Road Trip 2013

[gregsandor]

I'm in with my rural highway that I've been working on for several years.  You can see the old WIP from the previous contest at http://www.planetside.co.uk/forums/index.php/topic,10141.0.html

I haven't kept a detailed log, as I mostly work on this in bits of free time, and a lot of changes aren't noticeable since I last posted, many of them are things like better masks, a rebuild of the road system, new buildings and signs, new shaders etc.  I completely rebuilt the diner and have added some new buildings.  I originally painted out the large Industrial Pallet yard down the road from the diner, but the more I've looked at it the more I like it, so I'm putting it back in (have to dig out the cornfield I put in its place!).

The terrain now covers 12 km x 13 km, with 2-lane highways, lesser paved roads, and dirt roads.

Here's a render from last night with the new diner model.  The pallet yard is going in just beyond the treeline on the other side of the diner at the side of Hwy 52.

[attach=1]

So I've set up the Pallet Yard, built some tractor trailers and started ont he stacks of pallets, and just now ran across a news article with a photo of it going up in flames, the pallets and the building burned down.   Hmm, now what?

[gregsandor]

...

[Dune]

Nice to see this one again. Already very realistic. But what about the grid that seems to be present in the grass. Maybe due to a mask?

[Oshyan]

Agreed, very glad to see this again, and in a later iteration. It was one of my favorite earlier projects. The attention to detail and the sheer number of assets and scale of the scene is impressive.

I see "grid lines" throughout various areas of the scene actually. I'm guessing it's due to image-based texture tiling. I also see some softer-than-they-should-be (IMO) elements such as the tar seams in the road, which might also come down to image-based rather than procedural approaches. I think a mix of techniques is going to produce the best results here. Certainly image-based is going to be a key component of absolute realism, but Frank has demonstrated some terrific road wear and tar seams done procedurally, for example.

Lots of potential here, looking forward to seeing it develop.

- Oshyan

[gregsandor]

Nice to see this one again. Already very realistic. But what about the grid that seems to be present in the grass. Maybe due to a mask?

Oshyan called it:  I made the grass textures from photos I took and they were some of the first I made (several years ago) so they aren't "flat" enough to avoid the tiling.  Evening that out is on my list of things to do.

If I can find a way to place procedural shaders by masks I might be in good shape.  As of now, where you see a tar seam for example, is where there is one on the real road.  What I'm looking for is a way to use a lower resolution mask to place a clearly defined seam.  In the back of my mind I'm also looking for a similar combination method to more efficiently do the road paint.

Greg

[TheBadger]

I would love to see something like the OP image in a post apocalyptic treatment.

[FrankB]

This was also one of my favs in the last contest. And yes, why not use it again and perfect it? Good idea!
There are a lot of details to look at in this scene.

About the road texture: it's really simple to texture this procedurally, especially from this distance. If you have a mask for the road area, that is.

Cheers
Frank

[Hannes]

Can't wait to see your next images!

[gregsandor]

I'm trying to match the TG camera to a real world example.  In the attached image I have posted the EXIF information from a photo of the real location and linked each of the settings I've used in Terragen.

I want all the lines in the original real-world photo to match those of the model.  The test is successful if the camera settings are matched and the render overlays exactly on the photo.

Are these settings correct?

When I use 18mm instead of 27mm the perspective seems very far off.  I know I'm missing something since the 27mm setting is close but not exact either.

[Tangled-Universe]

I see that the reference is shot with Nikon D90 which has a sensor size of 23.6 x 15.8mm.
The TG camera has 36x24mm, so I think you need to adjust that (anyway).

Did that help or improve?

[gregsandor]

I see that the reference is shot with Nikon D90 which has a sensor size of 23.6 x 15.8mm.
The TG camera has 36x24mm, so I think you need to adjust that (anyway).

Did that help or improve?

Thanks.  I think it is an instant improvement.  Testing now.  So that entry for Film Aperature = Sensor Size?

[Tangled-Universe]

Yes aperture size is actually film size.
The TG camera simulates a typical 35mm film, which has dimensions of 36x24mm.
There are derivatives of 35mm like 24x18mm, but as you can see the Nikon D90 sensor size is just a little bit different.

These so called 'crop factors' for camera sensors make it that focal length is not equal to the 35mm standard, or 'full frame'.
The crop factor for a Nikon D90 is ~1.5.

Say you have a default TG camera with 31mm focal length and 60 degree FOV.
If you change the aperture/film to Nikon D90 size = 23.6 x 15.8mm then the crop factor ~ 1.5.
The FOV will change, since a different film/sensor doesn't make the lense zoom in/out any different.
It's the area of the sensor which changes and not the focal length.
So if you set FOV back to 60 you'll see that the focal length = 20 and 1.5 x 20 ~ 31.

The discrepancy in the calculation is probably because the crop factor is not exactly 1.5, because for that we would require a 24x16mm film/sensor.

In other words: with a smaller sensor you need less focal length for getting a similar FOV.
If you buy a 70-200mm telelens for 35mm/full frame then it is 70-200mm, but for crop factor 1.5 it would give you 112.5-300mm.

Briefly, film/sensor size affects effective focal length and FOV.
That should explain the difference.

[gregsandor]

So far here's the result of my camera matching experiment:

[Oshyan]

Very impressive!

- Oshyan

[gregsandor]

Getting the TG camera matched really made it easy to adjust the sky color and sunlight intensity to match too.