positional constraints over time as a space-time optimization problem in the tangent space of the curves driving the animation controls. Their method has the key properties that it allows for the manipulation of positions and orientations over time. https://www.lost-identification.com
Dela Neve did a breakdown of his concept piece Ruins Scouting made with Photoshop, 3D Coat, and Blender.
Hello everyone! My name is Dela Neve, I’m a Venezuelan/Italian designer currently living and working in Madrid, Spain. I graduated as an architect and have been working as a 3D generalist for ArchViz, but now, I’m making the move towards Concept Art which has always been my goal.
Today we will be talking about the process for Ruins Scouting and some of the decisions behind it.
My workflow changes depending on the tools I need for each task. With that being said, let me give you a general idea of how it mostly looks nowadays:
- Reference: I search online for references and gather everything that is relevant in a PureRef file. I then sort the images by categories I define and choose 4-5 images to be my “main” references.
- Sketching: I do some quick drawings exploring ideas. I don’t try to make them pretty, that’s not the intent, I never even present them publicly. Even later down the road, I always keep pencil and paper close to me in case I need to quickly doodle something.
- Block-in: I make some blocky 3D models to assemble a scene in Blender and start testing if what I have in mind makes sense and has potential. It’s great for checking composition, lighting, scale, and proportions early in the process. I think of this as 3D sketching and as in the previous 2D sketching phase, it looks pretty rough.
- Sculpting: I sculpt my assets in 3D Coat to the level of detail that makes sense for the piece.
- Assembly: I bring all the assets back into Blender and replace the block-in models with finished ones. I make all the changes necessary to the scene and change, move and adjust whatever I feel like to push it further.
- Rendering: From Cycles, I get all the render passes I need for the painting phase. I shared a peek into the passes I used for the recent image, you can check it out here.
- Painting: This is where everything comes together. I mix different render passes, add photos where needed and paint on top of everything until I feel it is done.
I’ll get deeper into some of the steps below.
3D in Concept Art
I think it is safe to say that at this point 3D is pretty standard amongst professionals. Concept Art is about ideas and design, not necessarily illustration, and 3D is just another great tool to work with. There are a lot of benefits to it, like being able to test and explore inside a 3D world or to easily iterate on a design without starting from scratch, thus being able to respond quickly to revisions from your Art Director. It is simply convenient.
I find it super fun and get excited by the possibilities, and for me, it is down to how much 3D vs 2D I want to do for an image. If I’m doing a heavy piece of architecture I’m comfortable with pushing the 3D to an almost finished look. This is not to say that you can’t have a full 2D workflow, but you should have a pretty good base for it in your drawing skills.
Modeling ruins can get overwhelming. They took a little bit of testing before I found a method I was comfortable with. I start taking a look at my Block-in scene in Blender as it helps me decide the level of detail each model needs according to the placement and importance in the scene. This is very important for me to make informed decisions so I don’t waste time on details no one is going to see far away in the distance. I then separate all the assets I need to sculpt on, export them as FBX, and bring them into 3D Coat.
For the sculpting part, I used mainly two ways of destroying my assets. For the big broken chunks, I took some rocks and subtracted them from the clean model. For smaller details, I chiseled away from the model with the cutoff tool.
Geometry can get super heavy in 3D Coat, so before I export them back to Blender I resample my pieces to a manageable size. I usually decimate them again while assembling my scene.
I was worried that too much broken parts could just make a very noisy mess. So in my little R&D, prior to making the final assets, I figured that I could keep the parts I cut off and have a broken version and a cracked (but whole) version for each asset. This meant that I could have an extra render pass with just the cracks but no missing parts, so I could mask back the broken pieces in Photoshop. This gave me a lot of control down the road and allowed me to explore the assets destruction more freely. This was a cool realization for me, and with the right approach, I could stop worrying about it not working out.
At this stage, I’ve already solved key problems like scale and size in my Block-in scene. My main problem here was how to make it look believable, but most of all, avoiding the feeling of a bunch of copy-pasted models. So while sculpting, I separated some of my models into different sections and made some destruction variations. Since I didn’t really have too many assets, it was pretty manageable and fun.
Let’s take a look at the wall piece to further illustrate this. At first, it was a single clean model, but once in 3D Coat, I separated it into a top, middle, and bottom section. I did 4-6 destruction variations for each. Now, my walls are a combination of 3 different pieces out of 15 which gives me a lot of options and flexibility for an arrangement I’m happy with.
Once I’m happy with all my assets and variations, which I keep in a separate Blender file, I copy and paste them into my scene replacing my Block-in models with finished ones. You can copy-paste geometry from file to file with Blender, which is super handy.
Painting in Photoshop
Throughout the whole process, I try to be mindful about what could be solved using photos, what could be enhanced by painting and what should be left untouched. Later in the process, I let all the work I’ve done so far inform my decisions, and I’m open to any self-editing that needs to be done. For example, I realize that the image wasn’t balanced to my liking, so after hours and hours of reframing my 3D camera, I changed the aspect ratio and added some more ruins to the right by copying and painting from what I had already rendered. Such a simple adjustment made it better already.
The painting starts by mixing render passes. From Blender, I get different materials and a variety of data and light passes I can use as selections or masks in Photoshop. My material passes are usually slight tone shifts on the same surfaces. I recently posted a little explanation on the passes I used for another image here.
Then, it’s time for photobashing. I knew the vegetation would be a challenge since there was so much of it, and that it was going to be all photos with some painting on top. That’s why I rendered it as big green simple forms, just to get an idea of where they would be and how they would intersect the other models. I used a vegetation render pass to paint it overgrowing the architecture and to define extra areas for the photos to be added.
At this point, it is very close to done. On a flattened layer, I go through the image, painting to fix and enhance whatever I might have missed. At the very end, I add some adjustment layers if needed and that’s it.
From the beginning I wanted clear sunlight hitting my focal point directly. That’s because I knew I would try to design it to help me define the ruins’ silhouette and separate them from the middle ground walls. That’s what I went for.
I usually start working on it as graphically as possible, defining a black and white two value read, just like I would make a 2D composition thumbnail. I try to keep it that way to have a very clear read to aim for, but sometimes I will introduce values in between as I need. In this case, I was also interested in exploring the background ruins’ silhouette against the sky and the negative space they make.
You can easily set up a scene like this by creating a light source and setting the World color to black and the light bounces to 0.
If you made it this far, thank you so much! I hope you found something useful for your own work. I’m happy to be able to share some of my process.
If you have a question, are interested in working with me, or just want to say hello, you can find me on:
If you have a question other people might be interested in, don’t be afraid to @me on twitter!