I’m always searching for a productive way of working, some way to create ideas and produce finished artworks that look half decent. I recently partly completed a course on level design (on Udemy) and this could be the way to go. Most environment artists are working for game companies. One has to produce environments with strong centre of interest and be well composed. There seems to be a lot of info on designing levels, not just the mechanics of modelling/rendering etc. It’s very concept and composition focused.
Some time ago I had a look at the Godot game engine, and I’m going back to that. The teacher of that level design course was implementing the ideas in Unreal and Unity, but I’d like to stay away from those engines for reasons I’m not really clear about. I think I’ll have to do a bit more work in Godot because it doesn’t have the vast asset library that they do, but all I really need is to get a playable character into a scene that I can import some assets into and I should be good to go.
A lot of my inspiration comes from Cindy Sherman. As a photographer I was inspired by her ‘Untitled Film Stills’ and did some similar work myself. Now it’s ‘Untitled Game Levels’. Moving with the times.
Lots of sales on at the moment. Udemy courses for $10 for example. I’ve purchased a few. And half off all kits at KitBash3D. Even at half off they’re still pretty expensive but I did buy one that seems a good fit for the kind of work I’m doing at the moment.
After spending years on nature stuff I’m moving more towards environment, as in the kind of environment one would find in a computer game or film. Built environments basically. Buildings. I’m looking at courses on level design. I think the last few posts here have shown that pretty clearly. This means I need to spend more time on modelling and texturing.
After dithering about for far too long I finally started on modelling with camera projection. I found a nice simple subject, used fspy (great program, that) to match the camera, scaled up a cube to fit and used Project from View to texture.
I’ve had a few issues with sfpy in the past, when it was called blam, but either it has improved or my understanding of how to use it has improved (probably the latter) and getting this result was pretty straight forward.
I’ve had a few focal length issues in the past, especially. My cameras use a crop sensor, but the lenses are generally rated (focal length) for a full frame camera. I noticed with this image that fspy identified the lens as 31mm (the stated focal length) but Blender gave the camera as about 48mm focal length, which takes into account the approx 1.5x magnification factor of the crop sensor. I guess what mattered is that the cube, when scaled up appropriately, fitted the reference image perfectly.
The scene needs a lot more work to complete, but I was really just getting the workflow sorted here, as usual. I wonder if I’ll ever put in the work required to finish something properly.
The background here is the HDRI I used to light the scene, not the original photo containing the building.
After much exploration (and frustration) I managed to transfer the texture from a projected view to a regular unwrap. Easier to edit I think. I’ve removed a car bottom right, not very well. I need to use BI to do the cloning. I wonder if anyone on BA will tell me how to do it with nodes.
One of my first digital photos – the old Melbourne Gaol, with a rendered building in the courtyard.
I’m copying all my media to a new NAS. I had a few teething problems but most are sorted now. It’s a long process so I’ve now got the Linux box doing the transfer so I can do something else on the laptop.
Shot from my old office in Building 51, comped in a building from the Cyberpunk mini kit from KitBash3D.
Here’s a pic I took many years ago from Building 51, 8th floor. I’ve replaced one of the buildings with a 3D render from the Brutalist kit that I just lashed out on. Not really fitting in all that well, not sure why.
And another, inserted rendered building into back this time. I’m really enjoying this approach – take a photo and enhance it with some rendered elements. This is very common in matte painting for film. Work is still pretty basic but at least I’m actually producing stuff and getting practice at compositing.
I’ve replaced a house in the photo with a rendered mediaeval house. Not totally convincing but a bit of fun anyway.
Melbourne now has it’s very own destroyed building. With a slightly sinister blue-ish cast despite the sunny day. Using View Layers in Blender makes this a snap.
I wasn’t getting the brightness and contrast on my rendered element (the tree front left) so I ditched the HDRI lighting and replaced it with a bright sun and a ground plane to reflect fill light. This allowed me to tweak the direction of the light more easily too, and I’m happier with the result. At least I’m learning something from experience and not just from tutorials.
This time I used one of my own photos from Tepoztlan as the background image, and composited in a render of an English Yew from xfrog into the foreground (left). I guess I could have just rendered the yew with transparency and composited it over the photo in GIMP, but instead I actually imported the image into Blender, used an Emission Shader, and rendered both image plane and yew. I was trying out using View Layers, which turned out to be pretty straightforward once I discovered how to prevent the HDRI from being rendered too. So I did composite the two view layers in GIMP, I suppose I could have done that in the Blender compositor.