Tutorial: Using a Raytracer for texturing

[AFTRS]

Tutorial: Using a Raytracer to create texturing

The idea to use a raytracer to generate any texturing material was taken into consideration if you do not have:

1. any usable picture material at all for the desired object
2. the object doesn’t exist in reality
or
3. you need some new materials designed
4. it is too complex to do it with an picture manipulation software

Why an raytracer:

The object or material you want to use is created in an “laboratory” which then can be used later on as an object as if in a real “Photo-Shooting-Session” indoors.
That means placement and illumination can be set accordingly to the special needs as if it wood be reality.
That’s was might be the difference to any picture manipulation software. CS4 is something in the middle.
In the end it is dealing with the question of low “paint-requests” to the graphic card.
If you own 3DSMax you can export to both, the small polygon-count 3D-Model with the more detailed texture using high polygon count rates on it, exported as an in the end reformatted DXT-xxx .bmp.
OK, if you have only GMAX and a picture manipulation software it would be very hard work.

I am using GMAX 1.2, Corel Photo Paint 12, Adobe Photo Elements 2.0. For those I have regular licensees.

Talking about points 1.. and 2. no CS4 will help you out.

So let ´s have a look what it might be useful for:
If things do not exist you can’t have any picture material.
I wanted to generate an 2.5 ton truck made in the 40th ´s to the late 60th ´s by both US and GB. In my scene it would best fit as a GB truck.
For the reasons of modern lawyers and the question of “licensing” it is sometimes useful not to be to exact.
Secondly I did not have any useful pictures of the original.
So I decided to implement something that would be “characteristic”.
Not to far away from reality but also not to be to exact on the real thing.

Pictures:

The truck in GMAX, not yet textured in all parts, the cabin will be the interesting part in this tutorial:

So, let ´s start with the cabin designed in the 3D-Frontend:

Being honest, this is much more easily done in the Moray 3D front-end software than in GMAX. These are only some differences, unions and intersections. Each part is simply to select. In GMAX the concept is different. But I don ´t want to claim about it. Let ´s stick to the theme.

The above cabin has not jet any “Glass”-Windows implemented, just holes.
This is done by intention.
We will see in a moment why:

The same cabin, just “walking” around with camera:

The difference is the illumination. It is the same object, with a different camera position but the same light taken.

Ranges therefore from ok to not acceptable.

If you change the light, you will get better results.

What we need is in the end a good picture from front, left, right, back, top.
So we will need to separate the 3D into Front, Back, Left, Right data set, using the appropriate camera position.
What I figured out concerning the window mapping, that there is an additional dirty trick:
Take a second picture of the according window with an angle of X/Y=15 and Z=45.
So, may be you might create a second data set for each of the above.

If You are interested in reading more about that kind of starting-point / way of implementation:
I would be very impressed if anybody would take care about it.
If for any reason yes then I would carry on with that thread.
Would you like to read more about it?

 

[AFTRS]

Part 2

Tutorial: Part 2 Using a Raytracer to create texturing

Thanks at the moderators for correcting the misspelling in the title!

In part two we will have a closer look on the “how to do it” in detail.

First off all we need some windows. This is an easy task, as the window holes are object differences. We only need to copy the “cutting” part and adjust the “thickness” and the material.
Glass is a standard texture with a pigment and a finish section. In the pigment section we define solid colour with at least 50% transparency, some filtering and a finish section for the reflection and ambient parameters. I added in the finish section some 3 percent of being metallic.
This is more of less identical in all raytracers. Works the same way in GMAX.

(Remember we can´t use it inside of GMAX natively on the object because of the polygon count and/or we cannot export to a xxx.bmp. Off course we could make a screen shot in GMAX to import it into any picture manipulation software, thus not having the possibilities to change resolution, anti aliasing, raytracing parameters. They did cut to much parts from 3DSMAx there. What way either, the principle stays the same for all of the software used.)

My standard resolution of the rendered images is 800x600 if I am using a 512x512 bitmap for texturing.

Don’t put too much work in the glass material now. AND YES: don’t use the background of the picture to simulate the glass, we will see in a moment why!

In my case it looks like that:

Now we change our profession from an engineer to a photographer.
We need 10 pictures, 5 for the day texturing and 5 for the night texturing:
Each of the four sides front, back, left right and the top.

YES, that is a second benefit of doing it that way. The raytracer will calculate all of lightning, shadows, reflections and so on. Much more precisely than any mask in a photo manipulating software could to. And we have nearly no additional work involved.

It doesn´t matter what part you start with. I did start with the left part.

Position the camera at x=0 and somewhere on the y axis so you get the object as big as possible concerning the z-Axis. Now adjust the “look at” point until you have clear lines and no dithering effects, falling lines, things like that.
Position the light somewhere on the y-axis with X=0.
Turn the render-backgroung to full white.

Make a test rendering. In most cases you now have to adjust the lights and materials slightly.

A word on the lights:
In most cases it is sufficient to use one light and adjust it to get only those shadows needed for for the 3D-Feeling. Then adjust the material with the ambient factor. If you have more than one light you get more than one shadow, what you might not intend at this point.
For the night texturing you might need of course additional lightning for lamps and so on. Using spot- or aerial lights turned out to be very practical.




Now, lets say we have the day-texturing done for the left side.
I then make a copy of the file for the night texturing.
Turn the background to be nearly black, not full black.
The reason behind that are alpha channels you might use later on. If you won’t make any use of them you do not need to take that into consideration.
In the example used here the colour of the cabin is very bright. So I deleted the light completely. The outer surface then still was to bright.

This brings up a different point of view:
The way you construct your 3Ds: If you make a difference the polygons of the original part will have the material of the cutting part at all of the sides that had been cut.

In the above example a “outer” cube had been cut with a “inner” cube. I have defined two materials for that. In the day texturing they are identical.
For the night texturing I now turn down the ambient factor for the outer part of the cabine.

In addition I did position an aerial light into the cabin and changed the light colour to be something like “Sand”. Then I adjusted the windows colour and transparency.

The results:

Now we have done already most of the work.

You may ask: why that?

Ok, here is the answer:
Copy your left night file to make the right part of the night texturing.

Now, select the camera. Change the prefix of the y-axis of the camera itself and for the “look at”-point to the opposite. If it is ”-“ make it “+” and vice versa.
We only have an aerial light inside the cabin, so nothing to work on with the lights.
There you go, second one ready.

Do the same with the day file, but also change the light parameters.

Results:

Next step is the front or back.

I decided to take the front.
So I copied the left day for the front day.

As my object does not have to different dimensions I started to copy the y-parameters of the camera and the light to the x-parameters and adjusted the y-parms to zero.
Then I adjusted the camera position very slightly to meet the best position.

You should pay attention to the colours. Because of the light position they can vary from the left and right side. Try to adjust the light and/or the ambient parameters so that you end up with a more or less equal colour in comparison to the both sides.
Some changes to the glass texture will be necessary, too.
If you like the result render it.
Now it is a good idea to open the left and front picture in your picture manipulation software.
Take the “Get colour” tool and check the colours in both pictures. This can help you to find gaps.
They only need to be “very similar”, not precisely identical.

The result:

Now we need the night texture for the front. To get the identical view we export the camera. You can do that because they it is treated as an regular object.

Copy the left night file as the front night file and open it.
Delete the camera and then import the exported camera.
As already mentioned this ensures the same view for day and night texturing, so it is much easier to position it later on in the xxx_LM file.
Adjust the colours and materials as need (always the same procedure than above).
Result:

OK, there you are.

For the back side we use the same trick as already done for the right side. We copy each of the front files as the new corresponding back file and then change only the x-parameters of the camera. This time we keep the y-parameters. Never change any z-Parameter.

Results:

The top view isn’t that interesting, it works the same way.

If we have a look at the mirrors it might be interesting to get them into the front view, or being able to see the other objects in FS through the windows. That´s part of using alpha channels. I did use them for the chains of my water tower, stairs and things like that.

If you like this could become a part 3.

 

[rhumbaflappy]

Hi Christian.

Yes, please continue the tutorial.

Dick

 

[AFTRS]

Thanks for the feedback,

anyway there are some additional things to be stated that one should avoid:

Taking the most worst possible angles of any view in FS into consideration things might turn like this:

Seems to be some how ok,

Turn the airplane and you end up with a mess:

So, if you want to have objects beeing able to be looked at from a nearby position under an angle of 45, 15 degrees avoid that duplicate windows from left or right, and so on and so on.

I had a DC3 placed next to it, looking out of the left window:
Works perfectly. Push it back some 10ths of meters, uups strange picture.
That has nothing to do with the method taken above, just the way FS represents the polygons.

 

[AFTRS]

Part 2.5

This is out of development an intermediate Part 2.5.

As mentioned above I wasn’t that happy with the way windows are represented.
So, for the cabin of an excavator, used as to be part of a crane truck later on, or as an individual machine, I designed something like the “Fuchs 300 cabin”, which has been very famous in the past in Germany.

Some pictures of only the cabin at the current stage:

Through the open window at the right, you can see some part of the white-blue truck in the background that I didn’t design at all. This one is just used as a very, very good reference, nice work from the college.

Continuing with the pictures:

The front window isn’t at all at the stage desired. Have some GMAX problems of material design. It is to “clear”, to “clean”, than to be used.

What I think, that using a good mixture of techniques is the way we are looking for.

The “pivot arm” that will turn round the rope of the heel in the back center will then be in alpha channel as also the rope. Don’t know jet if the arm will also be in alpha channel design. Think it would work, not sure at the moment.

Keep you informed, I just keep on trying.
Alpha channel in most cases means then ray tracing as I could implement it as simple planes.

Always keep in mind; These are screen shots of a Radeon 9000. So very old, but it works on that card without flickering, I think it is ok.
Things will look much more perfect on better grafic cards.