Hi All,
I've been following the twister tutorial on 3dBuzz. I'm now trying to render it using i3d volumetric rendering (with VEX Meta Cloud). So far, I've tried both direct rendering of the particles of the tornado and copying metaballs to particles. Both methods seem to work ok for still images however, if I render out a sequence the result is quite bad. You don't really see the motion of the particles, just some noise generated by the shader. Looking at the docs it says that you can use per-point noise option of the VEX Meta Cloud shader to avoid swimming of the texture. The problem with this however is that the result looks quite poor as you can see the repetition or pattern of the noise. Any suggestions? Is it correct in this case to use the per-point attribute of the shader or is there a better approach?

thanks

About 2 weeks I attempted the same, except I only tried the “applying metaballs” technique. The render times were very heavy and I also didn't like the results. A better way is to assign sprites to the particles, watch the sprite VTM's, very good start

pld: I did the exact same thing about a year ago, and had the exact same result hehe . Good looks in a single frame, but nothing good in an animation, no matter how hard I tried. Since that time, I've talked to a few houdini people about i3d and I also talked to a developer who said that i3d isn't really meant for particle work. It's possible of course, but i3d is more meant for static volumes (I guess ?). I heard that odforce.net has some i3d shaders meant for particles, but I haven't looked. However, even with complex shaders, I get the sense that i3d cannot match the power or ease of afterburn for max or hypervoxels for lightwave, so if you've used either of these I think you'll still be unsatisfied with the final result.

I think it's probably about how you build it. I think i3d and particles are perfect to use together.

I think i really comes down to tweaking, getting the right noise and turbulence frequency and balancing that with metaball sizes.

It certainly isn't a very user friendly process, and from what I saw of hypervoxels a few years back, possibly a little feature light.

As is usal we're also a little light on good docs/tuts that allow new users to get the most out of it.

dude, jesta, if a developer is saying that it's not really meant for particles, guess what, it's not really meant for particles. Feature light? What features _does_ it have is more the question. Rendering static volumes you say? Yes, that is one feature, correct. But to me, it doesn't even look like it has your standard ramp controls. I don't doubt that you could get color mapped to a ramp, but what about size, density, noise settings? I'm not saying it's terrible, don't get me wrong, it's a hell of a lot better than nothing. Still, when comparing it to afterburn or hypervoxels, what _is_ there is the correct question. It's not a matter of rendering and tweaking to get the correct results if you can't control anything no matter how many tweaks you do.

ReggieFourmyle
dude, jesta, if a developer is saying that it's not really meant for particles, guess what, it's not really meant for particles. Feature light? What features _does_ it have is more the question. Rendering static volumes you say? Yes, that is one feature, correct. But to me, it doesn't even look like it has your standard ramp controls. I don't doubt that you could get color mapped to a ramp, but what about size, density, noise settings? I'm not saying it's terrible, don't get me wrong, it's a hell of a lot better than nothing. Still, when comparing it to afterburn or hypervoxels, what _is_ there is the correct question. It's not a matter of rendering and tweaking to get the correct results if you can't control anything no matter how many tweaks you do.

Houdini doesn't have any volumetric tools like afterburn or hypervoxel. It does have some very low level tools, (i3d, volume shaders, metaballs, etc), that allow you to build effects that are similar to afterburn or hypervoxel. The problem is the current setup is very “do it yourself” and requires that you have some technical people on site. As with a lot of Houdini, the main problem is that the application doesn't ship with very robust shaders and the samples are very simple. When you have a deadline looming you don't have the time to spend 4 weeks putting together a good toolset of shaders and assets to help create volumetric effects. Relying on the ones that ship with Houdini can lead to a lot of headaches.

Sometimes developers dont see the big picture, and even if it is not meant to be used one way, clever artists/monkeys could make it “meant for it”.
I think it is the beauty of Houdini. If you are ready to get your hands dirty and have an open mind, things can work.
In fact Houdini is a Philosophy!

Sorry Reg. Didn't realise you had the inside track on this one.

Works just fine for me and a whole other bunch of dudes that have done i3d with particles at places like DD and Framestore-CFC amongst others.

But I'm sure you know best.

jesta: DD uses i3d? lol, so how do you explain storm and voxelB? And what exactly works fine for you? Are you saying that you can control the things I asked about that are easy to control in ab and hv? Density, noise attributes, particle size? If you can, tell me how. If not, stop posting.

wolfwood: so i3d _can_ be used with particles? I mean, you can actually ramp things like particle point size, and density based on life or position or something? When I looked, it seemed to me that some things were controlled by the shader, but others were controlled by the renderer, and didn't give any other options otherwise (except as always with the coding method). I'm just asking here, I'm not criticizing. I know that most of houdini can be powerful with coders on site, but that can be said with anything, as I've stated before. Mental ray can render volumes too, but I don't see anyone flocking to xsi or maya to do volumetric particle work with it because there's no real interface for it. And if we're in the realm of $$$ and time, I'd just buy renderplus and get air or prman.

ReggieFourmyle
wolfwood: so i3d _can_ be used with particles? I mean, you can actually ramp things like particle point size, and density based on life or position or something?

I apologize for the confusion, once again I wasn't clear.

Is I3D Houdini's version of voxelbitch, Afterburner, Hypervoxel? No. In fact Houdini doesn't have any tools which compare.

I3D is simply a 3D file format for saving out 3D data. That's all it is and nothing more. To add data to I3D you can use a variety of methods, including metaballs, particles or geometry. You can save any arbitrary data into the file that you like, rest positions, fields, density, color, age, etc.

Once you have your I3D file you can do anything you like with it. There are VEX functions which allow you to query/sample the volume. From within your shader you can query any data that you added to the I3D file, which includes the attributes listed above.

The point I'm trying to make is that unlike other 3D apps Houdini is the middle ground inbetween something like 3DSMax and completely custom inhouse software. In 3DSMax you have a ton of tools that are designed to produce pretty pictures with a fast turn around time, but is as only flexible as the plugin makers allow you to be. At the other extreme is completely custom inhouse software like VoxelBitch. Houdini sits in the middle. Cranking out the pretty pictures takes longer but allows you to be much more flexible without the need for a large R&D team.

What Houdini needs more than anything is wrappers around its default tools that are more artist friendly. The Auto-Rig assets are an example of this.

wolfwood: Thank you for clearing that up, you've answered basically my and pld's questions. And on that note, I'd love for you to relay this to the sesi developers. I was just talking to a developer the other day and I was trying to get him to realize how great the specialized (aka non-prodedural) additions are to houdini. It's great that houdini wants to be a fully procedural application, but the vfx world requires customizations for common problems, like the character tools. This developer didn't seem to realize the importance of specialized toolsets for common problems, but rather assumed that it was because I have no production experience with houdini. So it looks like it's up to the seasoned peeps to yell at sesi for me. I can only wish that there will be more additions included, or even optional in houdini over the years hehe.

Hi Reggie,

Sometimes, a “very specialized” toolset doesn't quite fit into the general scheme of Houdini where every operator should be an “atom” that can be assembled into a larger tool set. As for the non-procedural modeling tools goes, it may be nice to see Houdini have some of those functionality goes, but that may also imply that Houdini should become more and more like all the other packages out there. For SESI to develop the same type of non-procedural tool set that all other packages have done would almost be like reinventing the same wheel at SESI's resource.

About more than a year ago I've had the pleasure to work with one of the brilliant developer from SESI. At the time, I ask almost the same kind of question to him. One thing he mentioned to me is that (rephrased from memory) generally when there is a request for a tool, SESI likes to make it general, robust, and fit into the procedural schemetics of Houdini. When it's a functionality/feature that's similiar to other package, SESI likes to think of a better way to implement it into Houdini as oppose to being copy-cat. Sometimes, having a super-specialized tool might just not fit in super well. And sometimes, it's just priorities and resources to get something out.

Remember tho, that SESI is very small compared to all other major 3D software companies out there. But SESI has really brilliant developers/engineers there.

Hi Alex, (sorry for the long post)

I think I know what you're saying, and in the case of my modeling toolset suggestions, I can understand that that toolset may not be in the best interest of sesi. However, there is no way that that is true in this case. If houdini has one specialty, what is it? particle effects right? So why not have a full volumetric particle tool? They already have i3d in place, ready and working, so it can't be _that_ hard to make it easier and more powerful for vfx. Mantra already renders volumes perfectly, and there's already an interface for static volumes, there's just nothing for particle based volumes. In other cases, I can understand sesi not wanting to develop xxx toolset for this or that, but in the case of houdini's most powerful feature, I think it would be beneficial at the very least. And again, I can't imagine that it could be very hard seeing as there is already an interface for most of what's needed.

I understand that specialized tools don't always fit in sesi's plan of having a procedural 3D application, but there's a problem with this: Specialized tools are necessary for 3D production. A good example of this is the character toolset, which, from what I heard through the grapevine, is making houdini a hell of a lot more popular than DOPs or possibly even POPs, as it has made way for entire movies to be made with houdini (again, this is just what I heard, could be wrong). So if anything, I think a specialized toolset in this case, is what sesi wants, especially when it will increase the power if its main feature, particles.

Sometimes, having a super-specialized tool might just not fit in super well… Remember tho, that SESI is very small compared to all other major 3D software companies out there. But SESI has really brilliant developers/engineers there.

On this last note, I'd just like to say a few things -
Of all of the 3D programs out there, houdini is by far the best candidate to integrate specialized toolsets because it is so perfectly organized. If you don't need xxx tool, you never have to even see it at all. This is clear with the character tools: they're not going to get in the way of any particle people. That being said, I think if anything, houdini needs more specialized tools than other applications. Especially because of the “niche” market that houdini is in. It is clearly a program that does not dominate the 3d marketplace, therefore, it's chances of being purchased and integrated into an effects house rely solely on the strength of its specialties. Obviously, many 3D houses that currently use houdini would not use it if it weren't specialized for particle effects.

And one last note about sesi being brilliant -
I truly hope that you're right about sesi's brilliance. But it scares me that you describe them as such. My brother is brilliant, he has a 170+ IQ. He's also completely psychotic and cannot communicate like a normal human being. Guess what, he's a hacker too, sound familiar? If sesi needs anything, it's not brilliance, it's communication and socializing (so to speak) with the rest of the 3D world. Someone at sesi had the _truly_ brilliant idea to bring houdini into the mainstream, and I am with that person 110%. I think houdini kicks ass… But in order to fully walk the mainstream walk, houdini must make some changes. My modeling toolset suggestions were very good ones in the “mainstream” sense. They may not be popular with houdini users (aka most people reading hehe), but just about every non-houdini user would love what I asked for, and that would only help houdini as a whole. This is exactly what I'm doing now… I don't know why I waste so much time with this crap though, again, I'm thinking too much about the future of a program that I don't even own… I really should focus on getting a Fing job <sigh> =/

Seems like there's a bit of misunderstanding here.

While true that Houdini is a “specialty”, what I meant is, procedural paradigm. When you have a single SOP that does all the model operations like that of ZBrush/Max/Maya/XSI/etc, this what I meant by “speciality” that I speak about earlier – something that doesn't quite fit into Houdini's procedural pipeline. To have such single Model SOP that does the like of all other package may be great, I don't know. But this seems to break the whole non-destructive pipeline and much bigger HIP files and such. Again, I don't know. It's just that from your posts in the past, it sounded like you wanted Houdini to completely break its existing proceduralism. While other non-Houdini artists likes what you said, well, that may be true and something you posted may be valid. But then again, the Houdini artists/TD thinks much differently than the non-Houdini artist.

Now, don't get me wrong, I am all for Houdini to be even better and such.

Well, I have no doubt of your brother's brilliance. I personally have an IQ like that of a snail. Hence I don't understand stuff. I am turtle slow, to top it off. So, after seeing what the SESI developers did, I am amazed.