The Atmospheric Railway

This thread is an offshoot of the blender guru competition entry that I did recently, and I’m going to try to address some of the issues that were raised,
particularly those of composition and of lighting. At the time that I was working on the reconnaissance Reichenbach I used several different camera positions to
show some of the assets that I was using in the background, and in the process came up with some good ideas for new pictures.
It is one of these that I’m going to develop here.



A quite like this one so I’m intending to use it as a starting point, to address some of the issues head on. I’m going to start with composition and then move on to lighting.
As one of the reasons that I’m doing this is to improve my technique, comments are welcome, indeed encouraged.
The name by the way comes from a project worked on in the 1860s’ for a different kind of steam railway by Isembard Kingdom Brunel. http://www.ikbrunel.org.uk/atmospheric-railway
Apparently it was a disastrous failure but I quite like the concept, and so here it is after another hundred years of development.

Excellent start! The mountains and the water both look fantastic. I also really like the lighting/atmospheric fog. A couple of critiques, though:

  1. The trees in the background look too uniform and cartoony. Consider using the Sapling add-on to make more detailed trees (it can be simple, though, for so many particles).
  2. The texture on the side of the train (the red part) is kind of grittier than I would expect it to be.
  3. I’m not really sure what’s happening at the front/back of the train (that gray part). In the future, try to make it clearer what the function of that section is.
    Can’t wait to see where this goes! Good luck.

I’m glad to see your doing some focus renders of the many great elements that made up this very elaborate scene.

  1. The trees in the background look too uniform and cartoony. Consider using the Sapling add-on to make more detailed trees (it can be simple, though, for so many particles).
  2. The texture on the side of the train (the red part) is kind of grittier than I would expect it to be.

Funnily enough the lighting was part of what I was criticised for, as being a little too flat and unrealistic (though I must admit I myself quite like it).
However as I’m teaching myself have to do this from first principles, I will be undoing most of this later on, and starting from scratch as far as the lighting is concerned.
The plan is to set up classic 3 point lighting system and then go from there. I also agree about the trees, but I’ll deal with them later on.
The texture on the train is a placeholder only, in fact the modeling at the moment on this is extremely basic, but I shall be working on that only after I have sorted the composition, and done my basic lighting set up.
Last night I followed this tutorial by Andrew Price,http://www.blenderguru.com/tutorials/mastering-lighting/
Which I found extremely informative, particularly concerning the historical aspects.
I then went ahead and searched for the golden section and rule of three, and came up with these sites

(among many others). As usual where the Internet is concerned there is loads of useful information along with some of rather more dubious provenance.
Anyway to start with I shall deal with the golden section, and as far as possible get my picture to conform to it.
I realise that under normal circumstances you wouldn’t stick to this quite so rigidly but on this occasion I will do so
As that is the point of why I’m doing this in the first place.
(god that sounds pompous!) :mad:
Now by the time that I had finished “Reconnaissance to Reichenbach”, the view from this particular camera looked like this.



I realise that the towers were not in the original picture but on balance I think I shall keep them there.
This is the golden ratio diagram,


Put them together and they look like this, as you can see nothing lines up.


So now I am going to attempt to make the picture conform to it more directly.
(I’ll be back):evilgrin::evilgrin::evilgrin:

So here we have the same picture, now everything lines up with the diagram (in actual fact I did a lot of this last night, I am not really that quick).



I moved the camera so that the line of the railway now ends at the centre of the spiral, I also added what will later become
A station terminal, and I turned around the airship and made it smaller so that it fitted in more harmoniously with the general diagram.
Do you think it looks any better?
(personally I do but it will be interesting to see what other people think of this before I go any further).

It is a big improvement to the composition - not that I am an expert, but it just feels right.

Keep up the good work.

Cheers, Clock.

Good work :slight_smile:
Just one advice: while the “golden spiral” diagram is useful, you should mostly try to get a balanced composition. You shouldn’t be overly moving things to have intersections: if something looks good in a position, don’t put it on the spiral if it doesn’t actually make a better overall effect.
For example, i think you should move the train back a bit, or better add a wagon, so it gives a sense of indefinite (the viewer doesn’t know how long it is) and you don’t have the problem of making its back :slight_smile:

Right now, the picture looks quite good. I’d also like to note that you followed a method that is not described in your diagram: the rule of diagonal. You have a “not so accentuated and precise” diagonal going from top left to bottom right that divides the scene into a bright “sky” zone and a darker “land” zone. To me, it makes a good effect :wink:

For example, i think you should move the train back a bit, or better add a wagon, so it gives a sense of indefinite (the viewer doesn’t know how long it is) and you don’t have the problem of making its back

I like this idea so I’ve put it in.


Right now, the picture looks quite good. I’d also like to note that you followed a method that is not described in your diagram: the rule of diagonal. You have a “not so accentuated and precise” diagonal going from top left to bottom right that divides the scene into a bright “sky” zone and a darker “land” zone. To me, it makes a good effec

I was totally unaware that I had done that, I have to admit that the method I use is to move a camera around an already set up scene, until I see something I like, which is what I did here.
Having said that, it is looks good for exactly the reasons you stated, so in my case it must work subliminally.:o

Actually I think it looks worse, and it took me a minute to figure out why. The vertical elements at the left of the pic (which are the most in-your-face) are way off vertical. To my eye, this overpowers the composition and results in a worse overall effect.

I have to say that I do like the original pic in the OP too. I feel its main failing (apart from minor stuff like trees and train texture) is that the way the rail and mountains and shoreline are drawn gives the feeling of an amateur landscape shot where the camera hasn’t been held level. If it was changed a bit so it felt like the world was sitting level, I think it would be a lot better.

Actually I think it looks worse, and it took me a minute to figure out why. The vertical elements at the left of the pic (which are the most in-your-face) are way off vertical. To my eye, this overpowers the composition and results in a worse overall effect.

Here it is fixed,



I had to do this in photoshop using the lens correction tool, as the original distortion was caused by a combination of camera position and focal length. I tried moving the camera and adjusting the focal length,(for quite a long time), but eventually gave up, time to adjust in photoshop, about 40 seconds.
Obviously if I used this solution, then it would have to be at the end after everything else was done.
There is another question raised here, this distortion arises as a result of the use of technology i.e. a camera, and if you were going for photorealism, you could argue that it should be left in, as in this picture (taken by a proffesional photographer)


A renaissance painter would not have understood the concept of lens distortion, and would simply have corrected the picture, but in a physical sense it would be less realistic. Which leaves me with a dilemma.

  1. Do I adjust the picture.
  2. Do I leave it as is
  3. Do I abandon it and go back to the original image I posted. (I still have the file, but I would loose about 8 hours work, which was dedicated entirely to learning the ins and outs of composition.
    Crunch time
    What would you do?

to Gumboots, I am taking your comment as constructive criticism, so this is not sour grapes, I also realise there are as many opinions as there are artists!

My 2c: I’d go for what looks good. With the verticals more vertical I think it looks a lot better, for that picture at least. Not saying this would be the case with every picture.

Or, perhaps try something halfway between the two, like just enough off to make it feel like “real distortion” but not enough so the eye picks it as off, if that makes sense. Art is mostly about illusion anyway.

I think the ability to make such a decision is probably what defines a good artist. My problem tends to be that I always get so caught up in the minutiae that I tend to loose the bigger picture.

It’s a pleasure to follow your studies and steps.

To me the main difference between #4 and #5 with the Fibonacci spiral is a) depth and perspective and b) focus and guidance. In contrast to #4 the different objects in your image #5 appear to be at the same and relational correct scale and therefore the image gains more depth of field and realism. And the second big difference is the guidance it gives the viewer. The eyes are now wandering from the train over the airship to the city. The first tower is a bit of an obstacle, but it still works well. “Turbulence” has gone and the image is coming along more pleasantly.

Most probably you are right in that a painter would have drawn the poles completely upright. And that it is a photographic effect to see the poles leaning inwards. I’m mixed about that. But I tend to the “photographic” version, because it’s more according our modern viewing habits. Perhaps you could minimize that if you put the camera farther away and use a longer lens. But I understood that you tried that already.

Thankyou Minoribus! I have seen that there is a lot of enjoyment to be had as well as learning by following other peoples threads, (I have been doing that long before I was confident enough to post), so now I am attempting to contribute in return. It is very useful to get more than one opinion, and there is a lot of experience on these forums.

I think that I will go with the current version and carry on to the lighting. The distortion will remain for the time being, and I will fix it at the end in photoshop, (I will probably reduce it rather than get rid of it altogether).
As this is not a competition entry I have some extra freedom in using other programs, and that is usually the way I work.
As an example, this is a picture I did a couple of years ago as a book cover. In this I used Blender to sketch out the landscape and buildings, but then worked it up in photoshop, so by the end it was about 20 percent Blender and 80 percent photoshop.


The difference is now that I am becoming more confident in Blender, so I am looking at reversing that ratio in this current picture.
As an interlude While waiting for replies to the last render, I’ve done a bit of work on the railway carriages. I’ve removed the texture, for the time being, so it is just a default grey diffuse, but will work on that later.



So on to the lighting!

Good changes on the railway carriage. I think you made a wise decision in #10 to move the train up so the front is out of frame, I just like the way it is now composed.

The book cover looks exceptional. I think the end result is what matters, what software tools you ultimately use should not be a big factor, it is the end result that matters the most.

Okay, after having followed Andrew Price’s lighting tutorialhttp://www.blenderguru.com/tutorials/mastering-lighting/
and also this one by blender HD,


I have set up A classic 3point lighting set up for this picture, So I’m putting up a series of pictures showing the various component parts and then the varying combinations.
There are a number of issues that I want to raise here, because I’m not 100% clear on what certain aspects of blender cycles are capable of, and other people’s experience might be really useful to me on this.
The first concerns environmental lighting. This is the default that I had for this picture as I started with the sky made of a 360° environment map.( I have removed the textures as per the Blender HD tutorial as it enables the viewer to see where the light is falling more clearly, and I have also turned off the compositor to remove the mist).


I must admit I find this one of the most powerful aspects of cycles, being relatively easy to use and giving very realistic results. Obviously you are limited by the photographs themselves (in my case the ones you can download free from the Internet). But I am looking into modifying these or even creating my own,
as I am fortunate enough to to have an educational copy of Photoshop CS 6 which allows me to work with 32 bit images albeit only with a fairly basic editing functionality (It is still enough to do cut and paste or create or modify full HD images).
However for the purposes of this thread I’m going to set up a classic 3. Lighting array.
As follows.
Key light.


Light set slightly yellow to simulate sun shadow set to .25
Fill light.


Light set slightly blue to simulate bounced light from atmosphere. shadow size set to 1

Rim light on next post, because of 3 picture per post limit. Also slight delay as I have to go shopping:mad:

Ok here is the rim light.



There is a problem with this that I originally encountered in Reconaisance to Reichenbach, in that the light whose source is invisible is casting a strong reflection in the water. I got round this by unchecking glossy in the ray visibility tab which I have also done in the combined image below.


I think that this is looking ok as per the tutorials, but maybe it lacks drama.
Now I have also added a picture combining the 3 point lighting with the Environment texture,


is this gilding the lily I wonder? (the sun is in the same position as the keylight so the shadows line up). If it is too much, then is there a way of using the sky without it cast its own light, or is it better on its own without the 3 point lighting. My guess is that I will probably juggle with the strengths of all the lights to get a balance.
I am also minded to reduce the strentgh of everything so that interior and external light sources would stand out more as if it were late evening.
Also I have unchecked ambient occlusion, as it strikes me that it serves a similar function to the fill lamp.

Something along these lines maybe


Hi Carel

This is looking so good already, but that texture you are using for the sky (is it a HDR?) does not look correct.

Shaun

I second that. The sky texture looks like an approaching sand storm. That’s too much drama and it steals the show. And the bright areas in the sky look like the sun would be there - which is the opposite direction of your key light. I’m sure this texture is pumped through some nodes. How does it look originally?

Awesome book cover, by the way :slight_smile: