start to use the viewport more, every time you need to navigate your scene and select things (which imho is where you really spend time) will benefit.

 

If you are going through rigging, animation, previz, layout… i will be sooooo much more productive.

 

Edited January 21, 2014 by jordibares

[sebkaine]

The hard thing to solve is that everybody has his own idea of what is the best UI or way to work. I have found some people that are pure Max fan , other that only want to model in lightwave/ZB other like you guys who really love XSI. So one thing to keep in mind is that expectation beetween user about what is the perfect workflow can be pretty wide ...

 

I personnally love Maya UI the alt+nav + qwerty CTRL + 12345 layout + custom marking menu allow you to control all the software without having to move your hand from the keyboard , i really don't like what some lazy people has done to maya , but i am still a big fan of the initial philosophy of high ability to customize the tool and 10-7 years ago maya was a dream to work with ...

 

I think SESI should focus on a flexible way to customize the UI , where you can choose the way you work by keeping in mind that in terms of UI less is often more.

- how to do a maximum number of operation without moving your hand from the keyboards

- how to access any feature you like with a minimum number of click

- how much flexibility you have to choose the way you work and you organize your tools (shelves / tab / marking menu / etc ... )

 

Trying to replicate the workflow of app A or B or C , is not as efficient as thinking as a very flexible way to organize tools.

[Guest tar]

So what is actually missing or could be improved from Houdini then. Does it needs more keyboard shortcuts to a variety of the tools?

 

What's the feedback on the brush selection of polygons? Is it slow or needs more features compared to other software. 

I'm running OsX version which in Gl2 so I can't test it properly at the moment.

 

Edit: This was for Fabian's post "(selections, nothing to impress a modo user but I wish some of this was in houdini, the poly brush selection tool is atrocious (the video only covers some basics))"

 

Edit2: And also Sesi should improve some of the Sops like font, polysplit etc and fix the viewport loosing it's memory when you switch views - single, four, two, three etc

Edited January 21, 2014 by tar

[jordibares]

 

The hard thing to solve is that everybody has his own idea of what is the best UI or way to work. I have found some people that are pure Max fan , other that only want to model in lightwave/ZB other like you guys who really love XSI. So one thing to keep in mind is that expectation beetween user about what is the perfect workflow can be pretty wide ...

 

I personnally love Maya UI the alt+nav + qwerty CTRL + 12345 layout + custom marking menu allow you to control all the software without having to move your hand from the keyboard , i really don't like what some lazy people has done to maya , but i am still a big fan of the initial philosophy of high ability to customize the tool and 10-7 years ago maya was a dream to work with ...

 

I think SESI should focus on a flexible way to customize the UI , where you can choose the way you work by keeping in mind that in terms of UI less is often more.

- how to do a maximum number of operation without moving your hand from the keyboards

- how to access any feature you like with a minimum number of click

- how much flexibility you have to choose the way you work and you organize your tools (shelves / tab / marking menu / etc ... )

 

Trying to replicate the workflow of app A or B or C , is not as efficient as thinking as a very flexible way to organize tools.

 

I am of the opinion that Sidefx should be brave to change whilst keeping their philosophy intact, and that means that proceduralism should be on the top of the list always.

 

Selectability should be fully functioning, right now it is not (IMHO). Some of the ideas from Softimage like Supra mode, sticky keys is all compatible with any philosophy and it is not about mimicking shortcuts, but the behaviour and workflow the system forces you to use.

 

A good example of workflow enhancements would be in VOPs, you connect a float to a vector, houdini should add automatically the conversion node necessary to make that connection valid instead of the current approach. Still procedural but making the artists life easier.

 

I guess we could analyse till death but I think I don't know enough to fully discuss it, I rather point out a few things, give them some examples and how it is used in other packages and I am sure they will interpret it the right way.

Edited January 21, 2014 by jordibares

[Guest mantragora]

First time I run Houdini it was version 6 something. It was at the time when I was looking at each application trying to decide which one is for me. It was before shelf tools, when there was still start scene opening when you turned Houdini on.

Houdini was the ONLY application in which I couldn't figure out how to create ball . And that's how it started for me. All the other apps looked like copies just with different name. With ton of icons and menus.

Houdini was so clean back then. At this point in time the only application that still looks almost as clean like Houdini back then is Fusion (and maybe Unity). Later SESI started to make it a little Maya like and it started to confuse people. Thankfully they stopped. Don't try to make it work like other apps and complicate things. Cleaner is better.

There are some things in other apps that I would like to see in Houdini but Maya like customization is the last one I would like to see.

[Guest tar]

 

I guess we could analyse till death but I think I don't know enough to fully discuss it, I rather point out a few things, give them some examples and how it is used in other packages and I am sure they will interpret it the right way.

 

 

Thanks Jordi, very cool feedback!

 

I've always found coders like things to be dissected at a low level as that's the level that helps them to code it

 

Edit: Better put, it helps bridge the chasm between artist and programmer! As we usually want something that has unexpected implications for other parts of the software!

Edited January 21, 2014 by tar

[magneto]

 

Edit2: fix the viewport loosing it's memory when you switch views - single, four, two, three etc

 

+1.

 

Thanks Jordi, very cool feedback!

 

I've always found coders like things to be dissected at a low level as that's the level that helps them to code it

 

True that

 

Also not sure if it's me but if you have a large object like 20x1x1 box and you select a point and zoom to that point, orbitting the view uses the world origin, but not the current selection center, which prevents to orbit around the selected points (primitives, etc) in one swoop but rather in multiple small iterations of orbit and pan. In Max they have 3 settings for controlling the viewport Arc Rotate center and I never had a situation where I didn't want the current selection center ever. Current selection could be the object, the subselection, etc depending on which level you are in. Autodesk made this option the default a few years ago due to popular request.

 

It would be great if SESI also made the viewport orbit to automatically use the center of the current selection/subselection depending on the current level.

 

Having a separate undo buffer for the viewport would also be really useful. Max has this and it allows you to not worry about losing your view when you quickly need to change your view for selection, etc as you can go back to your previous views very easily.

 

Also homing the view shouldn't change the orientation of the view, but the zoom level. Having the option to reset the orientation is useful, but shouldn't be the default. Fortunately Houdini has this feature which is what I am using