TG2 Technology Preview issues

Started by commorancy, February 20, 2008, 07:27:09 am

Previous topic - Next topic

commorancy

I'm considering purchase of TG2 because of the render quality.  But, several things stand out as issues that I'm hoping are (or will be) addressed in the full release.  Hopefully, some of this has some been addressed in later beta releases.

Anyway, here are my points (not listed in any particular order):

1) The sun disc needs to distort (more oval and altered color) due to atmospheric distortion and refraction when at the extreme horizon edge.  Also, as someone else pointed out, the sun disc needs to reflect more accurately in water reflections when increased in size.

2) The renderer desperately needs the ability to render transparent objects and transparent water.  I also have a pine tree object that uses image mapping with the alpha channel and this tree cannot presently be rendered in TG2 due to lack of transparency support.  Of course, I could rework the pine tree object to model each individual needle, but that's adds a lot more polygon overhead. I would also like to try rendering interior spaces with exterior (and interior) light  using rendered glass as refraction and terrain visible outside.  Not possible without transparency / refraction support.   

3) Physical placement and building of terrain with the mouse (i.e., terrain by painting).  Being able to paint height fields and paint cloud patches would go a long way to allowing much more control over the environment.  Yes, a lot can be done programatically (and that's great), but painting allows for detailed finishing touches.

4) Needs more than 16 shaders per object.  Yes, I've read the thread about this issue and I understand the limitation.  But, when bringing Poser human objects into TG2, it's almost impossible to go through and cut down the number of shaders needed that Poser exports.  Worse, this goes back to #2, the eye lashes of Poser forms are made of an alpha channel image map and don't render correctly (among other Poser object rendering issues).

5) Crashing waves.   I would like to see an addition that allows for crashing waves along a shore line.  While I'm sure this can be done with an imported object and placed, it would be better to have support for this directly in TG2 as a function.  Of course, even an object is dependent on #2 and possibly even subsurface scattering to get the wave's light refraction to look realistic.

6) Export of the terrain (or other objects) in a standard format for use in other renderers or for reuse in other TG2 scenes.   The clips work, but I would prefer it to be exportable to OBJ, LWO or some other more standard format. This would allow people to create libraries of objects and sell or give away their created environments and individual objects (i.e., cloudscapes or mountain ranges) for use in more communities than just TG2.

7) Dockable windows.  When settings windows appear, I'd like to see a dockable tabbed interface.  Makes it easier than them just randomly popping up. I know the final interface is still in flux.

8) 4 port interface.  While the interactive 3D ortho view works, I'd still like to have a 4 port (top, left, right and ortho) setup (at least have it available as an option).  This allows for much easier placement of 3D objects in space and is more conducive to creating scenes containing objects.  For the ortho view, holding the alt key is unnecessary and requires two hands (i.e. cumbersome).  Instead, I suggest the following change:

In and Out: scroll wheel
Rotation: scroll wheel button
Up and down: right mouse button
Selection and movement of objects: left mouse button

Alt and ctrl can be reserved for object alteration functions like rotation, sizing or shear of selected objects.  Better, let the user decide how to map these functions within the app through preferences.

9) Wireframe imported object rendering in the rough render view.  While the cube does allow you to place an object, it doesn't allow for easy composition.  The viewport should at least render limited information (wireframe) about the object to make the overall composition easier in relation to the camera.

10) This one is a bug.  Populations don't work correctly.  If I use only one population, the population is constrained properly to the X by Y area that I specify.  If I create two populations, the application randomly decides to not constrain the population.  Instead, it places the population in a 360 degree circle around the centerpoint of the population rather than staying constrained in the rectangle.  This makes using populations difficult and unpredictable.

11) Depth of field / focal blur.  While a crystal clear image is useful, allowing for depth of field would allow for much more photographic and photorealistic images to be generated.  If there is a way to do focal blur in the Technology Preview, I've yet to find it.

12) Stars.  I'd like the ability to add stars to night or planetary scenes.

Anyway, that's my list so far.  Feel free to correct me on anything that I may have overlooked.

BTW, I don't need to know that these changes are in the later versions of Technology Preview (or any betas).  I'd just like to know if any of these features are planned as future enhancements.  Also, is there any date set on release of the full version of TG2?  While the technology preview works well, the limitations above prevent it from being worthy of the $200 price tag especially not knowing when the full release is coming.

Thanks.

--
Brian Wright

rcallicotte

Brian, I can't answer any of these officially, but the last two items on your list are available now.  As for the last thing about it not being worthy of $200, the final release will cost more.  $200 is for just now. 
So this is Disney World.  Can we live here?

zionner

I also cant answer these offically, and what calico says is correct on the last two,

However, just to point out if you look at the list of expected features/the small list of main diffrences between the current version and the beta, most of your other problems will be solved, from what I can see the only one's I cannot say for certain will be there is number 3,4,5(I dont know what 6 means :s) and 9

Also in my view, the sheer detail you can achieve with TG 2 gives it a well earned price tag of at least $200

commorancy

February 20, 2008, 02:56:58 pm #3 Last Edit: February 20, 2008, 03:01:49 pm by commorancy
Unfortunately for me, #2 and the #4 are show stoppers.  The lack of ability to render transparent objects (i.e., glass or transparent water) severely limits the ability of TG2. Combining this with #4 and TG2 cannot be used as a 3D renderer other than for terrain (and even then, not all forms of terrain.. like, for example, waterfalls or rivers).  There's also no way to easily create rivers or streams (other than by creating objects).

I have found that TG2 (through the nodes area) can export LWO and OBJ format files.  So, #6 may already be covered, but I'll have to test.  Not sure why it wasn't hanging off the main file menu, though.

Coming from a background of working with 3D modeling and rendering packages capable of CSG modeling, full fledged ray-tracing and radiosity (that cost around $400-$600) and that also support all surface types (transparent, bump maps, height fields, etc), $200 is a bit steep for the comparative missing features.  It's great that TG2 has so many mathematical functions and programming features, but that doesn't make up for the missing features in the rendering engine or the missing modeling/layout features in the GUI.

As I said, these issues limit TG2 TP to little more than a novelty landscape renderer and, for me, not really worth the $200 unless I know that a final release date is set and that most of these issues are addressed.  For example, I could take that $200 and put it towards the $495 cost of TrueSpace (which will render all of the surfaces that I need).  I could take that $200 and put it towards LightWave at $895 which does at least what TrueSpace does (and probably more).  And both of these 3D packages are available now.  For that matter, I could get POV-Ray and use it (which is obviously free) which does all of these surfaces.

If, as a programmer, you're planning to produce a renderer, then it's best to begin by supporting all known types of surfaces first.  I will say that there are two things that TG2 does well and that is 3D volumetric clouds and proper sunlight color balance.  There is no other renderer capable of producing clouds that look as good.  There are, however, a few renderers that rival the sunlight color balance.   But, the lack of transparent surfaces, limited shaders and hampered modeling/layout tools will, unfortunately, limit TG2's uses (even as a landscape renderer).

--
Brian W.

rcallicotte

TG2's main focus is terrain displacement and full-on planetary creation.  Nothing else out there will rival it, once it goes Gold. 

...and transparency will be available then, according to Planetside.  I've also read somewhere from the Planetside staff that the 16 texture limit will eventually be addressed.  By then, though, the price will be more along what you're talking about for Lightwave (as I understand the future pricing plans).

In other words, you could get this now for a great price.  It's already been used in a few movies and TV commercials.  If you wait until all of these things you want are implemented, then you'll pay the pricier price.  Not sure that would be in your pocketbook's best interest.  Your only trouble appears to be a lack of confidence that there will be a Gold version, I guess.
So this is Disney World.  Can we live here?

commorancy

You're going to end up paying for it anyway.  If the features don't arrive in 1.0, then you'll end up paying to get them in 2.0.  So, either way, it's going to cost you eventually.  For example, you pay the $200 now and get 1.0 and possibly limited features.  Then, when 2.0 is released, they'll sock you for the upgrade fee of another $200 or better.  So then you'll have paid $400 or better. I'd rather pay for a full featured package up front knowing it does what I want than pay $200 now not knowing when TG2 may see these features and then still be required to pay to upgrade to get the features I wanted in the first place.

Note that the $200 you pay now doesn't protect you in any way from upgrade fees later.  That $200 may promise you a 1.0 release, but it doesn't promise you the features in the release.  Planetside is not likely to promise to give you more than one release for $200.  So, expect to pay upgrade fees.

--
Brian W.

rcallicotte

What you don't understand, Brian, and might have found if you did some forum searching, is that Planetside has addressed the critical factors in what your calling 1.0.  It's actually 2.0 and will be the Gold release coming up.  There will be a Beta release for present owners of TG2 and then a Gold release after this next Beta release.  There will be no additional fees for Terragen 2.0 and this is clearly stated on their website.  BUT, if you wait until after it goes Gold and is no long the Technical Preview, you will pay extra.

It might be wise for me to let Planetside address your concerns, but these have all been clearly addressed in this forum, with the exception of a couple of your items that could just be a nice wish list for some day.  Most of the items on your list have been addressed by Planetside staff.
So this is Disney World.  Can we live here?

Cyber-Angel

Your point about crashing waves, which I assume you mean what are known as breakers are a vary complex phenomena that are not easily created with CGI: I repost here some thing I posted on Ashundar some time ago on the subject:

Waves like you describe are vary complex phenomena and would be extremely difficult to implement unless you modeled the underlying physics.

Breaking shallow water waves start far out at sea, with the sea surface been displaced by wind pressure, this combined with the centripetal force generated by the rotation of the earth produces waves.

The motion under the sea surface is a elliptical one, when the deep water wave approaches shore it increases in height until the force of gravity which is pulling on it means it can no longer support its own weight and the wave lip collapses in an asymmetrical fashion and this is what you see when a wave breaks.

There are many complex phenomena at work such as wave/ shore interaction that are not fully understood and so it would be vary difficult to produce an algorithm to render such phenomena without that data.

Terragen would need a 3D fluid solver probably based on a quad-tree data structure to do breaking surf, and the sea spray should be automatically generated.

This fluid solver would need to also consider:

Naiver-Stokes Equation
Reynolds Number
Law of Energy Conservation
Newtons Laws of Motion
Non-Ideal Fluids
Compressible and Non-Compressible Fluid Flow

those are just some of the things the solver would have to consider and since your talking about waves they would have to except back lighting and subsurface scattering, the water produced by this solver would need to consider multiple forward and backward scattering not single forward scattering as proposed in some of the computer graphics literature.


Regards to you.

Cyber-Angel   

Oshyan

Hi Brian,

You have a lot of good questions, concerns and points. I'll address them one by one. But I'm pleased to say that at least a good number of these will be addressed for the coming "gold" release.

As far as a specific release date, we're steering clear of them at this point. We realize it's frustrating not knowing when it may come out, but it's even worse setting a date and then not making it. We've had to disappoint people before and its very frustrating and unpleasant for all. The previous timeline had the final release scheduled for about now, following approximately 1 month of a beta cycle open to registered users. We're still working on the beta at this point, mostly hammering out some persistent rendering issues unfortunately, and you can expect a similar 1 month beta cycle once it is ready.

In regards to the price and capabilities, TG2 is fundamentally a landscape generation and rendering application. It's in the market with MojoWorld, Bryce, World Builder, etc. Certainly these other applications have features not available in TG2 at this point, but if you compare the end product I think you'll find that TG2's output is simply more realistic than any of the others. We've concentrated on the difficult problems that others haven't already solved. At times that does mean that desirable features like more robust object support have to wait a bit to get improvements, but without the robust landscape rendering TG2 would be just another also-ran, so it ultimately does make sense to develop this way.

Price-wise it's a steal at $200 compared to the competition. That's a pre-release price which will be increased once it goes "gold". If you purchase now you get free updates all the way up to (but not including) a 3.0 release. Whether you purchase now or later the upgrade price to 3.0 will be the same, so you're not saving any money by waiting, provided you do expect to buy at some point in the 2.0 product cycle. If your purchase is dependent on implementation of certain features, I'll try to address those below.

Now to your questions:

1: Sun disc distortion is indeed necessary for absolute realism. It's not something we'll be implementing for the coming "gold" release, but as our ultimate aim is total realism, it ought to make it into a future release.

2: Transparency is an important part of the improvements we have made for the upcoming beta. Refraction will be supported.

3: Paint-style terrain manipulation is a feature we'd like to add, but it's not a focus of development before the 2.0 release. There are lots of other very good and reasonably affordable dedicated terrain editors and we're unlikely to ever eclipse them in features, but we to intend to provide at least basic painting functionality for terrain edits as well as direct mask creation (e.g. population and texture placement). These things will come in future releases.

4: The 16 shader limit is one we certainly acknowledge. For our intended market and use of the product it is not generally a huge issue - the vast majority of vegetation and other natural object models use less than 16 textures. But we do certainly intend to remove this limit and support a much larger number of textures. It is unlikely that the "gold" release will be able to include this, but it will be one of our earlier updates following that.

5: As Cyber-Angel somewhat explained, crashing waves are an extremely difficult problem to solve. You have probably noticed that no one else has done this properly yet. Even cutting-edge research - where things are often accomplished that don't make it into mainstream applications for years - still has little to show in this area as far as realistic results. We'd love to solve this problem before anyone else, but it's a small enough part of the total landscape rendering feature set that it doesn't currently justify significant development time.

6: As you've found terrain export is already provided in the LWO format. You can export a specific area of terrain with the Heightfield Export LWO node, or export what the camera sees with the LWO Microexporter.

7: Dockable windows are a good idea. We've already got some improvements in the unreleased alphas that allow for better multi-window and multi-view workflow. I'll see what can be done in this area.

8: As I mentioned above the internal alphas now allow multiple 3D previews and node networks. These can be arranged any way you like. We may include presets that emulate the standard 4-view approach. Your control ideas are not necessarily bad for 3 button mice, but we need to accomodate 1 and 2 button mice as well (in the default setup), which is partly why it's setup the way it is. Customization of the controls will be possible in the future.

9: Wireframe rendering is definitely an important improvement for working with objects. It's not likely to make it into the initial 2.0 release unfortunately, but again it should be a focus for the updates that will follow.

10: The issue you describe is something of a bug, but it has nothing to do with multiple populations, which do incidentally work just fine. What you're seeing usually happens when you move the source object rather than the population. When the source object position is altered the population distribution gets distorted. This also happens with some of the built-in objects like spheres and it's definitely a bug in that case.

11: DoF is a nice feature we'd like to include. It will probably come after the 2.0 release with some other improvements in built-in post-processing such as post-contrast adjustment, exposure controls, etc. It would be based on z-buffer output which will also be available as a separate image output in the future. For now you can create a z-buffer type image using a Depth Shader and do DoF in post processing manually (as a number of images here have shown).

12: While a built-in "star shader" is not available (and will not be included in the "gold" release), you can easily create stars either procedurally (there is a tutorial on this here in the forums - try the Advanced Search), or by using an image map of a starfield (many are available for free online). Again there are many images posted here showing examples of both techniques used to create effect. You can also create nebulae and other space effects using built-in functionality.

I hope that helps you evaluate TG2. In making your decision on whether to buy the product you should be asking yourself if a dedicated landscape modeling and rendering application is really what you want. That is TG2's primary focus and it handles this area of rendering quite well. Other programs naturally do rendering of complex 3D objects better, and this will always be the case, even when we improve the object and texture handling. If object rendering of the highest quality and greatest flexibility is of most importance, you may be better off looking at another application. However at the same time don't expect Truespace, or even Lightwave, to offer you equivalent landscape rendering results, even with a significant amount of work.

Your best bet if you need both is to create a workflow that allows you to combine the strengths of two or more applications. Use TG2 to create and render your landscape as an HDR for example, then use that as an environment map in your general purpose 3D app and you'll get realistic scenery, and realistically lit and rendered objects. This is the kind of workflow many effects industry production houses use and it is essentially what TG2 is tailored toward.

Let us know if you have any further questions or concerns.

- Oshyan

RealUser

Quote from: commorancy on February 20, 2008, 07:27:09 am
5) Crashing waves.   I would like to see an addition that allows for crashing waves along a shore line.  While I'm sure this can be done with an imported object and placed, it would be better to have support for this directly in TG2 as a function.  Of course, even an object is dependent on #2 and possibly even subsurface scattering to get the wave's light refraction to look realistic.
--
Brian Wright


Here is an interesting interview regarding waves and so ...

http://www.vfxtalk.com/forum/vfxtalk-meets-guardian-waves-t6983.html?s=decdb72455cdf76a5279bc8d0dfd93f2&t=6983

Regards
Markus
Markus / RealUser
...................................................................
visit my Renderosity Gallery at
http://www.renderosity.com/mod/gallery/browse.php?username=RealUser
...................................................................

Sethren

February 21, 2008, 06:36:28 pm #10 Last Edit: February 21, 2008, 06:59:26 pm by Sethren
I still don't get why crashing waves are so hard to do. What you could do is have a system like the old terragen foam system where the foam and ripples are effected by the terrain mass and then implement a twisting type of displacement along the Y axis while that displacement is being run by similar methods used in the older water for terragen 0.9 therefore the displaced water grows stronger and widens along with the foam as it approaches the land mass and at the tips of the waves create a volumetric cloud for the mist that could somehow be attached to the water node along the ends of the wave tips. The cloudy mist also needs to increase in strength as the waves do but again using the old TG foam method. There could be function that limits the size of the crest height as it grows otherwise the waves may look to big in there proper scale. The twisting should not be that much different then what is currently being done with those hurricane/tornado methods. As the waves hit the terrain they break and disburse but the disburtion seems to be the harder part but i thought that the old TG0.9 foam disbursed as well. If this can be faked then problem solved.

It's like doing overhangs for terrain but adding a bit of a twisting effect to it as the twist reacts to the terrain coast.

I want to do waves do badly but if i am wrong about these methods then i guess the best thing is, is to import a mesh of a modeled wave and add cloudy mist along the edges. Modeling a cresting wave mesh would be a PITA.   :P


rcallicotte

Sethern, I hope you're right.  I have no idea how this would work, but if it can be done, maybe your ideas will spark a new node...or something. 

We want that 100 word sentence to count for something.   ;D
So this is Disney World.  Can we live here?

Sethren

I have no idea if i am ever right.  ;D

But it seems to me that if the older TG0.9 foam functions could be implemented with the possibilities of displacement then we should get close.

Also i was thinking that a nice idea is perhaps the foam could also act like the mist as it builds up it becomes displaced itself and fades out like the cloud edges would normally do.

The twist function idea is no different then what photoshop had for years now. I don't see how hard it is to be in a 3d context, i mean it's still displacement. 

Cyber-Angel

February 21, 2008, 08:37:34 pm #13 Last Edit: February 21, 2008, 08:40:07 pm by Cyber-Angel
Displacement, is not a substitute for fluid dynamics and never will be and is certainly no substitute for the micro-scale phenomena found in the swash/ back wash zone such as the micro voracities that you see when backwash has occurred, before the next incoming wave nor the small bubbles that are left behind from the foam.

The fact is that there is much to learn about the wave/ shore interaction and about waves them selfs; to really find out what work needs to be done then your talking about a full scale review of all the work published in the area. This is some thing that might help On wikipedia that has an overview.

Regards to you.

Cyber-Angel 

rcallicotte

Check out the hurricane / tornado thread in the Discussion area.  The tornado has been born and this leaves me wondering what could at least be accomplished in water (et al) via illusion, since what CA says here is clearly a state of fact.  This doesn't need to stop the momentum building toward something that could appear to be what we want.  All effects in movies are just that - illusion.  Or am I wrong about that?


Quote from: Sethren on February 21, 2008, 07:11:10 pm
I have no idea if i am ever right.  ;D

But it seems to me that if the older TG0.9 foam functions could be implemented with the possibilities of displacement then we should get close.

Also i was thinking that a nice idea is perhaps the foam could also act like the mist as it builds up it becomes displaced itself and fades out like the cloud edges would normally do.

The twist function idea is no different then what photoshop had for years now. I don't see how hard it is to be in a 3d context, i mean it's still displacement. 
So this is Disney World.  Can we live here?