Modelling and Rendering the TMP Enterprise

[Science Officer]

Hi Folks,

This is my first post on TrekBBS. I've been a member of the TrekMovie forum for a number of months posting attempts at a decent render of the TMP Enterprise. However I've really liked a lot of the work going on over here and it finally dawned on me that you would be the ones to ask for advice on how to advance my project should I hit problems.

A guess a little background might be in order. For a number of years I've had a copy the Dennis Bailey TMP Enterprise in Lightwave format, which I ported into Cinema 4D v8.2. Unfortunately the port was not good - texture maps were missing or incorrrectly mapped, the mesh was full of
holes, overlapping polygons, duplicate points and so on. Eventually I got all this fixed up and started to think about lighting, shaders and a fairly simple scene to try it all out. That was the point I joined the TrekMovie forum.

With helpful criticism and encouragement from Saquist and Scribbler on TM forum, I tweaked away until I got the render below:

http://img42.imageshack.us/img42/1618/ussenterprisencc1701fin.jpg

However I wasn't entirely happy with the result, broke the render down and discovered two things:

1. I'd messed up my inverse-square lighting and ended up with illumination that had no falloff.
2. The mesh didn't have sufficient polygons and consequently C4D's Phong Tag was unable to render a smooth surface. There were discontinuities in illumination and reflections.

Now admittedly, I've made life difficult for myself here as my major irritation is self-illumination of the Enterprise, particularly the saucer! The TMP movie and DB model used ghost lights - the former did this by using external light sources and dental mirrors. But I'm keen to get the
illumination working from the ships sources as designed.

Having corrected the inverse-square lighting problem, I realised I had a disaster. The bottom of the saucer was severely over-illuminated close to the centre and on the top of the saucer, deck 2 blocked light reaching the lettering (especially USS ENTERPRISE).

Since then I've been trying to overcome these problems and I've made some progress. This has come about by:

1. Remodelling the bottom of the saucer, which now allows the Phong Tag to produce a smooth surface.
2. Using a different lighting set up with each spotlight consisting of 3 smaller lights and a gel to prevent over-illumination.

The images below show what I've remodelled to date and the new illumination.

http://img7.imageshack.us/img7/6841/saucer.jpg
http://img3.imageshack.us/img3/8018/navdome.jpg
http://img42.imageshack.us/img42/4299/rimx.jpg
http://img59.imageshack.us/img59/103/saucerwithspotlightson.jpg
http://img3.imageshack.us/img3/2835/navdomewithspotlightson.jpg

Since then, I've added the spotlight rectangles to the navigational array (aka spotlight housings) and my light sources sit at the bottom of those rectangles. My problem for now is to reconcile those small lights with the larger rectangles. I've some ideas, but they may take a time to
see if they work.

That's all for now! Sorry for the long-winded post - I have a habit of doing it.

Cheers,

Science Officer.

 

[MadMan1701A]

Looks great, to me. Lighting is something I still struggle with, in Blender... I'm getting better, but it sounds like C4D has a much more sophisticated lighting system than Blender does.

In your top image, all that looks a little off is that the specularity seems high, on the edge of the saucer. How many light sources were you using?

 

[Science Officer]

Hi Madman,

I'd imagine that Blender has a good lighting system as it is an upto date application compared with my 8yr old C4D. I downloaded it a few months back to see if it was a viable replacement, but was immediately put off by the UI. I think I read somewhere that the next release was going to address the UI and if so I'll take another look. You have my respect for having the patience to use it!

The top image is a combination of lights and image-based lighting.

The Enterprise's self-illumination uses standard lights, which I got badly wrong as there is no inverse-square falloff. This makes the hull look flat with the quality of a drawing.

The Sun and the Earth are represented by simple colour gradients mapped onto a Sky object (essentially a big sphere). I used radiosity to get the illumination and allowed rays to see the object so that the Sun and Earth can be reflected off the ships hull.

The big specular highlight on the saucer rim is actually a reflection of the sun gradient. My error is that the gradient covers too much of the Sky objects surface. From Earth, the Suns angular size is roughly 0.5 degrees - I think my gradients was roughly 3-4 degrees.

My aim is to try and correct as many of these problems as possible. I have some idea of what I want to achieve as demonstrated in the image below:

http://img64.imageshack.us/img64/787/combinedtest1.jpg

This uses my first attempt at a new saucer bottom. It has a more detailed Earth (blue and white noise for clouds combined with a more detailed gradient). I'm going for a shinier hull, so that the ship looks like it fits into its environment. The green sections of the hull will use a less reflective look. This image of the Apollo spacecraft kind of sums it up:

http://img12.imageshack.us/img12/5898/apolloref.jpg


So I've got a few ideas, but as with all things it boils down to time. I think if I can realise what the above demo shows, I'll be very satisfied indeed!

 

[MadMan1701A]

Hmm, in your combinedtest1, I think you have gone WAY overboard with the reflectivity. The hull of the enterprise isn't a mirror finish... It's more of a matte surface... you shouldn't be able to see defined reflections in there.

I use a simple lighting rig, that seems to work pretty good... I use 1 really bright "sun" lamp, set at a value of 1.2 or so, for my main white light. If I'm giong to have a blue planet in the scene, I simply add another lamp, much weaker, but the blue color I want, and have it coming from the direction of the planet, obviously. That gives a little "bounce" of color from the planet, without having it necessarily reflect it. Does that make sense?

I think the "radiosity" you are using might be the problem here. Blender has something called "Ambient Occlusion" which basically brightens up the entire scene... From what I've learned, that is really only for indoor architectural type images.

I really like your saucer lights, though.

 

[Science Officer]

Hi Madman,

Reflectivity is a tricky one. I've only seen reflections used once on a CGI starship and that belongs to Vektor. But it makes the ship look so much more believable.

Problem is I can see a high reflectivity hull working - but being able to convey what's in my head in CGI can be really tricky. It's a matter of putting together lots of subtle effects that take away that pristine look. For example, the leading edges of the hull should look tarnished due to hitting atoms and molecules at very high speeds. Add in some minor space debris collisions etc and things look a lot better .

I went through this phase with the spotlight illumination. I knew that the lights had to use inverse-square falloff (afterall you can't entirely ignore the laws of physics). My first attempts at using gels on the lights were rubbish and counter-intuitive. But I knew what they could look like so I just carried on refining it.

It is very likely I won't be able to use C4Ds standard blurred reflections anyway because of render time. Throw in radiosity and the time goes into several hours, which is ridiculous.

However C4D comes with a bhodinut shader called Danel, which has its own reflection algorithm. What makes it different is that you can add a falloff to the reflection so that the greater the distance between some object and your hull, the lower and more blurred the reflections become. You can see a similar effect on the LHS of the Apollo Service Module image from my previous post.

This might solve the problem that you see and remove the details from the Earth reflections. It would be a plus to me as I'd see much sharper defined reflections coming from other parts of the hull. I'll have to look into it.

Cheers,

SO

 

[MarianLH]

Hmm, in your combinedtest1, I think you have gone WAY overboard with the reflectivity. The hull of the enterprise isn't a mirror finish... It's more of a matte surface...

If I recall correctly, the filming miniature did have a very glossy finish in TMP, which worked with Douglas Trumbull's VFX approach. But when ILM took over for TWOK, it caused problems with their bluescreen method, so they toned it down.


Marian

 

[Science Officer]

Hmm, in your combinedtest1, I think you have gone WAY overboard with the reflectivity. The hull of the enterprise isn't a mirror finish... It's more of a matte surface...

If I recall correctly, the filming miniature did have a very glossy finish in TMP, which worked with Douglas Trumbull's VFX approach. But when ILM took over for TWOK, it caused problems with their bluescreen method, so they toned it down.


Marian

I think that was the case. The guy who painted the Enterprise put some notes down on the following web page:

http://www.olsenart.com/strek.html

He described the ship as a "gleaming opal" and even for TMP the lighting had to be toned down.

For TWOK they had budget constraints and ILM went for a more comic book approach with simple diffuse lighting plus optical compositing of various ship elements. Compositing using a blue screen likely meant that certain elements (i.e. the blue deflector) couldn't be shown reflected off the hull. For TMP I don't think blue screens were used, so a more realistic look could be achieved.

Ever since it has seemed to me that most CGI shots follow the ILM look. If possible I'd like to see more reflections on the metalic hulls.

 

[Cary L. Brown]

Hmm, in your combinedtest1, I think you have gone WAY overboard with the reflectivity. The hull of the enterprise isn't a mirror finish... It's more of a matte surface...

If I recall correctly, the filming miniature did have a very glossy finish in TMP, which worked with Douglas Trumbull's VFX approach. But when ILM took over for TWOK, it caused problems with their bluescreen method, so they toned it down.


Marian

I think that was the case. The guy who painted the Enterprise put some notes down on the following web page:

http://www.olsenart.com/strek.html

He described the ship as a "gleaming opal" and even for TMP the lighting had to be toned down.

For TWOK they had budget constraints and ILM went for a more comic book approach with simple diffuse lighting plus optical compositing of various ship elements. Compositing using a blue screen likely meant that certain elements (i.e. the blue deflector) couldn't be shown reflected off the hull. For TMP I don't think blue screens were used, so a more realistic look could be achieved.

Ever since it has seemed to me that most CGI shots follow the ILM look. If possible I'd like to see more reflections on the metalic hulls.

Well, the trick there is to recognize that it's not a "solid" color... it was a pearlescent finish, which is sort of similar to (though not identical to) "metal-flake" finishes and so forth. That is... there's a lot of scatter occurring below that glossy finish. A pearlescent finish will not be a "mirror-finish," by definition. It should never provide a coherent reflection, but should show some tremendously brilliant highlight effects.

 

[Science Officer]

Hi Carly,

You are correct - a pearlescent finish would involve subsurface scattering. Backtracking a little, I think it is a matter of deciding what type of finish the hull should have.



I think we all agree that the hull is metal and therefore it comes with certain properties (low diffusion and high reflectivity). Those properties are modified by the surface texture:

• Very smooth/highly polished - low diffusion, high reflectivity.
• Rough/tarnished - higher diffusion, lower reflectivity.

If I recall correctly the former generally has a constant reflectivity until about 70 degrees when it increases. Not so sure about the latter, but I'd guess it varies across a greater range of angles.

The Apollo image kind of shows the range of finish I'm interested in. The command module is too reflective, but something between that and the service module would be acceptable. For the green parts of the Enterprise hull, I'd go a lot more diffuse and less reflective (probably the same as the service module).

For the moment, I'd like not to get sidetracked by the issue of shaders. I need to push on with the modelling whilst keeping a careful eye that the self-illumination works.

Looking at the DB model, I've figured I need to slightly modify the top of the saucer and maybe add a slight clover-leaf/cross style indentation to the top of deck 2. That should help InvSq illumination get to the required parts of the saucer and prevent over-illuminating the top of deck 2.

By the way, I really like your work on the TOS Enterprise. May I ask, where do you intend to go with it next?

Cheers,

SO.

 

[Cary L. Brown]

Hi Carly,

Just a quick aside... my first name has only 4 letters... no "L" in there. Now, back to your regularly scheduled program...

You are correct - a pearlescent finish would involve subsurface scattering. Backtracking a little, I think it is a matter of deciding what type of finish the hull should have.

I think we all agree that the hull is metal and therefore it comes with certain properties (low diffusion and high reflectivity).

Actually, that's a bit of an overgeneralization. I know, for example, that Andrew Probert has stated that his personal idea about the hull of the TMP Enterprise was that it was some form of ceramic/polymeric composite. And that it wasn't "welded" together, but actually "spun" almost in the way that a spider spins a web.

Personally, I tend to lean more towards the metal-based construction... but I wanted to point out that this isn't as clean-cut as you may think. And the finish for the TMP ship was really done with the intention of reflecting the approach I just described.

Personally, I see the hull of the ship as being a very complex matrix, but the outer layer is almost certainly a deposited ceramic layer (with the ceramic applied almost as a very thick "paint"). This is sort of a consolidation of "conventional shipbuilding" (taking into account the unique material characteristics of metals) and the protective characteristics of glassine ceramics.

Those properties are modified by the surface texture:

• Very smooth/highly polished - low diffusion, high reflectivity.
• Rough/tarnished - higher diffusion, lower reflectivity.

If I recall correctly the former generally has a constant reflectivity until about 70 degrees when it increases. Not so sure about the latter, but I'd guess it varies across a greater range of angles.

The Apollo image kind of shows the range of finish I'm interested in. The command module is too reflective, but something between that and the service module would be acceptable. For the green parts of the Enterprise hull, I'd go a lot more diffuse and less reflective (probably the same as the service module).

For the moment, I'd like not to get sidetracked by the issue of shaders. I need to push on with the modelling whilst keeping a careful eye that the self-illumination works.

Looking at the DB model, I've figured I need to slightly modify the top of the saucer and maybe add a slight clover-leaf/cross style indentation to the top of deck 2. That should help InvSq illumination get to the required parts of the saucer and prevent over-illuminating the top of deck 2.

Everything you just described makes good sense to me.

By the way, I really like your work on the TOS Enterprise. May I ask, where do you intend to go with it next?

Well, my main goal is to work out the relationship between the sets and scenes we've seen in TOS and the exterior we've seen in TOS, all in a reasonable, believable "fill-in-the-blanks" fashion. Once I've got it all worked out, I may put together a set of blueprints, with deck layouts as part of that. Those will be made from this model, not "based upon the model" but literally from it.

What else will I do with it? Well, I may choose to provide an OBJ format copy to a few select folks who I think I can trust not to just put it out on the net for "general use" but who want to play with it for... as Gep said one... "carving it out to show the nuts and boltsy goodness." Personally, I'd love to see some of my design decisions become "widely accepted" (location of the main engineering set, structure of the pylons and keel, computer facilities, arrangement of the shuttle landing bay, etc, etc).

But ultimately, my main goal is just to figure out the "best" locations for all the sets (or "real rooms" which may have been partially-accurately represented by sets) on the "real" ship.

 

[Maurice]

For TWOK they had budget constraints and ILM went for a more comic book approach with simple diffuse lighting plus optical compositing of various ship elements. Compositing using a blue screen likely meant that certain elements (i.e. the blue deflector) couldn't be shown reflected off the hull. For TMP I don't think blue screens were used, so a more realistic look could be achieved.

That's not entirely accurate. You can't have blue lights on within or on the model while shooting matte passes, but you can certainly have blue elements in non-matte passes. The matte passes are used to generate the matte elements, none of the other passes are. as such. If you're doing as many separate passes as ILM did, it's not a problem.

For TMP the Enterprise was primarily shot against black velvet. The moves were then repeated with the model unlit in front of a high-contrast white screen to pull off a hi-con silhouette for matting.

 

[Kaiser]

Cool stuff

 

[Science Officer]

Hi Cary,

Sorry about putting the "L" in your name. It's not a good start annoying one of the people whose work you like!

Thanks for the information - this is one of the reasons I joined the forum. I didn't know about Andrew Probert's ideas for the hull construction, but now you've mentioned it, it sounds interesting. I'd like to look into it at a later date. Are there any websites you can recommend for this type of information?

When I first started to render the Enterprise, it was recommended that I start with a familiar scene - such as in orbit around Earth. I was keen to get the balance between sunlight and earthlight correct on the hull, so one of my references was the space shuttle. At that point the material of the Enterprise wasn't so important as getting a reasonably faithful reproduction of a real spacecraft in space. So based on a few images such as this:

http://img6.imageshack.us/img6/3706/earthmoongpn2000001090s.jpg

I got as far as this:

http://img16.imageshack.us/img16/3686/hiresradiositynonwarpnoh.jpg

A "ceramic-looking" Enterprise (in that it is whiter). Not long after, I abandoned it to pursue the metal-hulled render in my first post.

Cheers,

SO

----

Hi DS9Sega,

I stand corrected. Like with Cary, that's now 2 useful bits of information I've learnt today!

Thanks,

SO

 

[Kaiser]

Hi Cary,

Sorry about putting the "L" in your name. It's not a good start annoying one of the people whose work you like!

Thanks for the information - this is one of the reasons I joined the forum. I didn't know about Andrew Probert's ideas for the hull construction, but now you've mentioned it, it sounds interesting. I'd like to look into it at a later date. Are there any websites you can recommend for this type of information?

When I first started to render the Enterprise, it was recommended that I start with a familiar scene - such as in orbit around Earth. I was keen to get the balance between sunlight and earthlight correct on the hull, so one of my references was the space shuttle. At that point the material of the Enterprise wasn't so important as getting a reasonably faithful reproduction of a real spacecraft in space. So based on a few images such as this:

http://img6.imageshack.us/img6/3706/earthmoongpn2000001090s.jpg

I got as far as this:

http://img16.imageshack.us/img16/3686/hiresradiositynonwarpnoh.jpg

A "ceramic-looking" Enterprise (in that it is whiter). Not long after, I abandoned it to pursue the metal-hulled render in my first post.

Cheers,

SO

----

Hi DS9Sega,

I stand corrected. Like with Cary, that's now 2 useful bits of information I've learnt today!

Thanks,

SO

that pic almost looks like as if Nasa really took a snap shot of it in space

 

[Professor Moriarty]

http://img16.imageshack.us/img16/3686/hiresradiositynonwarpnoh.jpg

Oh yeah. I am really digging the Ceramaprise, Science Officer... don't be so quick to abandon her. This finish really makes you see the lineage between the TOS Enterprise and her TMP incarnation.

 

[MadMan1701A]

I agree. That image is just about right. It's a little bright, though.

 

[Cary L. Brown]

I agree. That image is just about right. It's a little bright, though.

Exactly... the only "flaws" with that image are based upon the lighting. It looks like it's lit in a room with uniform lighting.

In a "space shot," you need very dim ambient lighting (representing starlight). You'll want a mid-intensity light source coming from the planet (to represent light being reflected from the planet surface... and probably, in this case, slightly tinted towards the blue/green range... but not TOO much so!). This light should either not cast shadows or cast very, very soft shadows (since you're getting "bounce" from the entire planetary surface, not just a single point source, after all!). And you'll want a single distant "directional light," tinted slightly yellow and significantly brighter than either of the prior two light sources, to represent the light from the sun, with this light source casting hard-edged shadows.

Just think about where the light would really be coming from. Look at the "background plate" you'll be using and figure out where the sun is relative to that plate, and place your "sun" in your render shot in that orientation relative to your render model. Look at the "center of mass" of the planet and place your planetary light source there.

There are other folks... many of whom frequently post right here on this BBS... who know more about this sort of thing than I do (I consider myself a "high-school-level student of lighting" right now, and a "kindergarten-level student of texturing"). But the concepts should be pretty easy to think through, huh?

 

[Gepard]

I'll just throw a "huge image" warning onto this post and have done with it. Hey, it's the art forum, after all.

To add onto/clarify/correct some what Cary is saying, we should make a distinction between what's physically accurate versus what looks good. A lot of CG space shots will do a setup with some kind of ambient occlusion global skylight combined with a brighter directional sunlight. It works well. It's worth noting, however, that it's a total cheat--starlight is far too dim to show up in a photograph with the f-stop settings needed to prevent completely whiting out the frame from the sunlight. It's the same reason you can't see stars in space.

The major factor in real-world lighting that gives the fill light in NASA photographs are secondary bounces and reflections, as the OP has correctly surmised. For all practical purposes, there's only one light source in orbit, and even the diffuse bounces from a planet don't seem to be bright enough to show up. Have a look at these two photos.

Note how in this one, the shadow side of the astronaut is brightly lit--being surrounded by white and reflective metal machinery, the sunlight is being scattered to fill in all those shadow areas and add depth to the image.

For this:

Pay attention to the sunlight coming from the left and the fill "light" on the right. At first glance, it looks like the fill is the ambient illumination I mentioned before, but if you look at the surface of the shuttle, you can see it's actually a very blurry reflection. Notice how it's almost impossible to tell the difference between the white and black tiles on the right side, and how the shuttle body is picking up the coloration of the continent below it. That's because it's not being lit by diffuse light but is reflecting the Earth underneath; we can't see a color difference between the tiles because they have the same level of reflectivity, independent of their color.

Note also that the back side of the bay doors on the left, which should be in shadow, are in fact brightly lit and have a bit of red spill on them. That's the secondary bounces again, picking up a bit of color from the NASA logo or American flag or whatever it is that's on that side of the shuttle.

I like to use the key/ambient fill setup myself, and I tend to render the two passes separately so I can curve the fill down in Photoshop and get the effect I want. It gives a decent effect of all those secondary bounces without the render hit of calculating them for real, and looks good enough if one isn't comparing against a real photograph.

It's essentially the same lighting rig used by Zoic for Battlestar Galactica.

For slightly more realism, I also like to use a specular-only spotlight to highlight the shadow side of a ship, after toning down the ambient fill even more. It's cheap, but it saves the time of setting up a realistic blurry reflection to do the same thing.

The effect is somewhat overdone here, but note how it has the same "glossy" quality as the right side of the above shuttle photo.

Because I'm using Final Gather for both of these images, there are a limited amount of secondary bounces, but it's set to the minimum number, which doesn't give a realistic enough effect without the additional fill. But some variation on all of this should be able to create an accurate lighting rig for orbital images.

Oh--it's also worth noting that as much as I'm harping on making sure you have the right light, knowing when and where to use blackness is also important. One of the major giveaways for a CG image, producing that glowing, plasticky look it sometimes has, is when light gets into places it shouldn't. I don't recommend turning the shadows off for a light unless you really, really need to – for render times or something.

Anyway, hope that helps. And stretches your monitor.

 

[Science Officer]

Hi Folks,

Thanks for all the input today.

Unfortunately I've lost the file containing the "ceramic" Enterprise as rendered in the earlier post. Given the errors I found in the later metal-hulled work, I quickly scrambled together various bits and pieces from earlier and later files and did a quick confidence render. Luckily I have something reasonable as shown below:

http://img687.imageshack.us/img687/2064/roughrender.jpg

The biggest change is caused by tilting the fake Earth to the same angle as the proper Earth image. The one in the "proper" render had the planets limb running vertically. It also lacks self-reflection and spotlights so the underside of the saucer looks very empty on the LHS.

The other image below was a later test of my much shinier metal Enterprise saucer. Unfortunately, I accidentally switched off the sunlight!

http://img251.imageshack.us/img251/5995/environmenttest.jpg

I think I need to figure out my priorities. If I pursue my shinier metal-hulled ship I'll need to carry on remodelling. Reflections and self-illumination will not hold up on the DB-model as it stands.

However if I focus on this specific camera shot, I should be able to combine my remodelled elements to the DB-model. That way, I know the spotlight illumination will work. I could then deconstruct both the environment and hull shaders to make sure they are working properly. I suspect that shaders will require the most effort.

Again, thanks for the input today and the thumbs-up for the "Ceramaprise". I'll try to take it all on board and make the best use of it I can.

Cheers,

SO.

 

[Gepard]

Your metal-hulled saucer would benefit from a texture on the reflection channel to capture that high-contrast aztec look; perhaps try copying the specular map and adjusting from there.