[AndyT]

I tried to construct two cubes according to the tutorial @ teamgt.
http://www.teamgt.com/forum/cubes01.htm

But I don't know how it works if one cube is further away than the other???

Which of the red points are the correct ones?
(I guess there's just one picture plane )
And how do I construct the cube then?
_________________
http://www.conceptworld.org

[Sumaleth]

The lower of the two construction planes is the right one. Now you just project the red dots straight down.
_________________
Art Links Archive -- Artists and Tutorials

[AndyT]

Thanks!
They are projected down now. I know now how to find the vanishing points of the cube.
And I know how to guess what its closest edge would look like.
But how do I construct it?

[edit]I think I figured it out![/edit]
_________________
http://www.conceptworld.org

[Ian Jones]

Hi andyT,

You are getting closer, its just the height judgments you are getting wrong. See the angle of the lower corner, you should never let it get below an angle of 90 degrees. It should almost always be an obtruse angle otherwise it will be a distorted perspective. This wouldn;t happen if you used the correct system for height measuring.

I don't have the time to fully explain it atm but I'll come back later with an example.

[AndyT]

Ian Jones - I know. It's was just a quick sketch to show what I wanted to know.

I've been struggling with the construction maily.
Now I can slowly try to take the next steps.


I'm working way to sloppily and I always rush things ... I have to concentrate and take one step at a time. I can see a lot of errors ... no need to take that one apart .

And my computer doesn't like these 5600x4200 resolutions
When I change the canvas size I wait forever ... just to find out that it still isn't big enough. Frustrating!!! I worked on a new one for hours and realize now that I have to give up because the vanishing points are too far apart.
Life sucks.
I'll look at your cone of vision explanation and see how I can apply it to this cube practice. I think you did explain how to avoid that kind of distortion.

Seems to come in handy. I just don't find it ... damn I'm sure I had the url somewhere . I'll search ...
[edit]
found it
http://forums.sijun.com/viewtopic.php?p=154436&highlight=#154436
[/edit]

[Ian Jones]

"The closest edge is as high as the the cube in the front view (if it touches the picture plane)! Right?"

Correct, although... this edge is a short distance away from the viewer and as a consequence it has diminshed in size due to perspective, so on the drawing it isn't actually the full size (the full size one would have to be right in your face at the point of your eye). It certainly can be safely halved or used as a measurinng guide for any other lines parallel to it, unlike one of those receding lines which you cant just use a ruler to find the half way point. In techincal graphics and engineering drawing they call this line a 'true length line', Because it is parallel and directly on the picture plane.

"I guess the only thing I can change is the distance of the observer from the cubes."

It isn't the only thing you can control. You can control the height of the observer... because you can be viewing from the ground, like an ant or from the normal height of a persons eyes. This is what I was talking about with using a correct height measuring system. Much the same way you construct the perspective by using the top view and tracing from a point to the observer and put a mark where it intersects with the picture plane, you can also do this with a side view. My scanner isn't working so I can't draw anything and I haven't got PS installed atm either.

The only way I can explain it is by taking a look again at your second attempt. See where you have placed your horizon line? Well that indicates to me that your observers eye is at almost twice the height of the front edge of the cube. Remember, that the horizon line is the equivalent of the height of the observers eye. So in this case we really aren't looking at the cubes at all, we are looking above them towards the horizon. Because they aren't being looked down on (by tilting the picture plane) they appear distorted if you push them too far below the horizon. I hope that helps to clear things up.

Moving on to your matte white cubes now...

The most important thing to do is sharpen those edges. At the moment they are blurry and the cubes don't have much solidity at all. Use the magic wand tool and make some hard edged selections which you can paint within. The values aren't totally corect either. The shadows are also incorrect, you made an attempt though which is commendable. The cube on the left has the light coming from a different angle than the cube on the right, just look at the angle of the shadows and compare them! eek!

Moving onto values...

If you are familiar with 3D graphics then you'll be aware of terminology such as radiosity and global illumination. In simple terms these talk about the way objects are lit by secondary illumination (or third, fourth etc...), this can be called 'bounce light' for our purposes. In your scene you have a single light source. How is it then, that the surfaces not being hit directly by light are not completely black? well as you have quite rightly estimated they do infact receive light that bounces back at them from the surrounding floor, sky and from each other. you have done well to observe that fact. It does however continue in complexity, because you can't just assume that a flat surface is actually one single value (as you seem to have done). You'd be hard pressed to find a flat surface without some small amount of value variaton. Look carefully imbetween your cubes. The leftmost cube has a flat surface that is about mid grey. The rightmost cube has a bright almost white surface that a viewer can assume is the one surface in the picture that is catching the most light. The problem I see here is that the leftmost cube should be recieving much more bounce light on its right facing mid grey surface, Simply due to the fact that there is a really bright surface almost facing it. This tells us that is should be receiving light bouncing off the rightmost cube. Also, as I was saying the value of the leftmost cubes face isn't neccesarily an even value. In this case towards the back of the lefmost cube, that face would be recieving a little more bounce light than towards the front of the face (the part closest to the viewer). Imagine you were a single point of paint on the cube towards the back of this face. Your job is to figure out how much light you are receiving and adjust your value accordingly. If you look perpendicularily to your suface (straight out), you would be seeing a bright, almost white surface to your left middle and to the right you would be seeing floor and that shadow from the rightmost cube. Now imagine you are a single point of paint closer to the viewer, the front of that face. you would hardly see any of the bright white surface of the rightmost cube, so you would primarily be seeing that shadow and an expanse of floor. At this point your value is darker than that of your neighbours, on the same face of the cube but further away from the viewer at the back. moving on to another surface, consider also that a point of paint at the base of the leftmost cube, the face that is in shadow and almost facing us would not be able to see as much as a point of paint higher up the surface. This is how you consider the effects of bouncing light. You imagine that you were a point on the surface and you figure out how much light you can see getting bounced / reflected into you.

I hope that made sense. Due to the fact I can't get a diagram going atm this all may be a bit confusing.

Goto Sumaleths link archive and look for tutes about painting matte white cubes. There is a great thread there somewhere that talked about all this with some good examples.

Keep trying AndyT, everyone gets frustrated as you are now... the most important thing is that you have the determination to keep trying. I'm lazy, so I admire your efforts lately around the board.

Please ask for any clarification, because I know my writing can be confusing!

[Ian Jones]

EEK! I should have broken up that big block of text into paragraphs. Sorry.

[AndyT]

I knew almost everything you said
Didn't I say

[Ian Jones]

sorry.

I obviously didn't read it carefully enough. Your shadows are correct! don't change them. I was using the assumption that it was a global light source like a sun, that would cast near parallel rays, sorry!

I do wonder though, why did you post if you didn't need any help?

[AndyT]

[Ian Jones]

heh, np.

Gradients... use the gradient tool, or a soft brush building up slowly like 15% opacity.