Realistic Space Landscape Project


A New Image

Hello all,

I have been using Blender for slightly over a month now, and thought I might would do well to come here and ask for general advice and critique.

The project which I am working on presently, is an attempt to create a realistic scene of a space landscape, based on understanding of actual geologic process, and not centered exclusively on a vehicle, although I have been considering the addition of some sort of unmanned lander vehicle to the scene at some point.

I have this so far:



The scene consists of a region, square, that centers on a large crater basin, with a central peak and raised rim. I have added onto the terrain several layers of craters, that the large one is somewhat old, but less covered than the modest area outside it. Smaller craters, as well as general surface roughness, originates from a bumpmap, but the majority was achieved with displacement modifiers. Subsequently I added a particle system of boulders.

The planet is created using several spheres, and volumetrics, this is not the best image I have of it, and it still is under work and variation.

Generally I am looking for comments, advice, and critique. Any tips are welcome.

Image rendered in cycles, texturing in GIMP.

I have been experimenting with generating craters in dynamic paint, using the wave and displacement settings. So far, only the former has really worked, waves kept causing great quantities of concentric rings, but not like the Mare Orientale, more like a sphere shaped puddle. And with too big of central peaks, I was quite unsatisfied. The displacement modifier works well for simple crater forms, and these two asteroids were cratered thus. The smallest craters are bumpmapped, but the bulk of it is geometry.


I also have been working on redoing the boulder population from the last scene. As I saw it, it had too few boulders, too big boulder, and too identical boulders. However, the next attempt to add many, small objects of differing sizes and shapes, caused Blender to run slow, and crash. A lot. I had, I think, too many polygons, and am looking into maybe doing it instead with a combination of bumpmaps for background, and actual particles for closer up.

Also regarding cratering, I am looking into how to create secondary craters and ejecta blankets/rays, but have thus only started and have little to show. Preferable this could be done through a in-Blender means. Maybe with the spreading of a dynamic paint colour to define the different texture and bumpmap density.


My latest render went fairly well I think, most of the modelling and texturing seems to have turned out nicely. I was a little concerned still with the boulders, but they seem generally to be working well as bumpmapped features only. I should add a few more mesh ones, especially in the foreground, still .I worked on getting a star together today, and am curious if anyone has a good idea on how to produce realistic stars for such scenes. This one was an image I generated in GIMP, and used to create an environment texture, featuring a newer version of the gas planet, here not seen. All of the craters now, from the bumpmapped ones to the large crater were created using dynamic paint.

It still seems somewhat lacking, and I am trying to figure out good ways to improve the overall image. As before, any and all critiques are welcome.

Have you considered trying a volumetric shader? could be interesting. maybe composite it with your current shader.

For me the best advice I can give to anyone is always work from reference. Find some good references images pick on, or pick parts from some of them that interest you and work by copy those part via studying them.

A first comment would be that your craters like crispness and sharpness, of course you need a lot more detailing if you aim for photrealism. Its also good to start with a concept, a quick scetch helps a lot unless you work from a specific reference photo.

here some references of moon craters

the last one is quite big so I give you the link , its also quite interesting if you care about a close up

http://img.phombo.com/img1/photocombo/2037/Moon_craters_Sabine_and_Ritter_from_Apollo_11.jpg

Obviously lighting will play a major role in the scene, in space because there is no atmosphere shadows are generally more sharp.

My post just deleted. Oh well. Thanks for commenting.

Kilon- I have been looking at references, perhaps too many, not at all focusing in on getting the details of one image or scene to manifest. Sketching out how I would like it too look, also would probably be helpful.

The sharpness I can try to improve upon, I can likely find a way in dynamic paint to do that (as it would not be feasible to say, add a sharp edge to thousands of craters, by hand), and will look into how to go about doing it. I wanted some parts of it to look a bit old and worn down by the incessant bombardment, but the crater rims do indeed look too smooth.

Extreme detail is hard to get, and I cannot in this image feasibly add another level of subdivision to facilitate heightened detail for close up shots, my computer is near its apparent limit for polygon count as it is. I was toying with the idea of building a separate object, a part of the rim, say, that would exist as basically a platform for the camera to sit on. There I could stick some boulders et cetera, and add a lot of complex details that cannot be achieved everywhere in the scene. But that has not yet advanced beyond my imagination.


The region as it was at the time of the previous posted render, showing the blister like quality of the crater rim, and the apparent lack of slumping inwards, despite the apparent age. The peak also looks abnormally smooth, very much different from the peak complexes of Tyco, or King craters. Illumination from the top.

Modron- Regarding the star, I did at one point try a volumetric star, but its success reflected the brief time I spent building it (that is to say that it looked really bad, but that it could likely work, and I shall try again). Currently, the star never existed in 3D, built from the start in GIMP, I gave it a sun lamp (lined up painstakingly) to add brightness, and illumination comes mostly from that, with a little also from the planet. I had not thought to try it in the compositor, but can look into that soon.

I was actually referring to the planetoid surface. I thought it might help give the look of fine particles.

Oh…

That is quite a bit different, sorry I misunderstood. Would that be using a thin layer of, say, extruded surface, using a high volume scatter density, or something else? I had also not really thought of that, but did try SSS, with generally poor results, too much reducing the shadows that I was going for.

I am fiddling with a volume test now, (first attempt looks, a little like saran wrap, I think…). One concern might be trying to make the scene still look sharp, as volume effects seem to cause a level of fuzziness. Perhaps in some areas such fine particles would do quite well. Thanks.

how about blending the render results in the compositor so you get the fuzziness of the volume and still get some fine detail from the surface shader

first take a deep breath and relax …

dont be so quick to give up on something. There are ways to make even a slow computer produce extreme amounts of detail, you just need to do what every great artist does… cheat.

There are several ways a) bump maps, can be painted and wont use your cpu , its possible to convert also actual geometry to bump map b) normal maps, are even better for details that stands out a bit more c) sculpt brushes, again geometry can be converted to an alpha map and alpha map to a sculpt brush, sculpting can detail a lot easier with high detailed than you usual mesh edit mode.

And the list does not stop here, you could use matte painting techniques that was described here in the 2014 blender conference

I am actually a huge fan of matte painting, just google it digital matte painters can take photograph and using the stamp/clone tool and other painting tools can digest photos down to pixel level and completely transfrom them or create their own from scratch with photographic detail. In 3d it can do wonders. Because Blender not only has stamp/clone tools, not only it has nodes for both materials and textures but a full blown compositor and sequencer that cut down your render times to an instant even with a super slow computer. Of course it does require a diffirent workflow and careful planning , plus several steps. So its not just a matter of hitting the rendering button, or raising your cycles samples. But then everything comes with a cost.

Actually many of those techniques are regular used in film industry , you can see them for example in the photorealism of Jurassic Park even though the computers they used at time are ancient compared to modern render farms.

The rabbit hole definitely goes very deep.

Yes, cheating/faking is the mark of a true artist, I have done a half year of work on an old Pentium 2.3 GHz (very slow) with no Gpu Capable of render in Blender and believe me, I needed a lot of cheating to get my projects to finish.

Keep up with what your doing, you will learn a great deal,

Jim

To Kilon and Journeyman. Wow, I certainly have my plate full of different techniques to try here. So far my bump maps were a couple of layers of simple grey scale maps, mostly made from the default dynamic paint setup, two for craters one four surface roughness, and one for boulders, but not like this. This certainly enables so much more.

Modron- I feel really silly, completely failing to understand what you were talking about at first, but I think that I finally got it; changing the materials to volume, re rendering the new scene, and adding/mixing in the compositor. Even if that is not right, I feel as though you have pushed me over the proverbial cliff in this rugged, complex learning curve that is Blender, unlocking an enormous range of new possibilities in the compositor.

I have a lot of work ahead of me,
Thanks all.

yes you got my meaning correctly. I don’t know what the result will be, because I have not tried it, but I am happy to be the person to have sparked your interest in the compositor. I avoided the compositor for a long time, myself, but I am not sure why. I have a bit of node anxiety I guess, but I am using it more and more, and it’s definitely a powerful tool.
(edit) it occurs to me that material nodes might be an alternate approach to the same end.
(edit) I just tested that out and it looks like the volumetric shader doesn’t work with material nodes for some reason. unless I’m missing something. It’s sure to work with composite nodes though.
(edit) this is the result of compositing of the default material + default with volume shader

So. I have been working quite a bit with the normal maps, which are quite amazing indeed. I had used one once in an Andrew Price tutorial (the bread one), but did not really know what was ordinarily viable from other methods, so the tool was hardly remarkable then. Now they seem incredible.
I had a number of issues where I thought that parts of the texture using the normal texture looked… really weird, though. I was unsure, however, whether that resulted more from it being really odd, or from my having looked over and over at slight variations of the same exact terrain. I hope it is the latter. Please tell me if you see it in the images below, both of which are the same model using a low(er) poly mesh and normal mapped texturing, in addition to some bump maps which add general surface roughness and boulders.


1)An image taken this morning showing crater 13 from a low angle. Small boulders are visible in the near and far ground. Inexplicably, most of these appear to be perfectly aligned squares.



2) Another image of the same region at the same sun position. Readily apparent are the abnormal wave like linear features. Apparently similar to sand dunes, these have no immediate explanation due to the thin atmosphere, but may be of volcanic origin, or similar to lunar scarps.

I have continued exploring the compositor, but have not achieved particularly grand results with it. Possibly I can readjust it and get the sort of softening effect you suggested (Modron), but so far my best results (to my surprise) resulted in adding faint colours to the terrain, blue and orange patches… why I do not know.

About the suggestion of using sculpt brushes, I made a couple, but they were fairly simple, and worked to either pull, or push the mesh only. Not both. I am looking for how to make more complex ones, and it seems promising in the longer term. At present, I only really changed how I went about building the large crater, where I have moved from being mostly dynamic paint, to a mixture of proportional editing, to create the essential shape, and sculpting after the basic form is correct. This (I think) contributed to a sharper look, steeper rim et cetera, but I would like to (at a later time, probably), adjust the textures to look more like the inward slumping features visible in many larger craters. Here, presently, that is only achieved via a fairly cursory attempt at sculpting it in.

I have also been reading much about crater formation, which is quite interesting. I hope to define a couple of parameters about the crater and planet/moon upon which it rests shortly, that I can incorporate some more details and proportions that exist consistently with each other.

Before I started this thread, the preferred method I determined for generating small craters was to, using GIMP, make an image composed of many layers of black dots This was the method I used for the craters on the first post. Individually tiled and offset from the rest, these would produce seemingly random patters of craters that could be used as bump maps or to displace the mesh.

Inspired by looking into sculpt brushes and some of the other texture copying ideas, I decided to do the same thing, using renders of simple craters instead of mere dots, and I here present the result. It went, I thought alright. Some of the craters (the small ones inside of larger ones) did not due to the fact that those craters do not add their values, and thus the rims should be about the same height as those on the unaltered surface. I am looking ways this may be fixed.


The light patch around one crater I included as a test of an idea about adding the ray patterns that are often visible younger craters. This was a supernova effect from GIMP used to define two materials on the surface. Next time I should do more visible rays and possibly secondary craters as here:


That supernova effect from GIMP provides a wonderful effect! This is a great thread to watch - keep up the good work.

Thank you. It is quite fun on this end as well.

Because it just turned out so well, I thought I would post a redo of a crater ray system that I made using only a textured plane.


It is incredible how much faster renders go with only one polygon. This took less than two minutes at three samples (an excessive number, I know).

hi,

it’s height map moune :

http://www.mapaplanet.org/explorer/images/kaguya_lalt_500.jpg

you can use in displace for more realistic render

add plane, subdivise 5/6 number
add displace texture and adjust the level of extrude !



what do you think ?

That would probably be a good idea, imagecorp. I would not want to use it exclusively, though, as it would be rather more difficult to integrate separate features in amongst the craters, when they are all on one image texture. So far I have only really tackled the cratering aspect of terrain, but in reality there are many different things going on on most planets and moons, even some asteroids. I will probably move on to covering more different features, like geysers and volcanoes, later.

It is also important to note that based on the conditions of the surface where a crater is formed things happen differently; craters of the size to have central peaks on Mercury, do not always have peaks on the Moon, owing to differing gravity. I probably should, however, use appropriately sized/scaled crater topography maps in place of my own textures, for displacement maps, but rather than using directly such images I could remix parts of them into layers to suit my needs, and integrate my own features.

UPDATE:
I am currently trying to use some the second of your images (not the one focused on a crater, which happens to be truly enormous), to create a base texture with general roughness but limited large catering, to build a more complex terrain onto. It works quite well, thank you imagecorp.

As I mentioned before, I had been intending to do some research into reasonable actual sizes and proportions of impact craters as these. It is quite interesting indeed, and I could discuss it more at length than I shall here, but I will be more to the point. Craters can be grouped into several classes, which in order of increasing size are: Simple (bowl shaped craters lacking peak structures), and Complex which is broken down generally into those craters that possess simple peak structures, to expanded complexes and rings, to the large multiringed basins, as the Mare Oriental, an image of which was provided by imagecorp. Depth to diameter ratios decrease as one moves from a simple crater to a basin, although actual floor level drops. The transitions between these classes depend on, largely, surface gravity of the object being hit.

Overall, based on the proportions I found and measure, I am thinking that I will go for an object with a gravity rather less than the moon, have a crater of between 40 and 50km in diameter, on a square base that will be between 50 and 80km (I at first was going for a landscape dominated by the crater alone, but am rethinking that some). The crater itself shall be between 2 and 1.5km deep, probably closer to two, with a peak maybe 50m below the surrounding surface.

That was probably the easy part.

Normal maps have been giving me a great deal of opposition, and my attempts to quell their uprisings have been difficult. During their use I have had terrains disappear entirely from the rendered view, act as though the sun was coming from the opposite direction, negate actual features, and do nothing at all. During their generation I have had problems with holy images (I found some suggestions that this might be from overlapping vertices, but on my UV maps, I did not see evidence that caused me to think that was the case), as well as weird bubbly warped looking ones. Attempting to fix all of this has been frustrating indeed, and is perhaps worthy of another WIP thread in itself. If anyone has any thoughts on how to troubleshoot this, that might be helpful indeed, I am getting together more images to upload.


A bit of real lunar terrain, generated from LRO data on my computer. Visible near the edges are patches where the normal map is transparent. This is simply a flat plain, were I to have displaced it to match the normal map, we would be able to see that the sun angle is reversed, as the shadows would be strange and contradictory.


A normal map. I had hoped for something that was more normal.

After imagecorp’s suggestion (an idea which Kilon alluded to before), I began looking into using actual textures generated from Lunar Reconnaissance Orbiter (LRO) data. MapaPlanet, a service provided by the USGS, provided access (albeit slowly) to this data, which I have been downloading, modifying, and turning into meshes (and normal maps). This has not gone badly, and the meshes themselves look quite excellent. It is also an easy matter to get colour maps to pair together for them. However, issues with normal maps are still irksome and have hampered this attempt. I have been able to modify and layer these fairly well with other feature that I want, and am experimenting with ways to layer them together in Blender.

(If you are interested in reading more about crater formation, a well put together website I found was www.lpi.usra.edu//exploration/training/resources/impact_cratering)