Hey, So far houdini is really sharp. I have a couple of questions that maybe you guys could help with.

-Manual, is there one? Or is the start-up videos and the help on each node everything there is?

-Poly modelling. It seems like I can almost never get the selection for the next operation rather than editing the current operation.

-Particle size. The ‘balls’ that particles start out by default as are ok, and I can turn them into points, but do they have a size parameter, and if so can you then use that to make them collide with eachother based on that size…

-Render speed. My experiments have led me to believe mantra is pretty slow, at least in comparison to modo. How do you turn on ‘gi’ like in other apps. Grey environment, gi on, nice overcast look….

I'm sure I'll have more questions later, but those would be a nice place to start. I'm coming from a LW/modo/XSI type background, so the whole super procedural thing is a little alien to me so far, but fun.

thanks.

Michael Blackbourn
-visual effects artist
The Embassy Visual Effects

-Manual, is there one? Or is the start-up videos and the help on each node everything there is?

There are plenty of videos available for learning in the “Learning” section of the website. There's also a lot more in the help docs than just the help for each node. But to answer your question. There is no printed manual that comes with Houdini.

-Poly modelling. It seems like I can almost never get the selection for the next operation rather than editing the current operation.

If you press the “s” key or click the selection icon at the left of the viewport, you shouldn't have any difficulty with selecting things so that you can then use the shelf/tab menu to perform your next operation. Note that if you're laying down nodes in the network editor rather than the viewport, you do a operation -> selection type of workflow instead of selection -> operation

-Particle size. The ‘balls’ that particles start out by default as are ok, and I can turn them into points, but do they have a size parameter, and if so can you then use that to make them collide with eachother based on that size…

I'm not sure of what exactly you're doing with your balls :twisted: but if you're in a geometry network, you can try using a point sop to control the particle scale attribute. You can use the interact pop to make particles interact with each other.

-Render speed. My experiments have led me to believe mantra is pretty slow, at least in comparison to modo. How do you turn on ‘gi’ like in other apps. Grey environment, gi on, nice overcast look….

I've yet to use Modo, so I can't comment on comparisons. There's a couple of ways to do GI / occlusion though. You can use an environment light (simple way) or the VEX Global Illumination shader (you should look up that one in the help docs since it's not as straight forward).

Allegro

There are plenty of videos available for learning in the “Learning” section of the website. There's also a lot more in the help docs than just the help for each node. But to answer your question. There is no printed manual that comes with Houdini.

I mean a reference document that starts at the beginning and explains everything with simple examples along the way. How houdini works, why it works the way it does, what are these chops and sops…. I guess even in the days of videos I prefer a ‘book’ of some sort, even if its a pdf and not ‘printed’…

If you press the “s” key or click the selection icon at the left of the viewport, you shouldn't have any difficulty with selecting things so that you can then use the shelf/tab menu to perform your next operation. Note that if you're laying down nodes in the network editor rather than the viewport, you do a operation -> selection type of workflow instead of selection -> operation

Ill have to try this. I just found I could rarely get predictable results in the group field with the list of selected elements for the next operation, I was getting extra edit nodes and things instead…

I'm not sure of what exactly you're doing with your balls :twisted: but if you're in a geometry network, you can try using a point sop to control the particle scale attribute. You can use the interact pop to make particles interact with each other.

Again, lots to learn here… but by default particles render as small spheres, all I want is a size parameter to go with the default size, I would have expected it in the location node that is creating them, but maybe it's a attribute you have to add, either way it wasen't easy to find….

I've yet to use Modo, so I can't comment on comparisons. There's a couple of ways to do GI / occlusion though. You can use an environment light (simple way) or the VEX Global Illumination shader (you should look up that one in the help docs since it's not as straight forward).

Modo's blazingly fast. I don't want to use an environment light, or occlusion, I'd like to use some type of interpolated irradiance like Final Gather in mental ray or Lightwave, or Irradiance cacheing like in modo. The VEX GI shader might be what I'm looking for, however the shading/VEX stuff seems even more daunting at this point than setting up dynamics and particle…. deep software.

Thanks for trying to help me out. I'll keep posting questions here as I run into stuff. Hopefully it will be useful to other new users as well. I've already figured out how to use proper SDS surfaces in the render and use a simple poly-subdivide SOP at the end of the SOP network set to render visible off. That allows a nice preview in the viewport, but dosen't dice up the geo heavily for the render, the true SDS work much better.

If your looking for a book, I would highly recommend WIll Cunningham's ‘The Magic of Houdini.’ Despite the frequent color commentary (which I don't mind, but many I know do), the book is extremely detailed and takes you through most of the important areas of houdini. Obviously, due to the magnitude and depth of the program it doesn't cover EVERYTHING, but you should feel confident in tackling most projects afterwards.

http://www.amazon.com/Magic-Houdini-William-Michael-Cunningham/dp/1598630822/ref=pd_bbs_sr_1?ie=UTF8&s=books&qid=1208969786&sr=8-1 [amazon.com]

Here's an example of getting those “grey environment renders” using the VEX Global Illumination SHOP. It's faster than using the Environment Light.

Mike_RB
Modo's blazingly fast. I don't want to use an environment light, or occlusion, I'd like to use some type of interpolated irradiance like Final Gather in mental ray or Lightwave,

Hey there! I hope you will like Houdini in time but in fact comparing Modo with Houdini or Mantra is pretty unreasonable.
Modo, as many others, is not fast, is incredible slow once you ask it to do something else then still imaginary with some sort of GI, which is nowadays rather trivial task judging form a number of solutions (including open source) out there. On the over hand asking Houdini for doing nice looking photo realistic house visualization can be hard - unless you use PBR in which case it's not hard it's just looonnggg

Saying that I must admit that Houdini lacks of ready to use fast GI solution, which is more a Mantra's setup problem then anything else. Once you know it, you can prepare your own shaders and tune Mantra to speedup things heavily.

The point is that so called “ultra fast modern ray tracers” usually introduce problems that are unacceptable for the industry Houdini was designed for.
Using of them become nightmare once we jump into moving, high resolution, high quality images. Many things that are crucial in such a case are absolutely absent in fast raytracing apps, like Modo.or many others freelence tools. Yet this lack gives it the apparent speed.

That's why quite a few facilities in commercial and games industries switch recently to Manta or similar products like 3delight, Air or PRMan. Not inversely. (btw Have you compared GI in PRMan and Mode?)

On the over hand fast GI or similar are not so important for films. Thus Mantra development was driven by a different set of needs. It doesn't change the fact I would love to have better GI setup for Mantra (for example via solid photons support which seems to be lately completely turned into PBR stuff what I really don't understand since I really want to use it in MP mode like many engines does.)

I saw once a person who was evaluating Houdini for a first time. He gave up after several minutes saying that Modo render GI faster then Houdini opens mplay (its framebuffer) - which ironically can be true in a bad luck of firewall problems, Windows Vista and ATI card.

Don't take me wrong but Mantra has so many serious advantages that a lack of fast GI setup out-of-the-box is a minor issue, at least for movies and animation.

Good luck!
Simon.

PS I've heard many good things about modeling in Modo. I don't think you'll be able to stand poly-modeling in Houdini

Thanks for the .hip, ill check it out.

And thanks for the book recommendation.

And Symek, while you have a point with the zillions of particles, or crazy deforming procedural geometry I don't think your assesment of modo, even in its current form is accurate. It's GI is fast, moving or not. We just completed our work on Iron Man using Lightwave to render and everything went very smoothly. We used GI and blurry reflections and ray tracing and area lights on all our shots some of which were rendered at 4k. Modo is the next generation LW renderer (written by the same author) and once it gets a few more features will be very useful. Even now it has true SDS surfaces, Instancing, micropoly rendering. Its AA system has been inspired by mental ray and renderman and gives you a great line quality very quickly. I'm hoping the next version will have the rest of the features needed to meet most needs (hell, if they can seperate it as a plugin for houdini even better).

Here's a test we did rendering in modo:
http://www.elementvfx.com/luxt02/crabhd720.mov [elementvfx.com]
micropoly displacement, physically based surfaces (fresnel and helmholtz reciprocity),blurry reflections, irradiance cached GI, soft shadowed ray traced direct light, physical sky, motion blur, DOF, painted footsteps… all native to modo. We did the animation in LW and baked out the deforming geometry and imported the hard body stuff in with FBX.

modo's not there yet. But I would love to be able to render in modo with Houdini….

M.

edward
Here's an example of getting those “grey environment renders” using the VEX Global Illumination SHOP. It's faster than using the Environment Light.

Thanks. I changed it to use full irradiance and gave it a latlong unwrapped HDR image as the environment image and that seemed to do the trick. It would be nice if there were controls to blur the samples together Irather than making the samples more accurate) but this is a good starting point. Everything we render is pretty much lit this way. Turning on the Irradiance Cache stuff in the mantra settings didn't seem to change anything and I'm guess might be for time smoothing rather than single frame interpolation.

Want to send me a realistic car paint shader built in here too (not a pre canned one, one that I can pick apart). With the various fresnel gradients piped into reflection or inverse piped into diffuse (energy conservation)… If you have one laying around that is……

You can use ambient occlusion with an HDR image, you don't need to use full irradiance for that. For your HDR image, make sure you use a .rat file.

edward
You can use ambient occlusion with an HDR image, you don't need to use full irradiance for that. For your HDR image, make sure you use a .rat file.

How would you get bounced light then? And where does one control the number of bounces?

Mike_RB

edward
You can use ambient occlusion with an HDR image, you don't need to use full irradiance for that. For your HDR image, make sure you use a .rat file.

How would you get bounced light then? And where does one control the number of bounces?

To interpolate your samples, save them on disk, control the way they are used (read/write/both) and control bounces you need to add appropriate parameters to Mantra ROP (Gear Icon of Parameter Tab -> Edit Render Properties ). Filter out (lower right corner) to select “cache” and “write” parameters and apply then to ROP.

SYmek

Mike_RB

edward
You can use ambient occlusion with an HDR image, you don't need to use full irradiance for that. For your HDR image, make sure you use a .rat file.

How would you get bounced light then? And where does one control the number of bounces?

To interpolate your samples, save them on disk, control the way they are used (read/write/both) and control bounces you need to add appropriate parameters to Mantra ROP (Gear Icon of Parameter Tab -> Edit Render Properties ). Filter out (lower right corner) to select “cache” and “write” parameters and apply then to ROP.

Thanks. I'll file this away to reference it when I get into that later. The ambient occlusion method works ok for now.

Alright, next question….

I want to build a shader/material (not sure of the naming in here yet). Basic carpaint. Both silver, and black.

Black carpaint has 0% diffuse, 5% facing reflection and 99% edge (glancing reflection) with a fresnel curve controlling the change over incidence.

Silver metallic carpaint: maybe 20% diffuse (grey) and another 40% blurry reflections (very rough) with the same fresnel sharp reflections on top. However, in this case the fresnel reflection need to subtract (invert multiply is better) from the diffuse and blurry reflection layers so we don't get more than 100% total light reflection (diffuse + blurry + fresnel sharp)….

I tend to stay away from Spec as I find I can't ever get a realistic look reflecting lights using spec, I instead tend to parent reflective geometry with a texture hdr of real world light scrimms to my lights. So I need blurry reflections on everything.

I'm assuming this isn't built as a precanned shader already, anyone want to link a quick demo of the network to get me started?

Mike_RB
And Symek, while you have a point with the zillions of particles, or crazy deforming procedural geometry (…)
M.

Really nice crab. Nothing to argue. If it works for you, it works. Art is made by people, not machines. I'd like to believe you, but I don't until I'll see it.

In fact I've just wasted three days working with a company that was using wrong tool (although great in many respects) for a wrong job. It was laying down on its knees after eating 4GB of RAM and constantly hanging.

Mike_RB
Thanks. I'll file this away to reference it when I get into that later. The ambient occlusion method works ok for now.

Alright, next question….

I want to build a shader/material (not sure of the naming in here yet). Basic carpaint. Both silver, and black.

Black carpaint has 0% diffuse, 5% facing reflection and 99% edge (glancing reflection) with a fresnel curve controlling the change over incidence.

Silver metallic carpaint: maybe 20% diffuse (grey) and another 40% blurry reflections (very rough) with the same fresnel sharp reflections on top. However, in this case the fresnel reflection need to subtract (invert multiply is better) from the diffuse and blurry reflection layers so we don't get more than 100% total light reflection (diffuse + blurry + fresnel sharp)….

I tend to stay away from Spec as I find I can't ever get a realistic look reflecting lights using spec, I instead tend to parent reflective geometry with a texture hdr of real world light scrimms to my lights. So I need blurry reflections on everything.

I'm assuming this isn't built as a precanned shader already, anyone want to link a quick demo of the network to get me started?

Note, that VEX GI Light Shader doesn't use many, useful options. Most of people here uses their own GI/AO implementation with point clouds, bend normals, blured hdri maps etc.

As to car shader:
This is some work of our friend, good starting place:
http://www.sidefx.com/exchange/info.php?fileid=352&versionid=352 [sidefx.com]

All calculations you talk can be done both in VOPs and in VEX rather easily.

cheers,
sy.

SYmek
In fact I've just wasted three days working with a company that was using wrong tool (although great in many respects) for a wrong job. It was laying down on its knees after eating 4GB of RAM and constantly hanging.

Ouch. What tool? BTW I'm not putting modo forth in its current form as a realistic killer rendering option. But I think the direction they are going is leading to a place where it might be there in the future.

Thanks for the car paint stuff. I'll take a look.

http://www.elementvfx.com/WebDemo/smoke_h9.mov [elementvfx.com]

Just trying things out, are shadow mapped lights the fastest way to light fluids?

Yes - very much so. Nice progress, btw.

Any way to get these cool vapour like vorticle things to work on normal particles? I was tweaking the various wind noise parameters to try and get cool smoky swirlies with normal particles but while neat, it didn't really looks right. Also when using winds with noise to mess up particles can you animate the ‘roll’ of the pattern, so it eveolves over time and not all the particles follow the same pattern in the same place?

Mike_RB
Any way to get these cool vapour like vorticle things to work on normal particles? I was tweaking the various wind noise parameters to try and get cool smoky swirlies with normal particles but while neat, it didn't really looks right. Also when using winds with noise to mess up particles can you animate the ‘roll’ of the pattern, so it eveolves over time and not all the particles follow the same pattern in the same place?

maybe this will point you in right direction:
http://www.sidefx.com/index.php?option=com_forum&Itemid=172&page=viewtopic&t=9878&highlight=particles+field [sidefx.com]


hth,
sy.

Thanks again for the link. Here's just a ton of particles with 2 layers of wind with thier offsets animated. I love being able to dial up the birthrate to 30000…. nice.

http://www.elementvfx.com/WebDemo/foam.mov [elementvfx.com]