How to make ST ships look suitably huge?

andar_b · April 2012

I've been pondering how to make the Defiant model I've been working on look 'right' when rendered. The textures aren't bad, and most of the details are screen accurate, but it still looks like a Happy Meal toy when rendered. I've done my best Google Fu but it's hard to find good information on the topic.

From what I've seen, Trek usually has an infinite focus on space shots, I think? There doesn't seem to be any significant depth of field anyway, nor depth queuing except in nebulae and dust clouds. Upon watching DS9 and Voyager again, I noticed it was usually quite hard to tell the difference between the CG models and the physical models, and I think that is in part due to the limitations of Standard Definition television making it easy to disguise lack of detail. On the other hand, the NX-01 always looked like CGI to me, and I can't explain why beyond the use of HD.

One thing I've considered is how the physical model of Defiant has streaks and smears on the hull, like it was probably coated with a dilute wash of dark-colored paint and wiped clean. This left it with a combination of the Trek 'clean' look and the grimy look of the battle-hardened ship she was. I can't see a satisfying way to simulate that technique in 3D, though AO does help a very little bit.

Maybe my best bet would be to avoid high resolutions? I guess this is a case of 80% of the work being the last 20% of the job. I don't have a background in traditional art or photography, so I don't know what sorts of camera, lens and film simulations I could do to help my renders look like the 'real thing'. So far I've been working in the original Blender renderer, in case it matters.

Chris2005 · April 2012

andar_b wrote: »

I've been pondering how to make the Defiant model I've been working on look 'right' when rendered. The textures aren't bad, and most of the details are screen accurate, but it still looks like a Happy Meal toy when rendered. I've done my best Google Fu but it's hard to find good information on the topic.

From what I've seen, Trek usually has an infinite focus on space shots, I think? There doesn't seem to be any significant depth of field anyway, nor depth queuing except in nebulae and dust clouds. Upon watching DS9 and Voyager again, I noticed it was usually quite hard to tell the difference between the CG models and the physical models, and I think that is in part due to the limitations of Standard Definition television making it easy to disguise lack of detail. On the other hand, the NX-01 always looked like CGI to me, and I can't explain why beyond the use of HD.

One thing I've considered is how the physical model of Defiant has streaks and smears on the hull, like it was probably coated with a dilute wash of dark-colored paint and wiped clean. This left it with a combination of the Trek 'clean' look and the grimy look of the battle-hardened ship she was. I can't see a satisfying way to simulate that technique in 3D, though AO does help a very little bit.

Maybe my best bet would be to avoid high resolutions? I guess this is a case of 80% of the work being the last 20% of the job. I don't have a background in traditional art or photography, so I don't know what sorts of camera, lens and film simulations I could do to help my renders look like the 'real thing'. So far I've been working in the original Blender renderer, in case it matters.

Well, you figure, the Defiant is to the Enterprise E as a Cessna 152 is to a 747... next to a 747 a Cessna 152 looks like a toy...

The easiest tell between the CG Voyager and the model model, was the colors, the CG model had more gradient color mapping, with the deflector dish, and warp grills, and the bussards, and overall lighting.

The Defiant was a little more tricky, especially later and in First Contact...

Well, the vessels in Star Trek are really big, and distance between objects also can affect DOF...

For example here:

The focus is not directly on the shuttle, in terms of center frame, however, the focus may be adjusted to the shuttle, which might create a small blurry look to the Enterprise... but since it's so big and close... it might not be as blurred out as one would expect, of course, seeing this shot in blu-ray quality might change how this shot looks overall.

Like here, despite the distance between the Cessna and the 747, both are still more or less in focus, one might be slightly more blurry, but it's negligible.

AirlinersNetPhotoID336520.jpg?m=1289692836

AirlinersNetPhotoID336520.jpg?m=1289692836

However, I'm by no means an expert in photography either.

andar_b · April 2012

Nice comparisons

I haven't tried to pair it with any other ships yet, I'm just looking for techniques to help give the appearance of scale. I'm at the point of knowing something looks wrong...but not what or how to fix it.

nightfever · April 2012

You can achive a proper scale by finding a good perspective. There is a simple principle in photography: Shoot an object from below and it will look big, shoot it from above and will look small.

IRML · April 2012

you are going to struggle with scale with a lot of trek ships because the window sizes are too big

depth of field shouldn't be a factor, unless you have something small close to the camera with a large ship in the background everything will be in focus, even the smallest shuttles in trek are bigger than a car, so think about how that works in focus with something like a building in the background

Aresius · April 2012

Beyond making the textures fit to a certain resolution, I noticed that lighting (luminosity, direction and focus) are also very helpful to create the feeling of "right size".

The main problem you get when looking at the Trek ships is that they're sometimes unscaled due to the fact that they used models from various sizes. Just think of the B'Rel.
In some scenes it seems giant (like when Kruge faces the Merchantman, or Kirk decloaks over the whaler), on other settings it seems small compared to the Enterprise (when Kruge tries to take on Kirk).
And the irony is, that in the TNG episode "Matter of Honour", it's the same ratio, despite the Ent-D being much bigger than the Ent-A. Which would mean the B'Rel is bigger as well. And during "Way of the Warrior", they're even smaller than usual, almost fighter-craft size.

evil_genius_180 · April 2012

nightfever wrote: »

You can achive a proper scale by finding a good perspective. There is a simple principle in photography: Shoot an object from below and it will look big, shoot it from above and will look small.

Yeah, that really makes all the difference. Of course, it's impossible to completely avoid top shots, but you can work the angles so that they at least look a bit more impressive (big.)

Normally, I'd also suggest adding rooms behind the windows to add scale. But, being as this is the Defiant, that's not really going to be effective.

andar_b · April 2012

Thanks all. On the window point, of course there are a minimum of windows on the Defiant class, and there's no definite proof that the 'windows' on the underside are even windows at all.

The main problem you get when looking at the Trek ships is that they're sometimes unscaled due to the fact that they used models from various sizes.

You could have easily used the example of Defiant herself, and how the graphics department couldn't decide how big she is, which combined with the same problem in the case of DS9. If you take DS9 as big enough to dock with a Galaxy class, then we'd have the problem of the itsy bitsy Defiant.

Beyond making the textures fit to a certain resolution...

As in enough detail but not too much? The current textures generally look good to me, there's a mild random 'Aztec' pattern just visible over the paint, and only in the right light, and I've added some dirt maps for distressing. The paint on the lettering and pinstripes is raised just a smidge, and is more specular like on the model.

I think, at the end of the day, it's the problem I've often seen with 3D models - they look too 'clean' even when textured quite well. No matter how tidy it is, a hand-made physical model will always look 'less perfect' than a 3d mesh, I think. It's kinda like the 'uncanny valley' in human modelling. Once you get it to a particular point, it's hard to tell the difference between life and art, but there's just something 'wrong' about it. I'd like to overcome this, but not sure it will be within my grasp at this point.

Doomsponge · April 2012

I've found a bit of ambient occlusion helps too. I don't know exactly what it is it does, but I was told (might be wrong here, if so my apologies) that it adds shading in the recesses. Basically, it 'upscales' the shadow on the model. Plus, one thing CG mo9dels have that physical models often don't is that it's impossible to get an entirely sharp angle in reality because of things like paint, et cetera. Almost always they'e'll be a tiny, barely noticeable on inspection, bit of 'bevelling' or such that; s caused by the moulding or painting processes. It's not that noticeable physically, but the light reflects differently off it.

The other thing that's easy to fall into is trying to light it too realistically. It sounds counterintuitive, but because of how models are usually displayed on TV, we're used to lighting maps that aren't entirely realistic. For instance, the window lights (again, not such an issue on the Defiant, to be fair) tend to be brighter than the ambient sunlight, which in reality wouldn't be the case, but when it's done how it 'should look' in reality, the lighting in the windows looks too dim.

Aside from that, I'm not sure. Maybe changing the depth of the panel lines, maybe altering the shadow types or bump mapping.

MadKoiFish · April 2012

I have to argue with the window lights, the apparent brightness is strictly dependent on exposure of the exterior. What has been said assumes everything exterior to the ship is shot with ambient light close to our own "daylight" TO which is likely is not. SO in the mentality of photography when you adjust an exposure to show detail in a dark situation anything lit lighter will blow out or overexpose hence why lights in those trek windows are so bright.

So in 3d we often just set things up with daylight like settings on the exteriors of our ships and have to overblow the interior lights to mimic what we have seen in efx shots.
Some examples of buildings exposed for exterior detail showing lights on inside.

So in the end I think we need to remember to think that we are seeing what we do in 3d as looking through the camera, not with our own eyes. Altering "exposure" via lighting, you need to balance everything else in a scene.

A way to avoid this and I am sure some do is to set up real lighting values and stick to those and only use the exposure controls within your app. But at a hobby stand point I do not know if anyone would or does go this far. But again none of this really helps none in altering a sense of scale. Just a sense of realism, which is subjective based on opinions of what your willing to judge reality as. Example, trek most scenes (least studio stuff) contain way too much ambient light for reality's sake. But one could argue reality based on if it appears as a cartoon or as a real photographed object regardless of the lighting.

As for the subject of the thread having no examples of what the issue is makes it even more vague to answer. I found most cgi lighting fails is due to the apparent size of the lights used on the ships. Studying how light interacts with large objects vs smaller ones and how large or small the emitting source of light is can cause a alteration of scale. I have found too most images that scream cgi are the ones with led like sharp focused shadows. Also field of view can affect this. Most apps represent this in a 35mm camera lens system. Whereby 55mm is closer to human eye. Where 20mm will begin to fish eye, showing more "view" and 200mm flattens and narrows the view. Focal length, is often miss represented as cropping an image and rarely have I seen its distorting effects mentioned EG

THis is just not how it really works. As you increase the from say 100~500mm your crushing the apparent distance/depth if the image making it look flat.

Not the best example but in the first sets of images you can see how distances seem not as dramatic.

http://www.ormsconnect.co.za/2011/11/comparison-of-tele-to-wide-portraits/
some really good examples.
In photography or miniature photography this use of lens is often used to single out an object or to make a smaller object appear larger or to a specific scale.

Also to sell a image is noise and distortion. Look at older photos of efx scenes, you'll likely find chromatic abbreviation, grain, and vignetting. Maybe even some flaring IE inconsistent contrast or tonal value across the image, to full on bloom or flares.

finally relationships, I think chris sorta touched on this. TO give scale you do need something to relate to. This does not help when your rendering out a single ship with a star background. This is when the above mentioned things should help a little but you will always struggle with a singular subject image since any tells will be plain to see.

andar_b · April 2012

It sounds counterintuitive, but because of how models are usually displayed on TV, we're used to lighting maps that aren't entirely realistic.

I understand perfectly. We're used to how it's 'supposed' to look, so when we see something as it would look in reality, it looks odd. Sorta like how gunshots on TV sound completely different than reality. Likewise, photorealism is more about making it look like a photo would, as opposed to looking like REALITY would.

Chris2005 · April 2012

Yea, the shadows often portrayed in Star Trek, are how shadows react in an atmosphere, since it'd be hard to film a miniature in a vacuum, at least under such a budget, even in CG, the shadows are often shown as area shadows or something similar, to generate soft shadows, that diffuse with distance, which in reality isn't how shadows in open space would appear, even the shuttle, and the Hubble Space Telescope, both are in low earth orbit exhibit sharp contrasted shadows... in open space shadows are also sharp, not quite as sharp as Raytrace shadows... as they have some smoothness around the edges, but they don't diffuse with distance, as the light isn't diffused by much or any atmosphere... which is why shadows are sharp on the moon compared to here... and is mainly why I use raytraced shadows in my renders... although they're not aesthetically pleasing to what most are used to seeing in Star Trek... but here again, lies the limitation of not being able to shoot in a vacuum... and the style was carried into CG shots as well... they seemed to get sharper in Insurrection, Nemesis, and Enterprise... more or less...

MadKoiFish · April 2012

What I referred to with shadows was not the effect of particulate matter (atmosphere gast dust) on how light interacts, but more to do with size of the emitting source and size of the object being hit. Place a can or other tall object on a white table and place a light to allow it to cast a shadow across the table.

In cgi the tell is when you have a shadow cast by say a near nacelle on a far nacelle and it is hard crisp and does not alter no matter how close or distant is it from the source, this is a tell to cgi. It also ruins scale. Since distance from the source of the shadow and the shadow itself denotes scale.
This is all through the effect of lensing. Same principles as focus on a camera. (See pinhole cameras in google)
The same can be said for shadows that are too evenly soft.

Atmospheric affects the diffuse light, so it will light an a general way, everything. This is where a hard vacuum comes into play. BUT as we know space is not a void of nothing, there are places where dust gas and other particulate matter would produce this diffuse light. See the hubble images of star birth regions all that light is not emitted photons but light bouncing on gas and other "crap" BUT yes in general you would not get all that AO GI and other fancy make it look more real trickery we often employ in 3d. I use these techniques to ad "dirt" or noise to a object to give that sense of scale.

However atmpospheric and air grime, can affect a shadow in how soft it is or diffused. This is all due to light being diffused and bouncing about and lighting an object from more than a single source.

Ray traced shadows, I use these often. If not I will use area shadows. I never use shadow maps. Mostly I use ray because I can control the size of the emitted light source and the parameters of the softness of the shadows. NOW each application will have it's own shadow "stuff" and settings and terminology as well. Hell in 3ds max add in photometerics and specialized render lighting gizmos and it is endless.

Not related to scale though but lighting, I never use 0,0,0, black as a shadow value, I always fall to a dark near black in a blue green red or rust tone. While this might not be realistic in space where you would not likely get this effect I still use to it trick the eye.

Chris2005 · April 2012

MadKoiFish wrote: »

In cgi the tell is when you have a shadow cast by say a near nacelle on a far nacelle and it is hard crisp and does not alter no matter how close or distant is it from the source, this is a tell to cgi. It also ruins scale. Since distance from the source of the shadow and the shadow itself denotes scale.

However, in the vacuum of space, that's how light behaves... the shadow would be more or less sharp... it may be smooth around the edges, but it wouldn't diffuse.

MadKoiFish wrote: »

Atmospheric affects the diffuse light, so it will light an a general way, everything. This is where a hard vacuum comes into play. BUT as we know space is not a void of nothing, there are places where dust gas and other particulate matter would produce this diffuse light. See the hubble images of star birth regions all that light is not emitted photons but light bouncing on gas and other "crap" BUT yes in general you would not get all that AO GI and other fancy make it look more real trickery we often employ in 3d. I use these techniques to ad "dirt" or noise to a object to give that sense of scale.

Yes, space is not devoid of gases, etc. but most of space is a vacuum, at least that we have explored, and there's not enough gasses in open space to diffuse shadows with distance, unless one is in a nebula, etc. but since we've never been in a nebula, hard to know for sure.

So, the shadow cast by one nacelle would be sharp... unless in a nebula, etc. You can even seen the shadow from the ISS cast on the moon...

MadKoiFish wrote: »

Ray traced shadows, I use these often. If not I will use area shadows. I never use shadow maps. Mostly I use ray because I can control the size of the emitted light source and the parameters of the softness of the shadows. NOW each application will have it's own shadow "stuff" and settings and terminology as well. Hell in 3ds max add in photometerics and specialized render lighting gizmos and it is endless.

I can never find a setting to soften ray traced shadows... of course, I use fR-Raytrace Shadows... renders really fast...

MadKoiFish wrote: »

Not related to scale though but lighting, I never use 0,0,0, black as a shadow value, I always fall to a dark near black in a blue green red or rust tone. While this might not be realistic in space where you would not likely get this effect I still use to it trick the eye.

That I'm not sure of... however, I suspect if one is close to a light source, the light bouncing off reflective materials, would cast a dimmer shade of said color...

andar_b · April 2012

However, in the vacuum of space, that's how light behaves... the shadow would be more or less sharp... it may be smooth around the edges, but it wouldn't diffuse.

What Koi is saying reminds me of when you see multiple shadows cast by many small light sources close to each other, the shadows almost seem to blend into one diffuse shadow. A suitably large light source could cast softer shadows depending on distance and such, some points on the emitter would reach areas where other photons would be obstructed, and I think diffraction would have an effect as well.

I can never find a setting to soften ray traced shadows...

In Blender, for instance, there's an option for softness of ray shadows on each ray-capable light, as well as the number of samples. Unless you up the sample number, the softness value doesn't do anything.

I'm getting a little better results with slightly soft ray shadows, and I've fixed a lot of places there were errors, but there's still that sense of fake-ness to it, but after watching the remastered effects for TNG in another thread, which were done with an excellent model, I can see that some of that sense is even visible in such professional work.

Chris2005 · April 2012

andar_b wrote: »

I'm getting a little better results with slightly soft ray shadows, and I've fixed a lot of places there were errors, but there's still that sense of fake-ness to it, but after watching the remastered effects for TNG in another thread, which were done with an excellent model, I can see that some of that sense is even visible in such professional work.

Yea, even though I've used GI, AO, and film grain in one of my TOS FX projects, it still looked fake...

Well, TNG remastered still uses the original models for the most part... for the CG shots, they're using Tobias' CG model...

IRML · April 2012

MadKoiFish wrote: »

A way to avoid this and I am sure some do is to set up real lighting values and stick to those and only use the exposure controls within your app. But at a hobby stand point I do not know if anyone would or does go this far.

I do this, well worth the effort I think

MadKoiFish · April 2012

Chris2005 wrote: »

However, in the vacuum of space, that's how light behaves... the shadow would be more or less sharp... it may be smooth around the edges, but it wouldn't diffuse.

Yes, space is not devoid of gases, etc. but most of space is a vacuum, at least that we have explored, and there's not enough gasses in open space to diffuse shadows with distance, unless one is in a nebula, etc. but since we've never been in a nebula, hard to know for sure.

So, the shadow cast by one nacelle would be sharp... unless in a nebula, etc. You can even seen the shadow from the ISS cast on the moon...

I can never find a setting to soften ray traced shadows... of course, I use fR-Raytrace Shadows... renders really fast...

That I'm not sure of... however, I suspect if one is close to a light source, the light bouncing off reflective materials, would cast a dimmer shade of said color...

I am talking about soft shadows resulting from a lensing effect NOT particle diffusion. End of story. What your talking about is soft shadows produced by refracted light bouncing off a atmosphere thus creating GI and AO. Diffuse light or GI and AO lighten the black details or add a HDRI effect and make for extremely fuzzy shadows. And are often fairly consistent at the source of shadow. Again look at the example images. Note the sharpness of the shadow near the source. This happens no matter what the environmental conditions. This is a fact of how light works. Second this is something you can do at really small scale. SO any atmosphere or diffusion created by such would be negligible. A vacuum would make no difference to things. Truth be a proper vacuum can be produced in a jar, that still would not affect light or photons passing through that jar.

Examples of moon landing images showing shadows lensing to soft edges. A real btch to find undoctoerd unshooped up pic or one not cropped up due to the low angle of the sun.
I have seen ISS images with this same effect occurring on the long structures.

not the best since it is too evenly soft.

deep shadows are suspect here do ti obvious shooping to clean noise or likely flares.
KEEP in mind this effect is reduced the more oblique the angle to the object so shadows cast nearly above the object will be the sharpest.

ISS casting shadow on the Discovery. Note the "contrasty" shadow.

The soft shadows I am referring to are caused by the object casting the shadow lensing the light. Research things before calling someone a liar. Look at pinhole cameras on google and how they lens light by obstructing it through a hole. The principal works in earth as it will anywhere else, if it did not no one would be able to take photos outside of a atmosphere or under pressure.

Whole space is a vacuum argument
http://answers.yahoo.com/question/index?qid=20070802214004AA5DZG0
yes, poor source but plenty of links and research, it is also moot for this subject since it sorta does not affect scale of a scene.

Ray trace shadow lensing (whatever the proper jargon is) can be affected by the size parameters. In a mr omni or omni it is often located under "area light parameters" setting is radius.

Also with the whole shadow softness thing it is a fine balance since too much or too little will kill it. Your light source is most likely a sun, so your emission point is HUGE. Playing with a small flashlight and a can or other tall freestanding item on a table is a vastly different scale. But no softness at all in many renders I have seen totally kill it. I will render out some stuff for example with a singular light emission point. KIM if your shadow is directly or nearly under the object it will not soften much with distance. That shuttle image above I think the iss was quite some distance away. ISS pics are hard too cause there is just so much reflected light everywhere it dims shadows horribly. (that and they often shoot only planet side of the bloody thing)

I should probably proof read all this but hopefully it is clear enough. If someone knows the terminology or jargon please fill in cause ATM my brain just is not remembering what this is called. I do remember way back in JC studying this in physical since classes and photography classes. Seen it on countless edutainmet and sci education shows as well.

MadKoiFish · April 2012

Here are some sample images in max. One is no light source size. sharp hard shadows no matter how parallel the light source is to the shadow receiving surface. Another with it at 10ft. Pole thing is 26fX90ft light gizmo or source is about 180ft away and about 120high. Further out the softer the shadow will become. Larger the emission size the softer it will get. Balancing this effect is important.
Again since it is only a single light and only ambient light permitted is that of the reflected photons from the 2 objects, the pole and the ground. Nothing else.

clicky for full rez since the forum will adjust it.

I hate to bring this up, but that image name says transit ISS moon, that is the ISS not a shadow.
Read the article the image is pulled from earth based photo of the ISS passing in front of the moon.

http://www.universetoday.com/82013/international-space-station-on-the-moon/

It is really cool guys are able to do this, stunning.

IRML wrote: »

I do this, well worth the effort I think

I just saw your reply IRML, issue I mention feasibility is setting up the "sun" and spacing that gizmo the distance from the ship with the proper values. I know in architecture some guys are so anal about proper light and reflection on a render they will produce a full dome of the building location and sample or set up the light to a specific setting for various times of the year and weather patterns.
I really would love to get my blasted workflow set up so that I could do this, cause the biggest frustration is how max manages this crap. Exposure settings seem to do thing out of logic of what you expect from a photography stand point. Values from say scanline to MR to photometric do not at all seem to interrelate. This may be why many where light is a concern use photometric only. I just do not have the computing power to live with that though.

Anyhow does LW allow for this sort of thing? IE using kelvin readings and using light measurments instead a some generic unit like 1.0

(sorry I know it is sorta off topic)

MadKoiFish · April 2012

Ok last post I SWEAR! lol this here is just with a model to give a better feel than just some pole thing.

Really thinking about things when we render in cgi it really is just a vacuum. Because of that we have to resort to trickery to fake atmosphere dirt and gunk in the "air" Ding this through GI, AO, and the like. Worst is adding atmospheric haze and "god" rays and similar particle effects. That really is the largest impact a atmosphere has, distortion or lensing of the light rays, and creating distance haze that disrupts clarity.

Ok image, first is with the softening, second none looks fake. Final has too much and appears as a toy on a desk. Now there are other things you can do to the image to enhance the individual feelings of size but this is to just demonstrate how the change in the light object size affects the scale of the object in question.

Again click to get the full size,

Question too should I add spoiler tags too? There just might be some dial-up users out there still?

And now that the west coast is waking up I shal go to uh, BED! lol

Anyhow post up some examples so we have some basis for more suggestion of what needs help. Or what works or does not. I think it would speed things up and flatten out that learning curve some.

MKF-

IRML · April 2012

MadKoiFish wrote: »

I just saw your reply IRML, issue I mention feasibility is setting up the "sun" and spacing that gizmo the distance from the ship with the proper values. I know in architecture some guys are so anal about proper light and reflection on a render they will produce a full dome of the building location and sample or set up the light to a specific setting for various times of the year and weather patterns.
I really would love to get my blasted workflow set up so that I could do this, cause the biggest frustration is how max manages this crap. Exposure settings seem to do thing out of logic of what you expect from a photography stand point. Values from say scanline to MR to photometric do not at all seem to interrelate. This may be why many where light is a concern use photometric only. I just do not have the computing power to live with that though.

Anyhow does LW allow for this sort of thing? IE using kelvin readings and using light measurments instead a some generic unit like 1.0

(sorry I know it is sorta off topic)

it's not an exact science for me, I use kelvin for colour but I sample brightness from hdrs, maybe I should try and find a way of getting stuff like lux/lumens/candela into LW

I understand your frustrations about max, I'm using it at the moment for converting my enterprise-e and it's a pain, LW could handle things better too, for the most part I just try and set up realistic values and leave all the exposure adjustments for after effects because it handles it much better

this sort of setup does wonders for post effects too

EDIT

MadKoiFish wrote: »

Question too should I add spoiler tags too? There just might be some dial-up users out there still?

they load when hidden anyway, spoiler tags used in this way are just visual, just for keeping the thread clean when there are tons of images in a post

Chris2005 · April 2012

MadKoiFish wrote: »

I hate to bring this up, but that image name says transit ISS moon, that is the ISS not a shadow.
Read the article the image is pulled from earth based photo of the ISS passing in front of the moon.

http://www.universetoday.com/82013/international-space-station-on-the-moon/

Fascinating... well, that ISS image came up when I typed in 'ISS Shadow' in Google Images...

Never said you were lying... it's just that every source I've heard, from people who are very scientific people, have said that about shadows, that shadows in space are sharp, they never mentioned that 'lensing' effect.

The shadows I was referring to, is in the second render of the cylindrical piece, that's the kind of shadows seen throughout Star Trek, which I've also read, are often cited as being unrealistic to space, again from those I've talked to, something I read, or heard in a documentary.

However, I didn't know about the 'lensing' effect, learn something new everyday.

Melak · April 2012

IRML wrote: »

I understand your frustrations about max, I'm using it at the moment for converting my enterprise-e and it's a pain

photmetric lights take lumen, lux or candela values for intensity
, and kelvin for color; the photographic exposure lets you tweak everything a camera would and more. In what way do they not behave as you'd expect?

Chris2005 wrote: »

it's just that every source I've heard, from people who are very scientific people, have said that about shadows, that shadows in space are sharp, they never mentioned that 'lensing' effect.

The shadows I was referring to, is in the second render of the cylindrical piece, that's the kind of shadows seen throughout Star Trek, which I've also read, are often cited as being unrealistic to space, again from those I've talked to, something I read, or heard in a documentary.

However, I didn't know about the 'lensing' effect, learn something new everyday.

I haven't heard it called lensing, but it's simply what happens when your light is not a single point source, but instead a surface where each point on it is acting as a point source casting light in many directions

for example, a light emitting plane, represented by 1, 4, 16 lights, and an actual plane with 64 samples:

As MKF said, the softness of the shadows depends on the area of the light and the distance of the shadow casting object.
There may have been some artistic license involved in how large their lights were in star trek, if you are sufficiently far away from a star it will essentially act as a point light with crisp shadows, but generally soft shadows happen everywhere.

IRML · April 2012

Melak wrote: »

photmetric lights take lumen, lux or candela values for intensity
, and kelvin for color; the photographic exposure lets you tweak everything a camera would and more. In what way do they not behave as you'd expect?

I'm by no means an advanced user, but nothing works like I'd expect in max

Melak wrote: »

I haven't heard it called lensing, but it's simply what happens when your light is not a single point source, but instead a surface where each point on it is acting as a point source casting light in many directions

for example, a light emitting plane, represented by 1, 4, 16 lights, and an actual plane with 64 samples:

As MKF said, the softness of the shadows depends on the area of the light and the distance of the shadow casting object.
There may have been some artistic license involved in how large their lights were in star trek, if you are sufficiently far away from a star it will essentially act as a point light with crisp shadows, but generally soft shadows happen everywhere.

what you're describing there is umbra and penumbra, like you'd get with an area light

when MKF is talking about 'lensing' he means the diffraction of light in what would be an otherwise sharp shadow, which is a different softness altogether, it isn't related to the size of the light source, rather the light being bent as it passes through the object casting the shadow

it's most often seen with sunlight shining through tree leaves, I've not fully mastered it in LW, I imagine you'd need a physical renderer to reproduce it properly

andar_b · April 2012

Well, I didn't expect this to take off quite so well, but here's my meager attempt to get that Defiant mesh to render in a way that looks something like it did on DS9.
Defiant with Credits.jpg
I've added the credits to my compositing finally, as well. Forgive the strange registry, it's a UV map error I haven't fixed yet.

MadKoiFish · April 2012

The biggest point though no matter what creates it is that is that a shadow will fall off from sharpness, how much and how soon is the issue and that is it not 100% from the point source. Hard shadows as we experience them in cgi are a null point and very unrealistic IE light source emitting from surface 0.
Truthfully umbra, penumbra I think are in affect with the lensing as mentioned in IRML's comment (dappled light) The dapple occurs because the sun is not a point. Opening and closing of the "hole" focuses that projection. I think these are all related phenomena I just cannot find a online connection.

Atmosphere will tint your shadows. Now for breaking realism, I maintain this "hue" not blue per say but some sort of off black hue. It is a trick mentioned to me by others and something I found helped me match the objects to a scene EXP if using a pre-existing plate.

Photometrics so far for me act a bit odd with materials, and mostly are not something easily adopted from older systems. IE nothing translates over so you have to start over on most everything, materials and other settings. But worse is the times required to output for learning. IE render wait 5 to 10 min then fiddle a setting, render wait etc. :O

Ok subject matter.
Other than the odd lighting and lack of shadows, such as the bussard cowling, where the saucer ends inside it is very dark there should be a continuation of this into that cowling. It almost appears to be a render with no shadows.
But besides the light the textures are very cartoon like. All diffuse hardly any surfacing. I am not sure what you can do other than adjusting all the materials on the ship. Here are some renders I have of the same model in max.
You could do is desaturate the effect though post and via lighting in app.

Here are some passes I have done with the model.
I have opted for a cool cast to the model over the warm cast of your example.
first is a replication roughly of lighting via a single omni
Changes I would make to this would be to pick a more flattering lighting, something more dynamic and dramatic. The lighting here show some weak spots in the model or ship design namely the flat faced bussard cowls, it is too lit up in there showing a lack of details and the noise added to the textures in there, it also kicks up the light on the housing itself near the pulse phasers and the seam of the join of it to the saucer.

Image 2 I relocated the sun and adjusted the camera to a 35mm to create more distortion in the model so it was less flat, making it seem a bit larger. I also turned on exposure control, this took the colour saturation down some and allowed control over high mid and dark tones. (this can be done in post (photoshop etc) but I prefer to do it in app, LAZY lol)

Image 3 is just a more dramatic angle. There are other things Id adjust or alter like comping a AO layer ontop to jump details some to better match DS9 vfx.

MadKoiFish · April 2012

Post processing,

raw diffuse render
AO is added (I render mine as a full ship ao pass so white model white bg with just the ao shadow bits)
ship lighting is added, things like bussards marker lights windows etc. This is split up and blurred some to add a bit of glow.
glare is added (lens flares on the marker lights. I do this within max.)
Then I adjust the diffuse image and add light bloom to the ship.

Lens effects and icing. Not everyone is sold on these and everyone has their own opinion. I use it somewhat to give a signature to my work. But if one wants to mimic the DS9 as close as possible some chromatic abbreviation and noise will be needed.

I add noise, generated via after effects to mimic a specific camera. (I have not sourced ones outside of the defaults yet or found any resource of existing plates, if someone has resource please post. I will upload mine as a example.) I use this type of noise/grain because it seems more real than generated noise/grain from PS psp or gimp and the like. If you HAVE to use those generate it and Gaussian blur it some. (note mine here is subtle since I am not generating a special one to match the DS9 vfx, just an existing file)
next is the lens glare
then stars
this one is stars without the glare
chromatic abbreviation is added. I do this to mimic photos and the like.

With all of these things you can go into excess, too much of one or the other can ruin things. And always it is really individualistic choice too. Some people hate any lens flare effects, other loath bloom and yet others rage over chromatic abbreviation.
There are some issues for example the marker lights and some of the ship lighting does not take to my workflow and seem to need adjustment. Example the lights on the tail should be oranger the lights in that recess in front shouldn't be so strong etc and so on. But I wanted to show an example without altering the mesh at all. Really the age of this model and that it IS a conversion it is still fantastic.

random ds9 vfx examples:
some grime from print or video conversion is to be expected. But it is best to sample from the era of the ship or the look you are after. I know I have gone past the whole scale thing to something more but I think the whole is built of small things one after another that create the look or feeling one wants from a image.

DS9 had a really moody feel to the vfx, when the ship appeared in the films it had a really different tonality to it. The harder shadows crisper blacks and hight contrast between colour.

andar_b · April 2012

Another great post MKF, thank you for your effort. I have to agree that this is more than an issue of size, but creating a render that looks, at least to an extent, like we expect the ship to look on film.

I'm not quite sure what you mean by

All diffuse hardly any surfacing

At least in my renders, I've added Dodds' original bump maps, plus a light pass of an 'aztec plating' that gives a bit of specularity and bump mapping, but it's subtle unless the render is in close. I would love to add some of the streaking that the physical model has, but it would require texture painting the entire model and I'm not ready to do that yet. If I were to redo all the textures, I'd probably be better off modelling the whole ship, so I understand the topology, the conversion from Max to Lightwave to Blender has really left a mess in some places.

In my last render, there are a pair of area lights, one bluish and one yellowish, lighting from above and below. I can adjust the lighting, of course, which would clearly help.

Aside from that, in the Compositor nodes, I added a touch of noise and defocus, bloom for ship lighting and specular highlights, and very mild lens distort and chromatic abberation. I hadn't thought to crack open After Effects and add a real film grain filter, but that might help with the illusion. A few other things your last screencap shows me are that the white lights are actually almost amber, and the red nav lights are almost white in the center, whereas mine look like red LEDs. That might help a little. Overall, my ship lights are too bright. Also, that's one of the only shots I've seen that captures the inside of the nacelle housings from the rear.

Mine look good, but they don't have the oblong shape in the middle.

Pearse · April 2012

A masterclass.

The bloom is a touch too strong for my personal tastes, but I think you've got the level of chromatic aberrations just right. Some people go way over the top with it and I think in a few years it'll really date their images. You'll look at a render, and completely aside from the complexity of the modelling, or quality of the lighting, you'll see those chunky bands of colour at either side of the image and say "ah, so this is from 2008-2015 (or so)".

With regards to lens use, I recall reading that for TMP they used a lot of wide angle lenses for the model shots to increase the apparently scale, while also getting very close up. It's most noticeable in the Orbital Office shots.

And lately I've been taking to turning down the density of my shadows to around 60-70%, which can give the appearance of bounced lighting without any added render time.

MadKoiFish · April 2012

What I mean about surfacing least in that particular render is there is no visual tells about what it is made of. It almost looks like a gi only render. There is no sense of how glossy or matte the hull is if it is reflective or not etc. The patterning on this ship does not help much. (the ship itself not the model, some of the vfx shots are a bit toonish) The lighting is likely a cause of this since it seems to evenly light the whole model. It also appears neither is casting any shadows. My examples above just use a skylight to get a small touch of GI (set to .02) and a single omni. Now I have used more than one source to mimic say a nebula or some other fill light. BUT it was always at least half of the power. IE the sun or key light is set to a unit of 1 the fill will be set to say 0.4 to 0.1

I just use a noise layer over everything set on screen then adjusted for opacity. (how much or little) You do not need to use aftereffects to get this though I am sure someone has a lib of noise plates.
http://img812.imageshack.us/img812/4629/noise01.png
I need to locate the original "noise" image since this is 1080p but it should help a few without AE.

3ds max model

The lights, yes a fall off was used to produce that white to red look. The model is native to max, so my raw renders is how it was intended to appear by the modeler. Likely the "mess" is from the conversion from max to lw? The model uses a lot of optimized surfaces.

Reading the readme this is v1.4 and some things were altered, not sure if these were carried over to the lw poser blend files at all. Readme is dated 2005.

Yes, My last few posts are about how to compose a final scene but I felt that this info related to scale since it helps sell the feeling of size. If some of the studio shots sell size without reference to other objects or a background of anything but stars then maybe copying all the minute details is worth a study.

Chromatic abbreviation, I sorta boosted everything in these renders for visibility given I was rendering at 720p and final images are 720p. Though with mostly stars at the corners it is not as stand out as if I had bits of ship there or another object. Proper chromatic only occurs in the corners of an image unless it is a printing error etc. The images that fail to sell are the ones where it spans the entire image. Luckily for me photoshop has a filter for that. :P Prior you had to shift the colour separations and manually pincushion it etc so on.

I also recommend rendering out at least 2X the final image size. Least in my experience it makes things look less fuzzy in the end results. These examples however were worked in 1 to 1.

As with anything time takes a toll, I can bet in the future 10-20 yrs grain will be as passe as a sourceless lens flare is today. (good ol photoshop and the early 90s)

TOS was no stranger to extreme wide angle either. It produced some dramatic shots of the ship.

japetus · May 2012

Thanks MadKoi for the great descriptions and images! I too have struggled with getting some of my models to "feel" huge enough. I know it's a factor of lighting, camera angle, and lens, but it can be pretty tricky sometimes to figure out just what the problem is. These would be great bundles for a tutorial on the site I think.

Greetings!

How to make ST ships look suitably huge?

Posts

Greetings!

Quick Links

How to make ST ships look suitably huge?

Posts