RSS Feed
IndexHellkeeper's Archive
Hellkeeper, 08/07/2011, 13:50
Post-Mortem DmRiot
As promised in my last note, here is the post-mortem of my last map and its chaotic development.
Two years were necessary to build this map, yet it's nothing impressive. The reason is that these two years were mostly spent thinking about it and failing to build anything, with only the last months actually spent mapping anything out of the pile of ideas and failed designs I had accumulated over time.
On 15th June 2009 at 19:57, I rebuilt and saved the final build of DmExar, with a new music and a collision bug fixed. The few weeks afterwards were spent looking at it with a fresh new look - and finding two more problems: a misplaced corona and an almost invisible texture problem caused by wrong brush order; these were never corrected as I did not feel like releasing a third version when those were barely noticeable issues. After that, I wanted to do another U227 map but had no idea what to make.
I was quite busy at the time, yet managed to plan a medium-size layout which, at the time, seemed quite nice and original. Many parts of it were hastily drawn in the margin of whatever piece of paper I could find. For personal reasons, I was not able to start working on it immediately, and it's only in January 2010 that I decided to start actual mapping. I had decided to use a classic yet twisted theme based on brutalism and totalitarian architecture, which, now that I think about it, is not entirely unrelated to the landscapes of ruined London in the 1984 film 1984 I had just watched. I wanted massive concrete structures with little decoration and simple flat surfaces, the opposite of Exar. I also started thinking about style and fixtures, and that's when I designed the concrete eagle in BSP.
It took a few months of fiddling to get the size of things right, before I started the map itself in April. It immediately became clear that though the layout looked good on paper, it didn't actually work in 3D. There were only two real "rooms", with many corridors on the sides, but since every zone was both small and open, they usually felt unnecessary. Several locations were completely pointless, the main fighting area itself was a large flat zone on only two levels. I was enthusiastic at first, as I believed whatever looked bad on paper would turn out better in-game, but since the plan was more or less devoid of straight lines, places which looked close on turned out to be impossible to link together propertly in the game. I spent two weeks trying to salvage the floorplan until I finally dropped the idea completely, foolishly thinking that I would have better ideas in the near future.
After months of idleness as far as mapping was concerned, I came back to it and tried to see if I could save parts of it, including the room with the eagle. I liked this area more than any other part of the map and it was the only room I had already started to embellish. I deleted the northern section and the upper levels for which I had no hope. Meanwhile, beta versions of the 227g patch were delivered to testers by Smirftsch, who gave me access to them. I finally abandoned the layout in its entirety in May, when I was given the task of writing a basic help file for UnrealEd. Seeing how DmRiot seemed doomed, I agreed to help to keep myself busy. By now, 2010 was about to end and I felt quite anxious, as it would be a full year without a map in sight and it would be the second year since I discovered Unreal I would not release a map (the first one being 2007).
Starting from the eagle room, I tried to design new layouts, but only one was completed. Even before it was finished, it was clear to me it suffered severe issues, making it even worse than the first draft. I looked at ancient aborted projects from 2007 but it didn't spark better ideas. My attempts at integrating these elements into outdoor terrains lead nowhere either. In the end, I abandoned Riot completely. After more thinking, I started a small DM project which quickly died out, as it was plagued by bugs. It was followed by project VIL, an industrial-themed realistic DM map with static-meshes, custom textures and detailed geometry, but it didn't go beyond a single room. Though I keep it somewhere on my hard drive, it's clear that nothing will ever come out of it without a complete overhaul and actual planning. 2011 was now upon us, and 227g was released in April 2011, more or less a full year after I had started working on Riot, which was now in my Aborted Projects list.
Being quite depressed and badly wanting to map something, I came back to the eagle room. Unlike before, I took only the room itself and nothing of the surrounding areas. It had previously been the inside of a small building surrounded by large volumes and great spaces. I now considered the entire idea flawed and I kept only the room itself (the inside, none of the brushes that made the outside). I started building the north-west room, improvising. After that, the east room was built with difficulty and the project ground to a halt when I had to build the southern part of the level. It took days and days of thinking, trying things and talking with Jival to complete the painful delivery of the southernmost room. After that, I was more than fed up with this and the last room was hastily slapped together with minimal planning, reusing the stairs from the main room of DmExar. I had decided earlier in the process I would make it a rainy map with muddy terrain on the lower level, and I was still using the brutalist totalitarian design I had adopted in 2009. After cutting out the window part of a texture from UT, I created most of the few decorative panels with Tarquin's extruder brush. The ground itself was made with static-meshes. As 227h had been released in the meantime (being the second version released during my work on this map - which says a lot about my workflow), static-meshes enjoyed a solid support. I used the BSP based terrain tool to create them. In 227h and subsequent versions, surfaces with the FakeBackdrop flag are deleted when a brush is turned into a static-mesh, a feature I used to delete all the surfaces of the terrain brush except the tessellated floor whose normals I flipped in the mesh browser. Once converted back into a CSG brush, the result was a tessellated sheet, vertex-edited into a muddy terrain. A custom ground texture was applied, aligned to floor, and the whole thing was converted back into a static-mesh. Each room has one sheet. The eastern stairs were also converted to static-meshes and the BSP underneath it set to FakeBackdrop as a BSP hole appeared there which would not go away despite all my efforts. I also decreased BSP cuts with the building sliders, as the entire structure is very inefficient and prone to BSP problems.
After that, only cosmetics were applied: a couple of projectors with loosely fascist-looking images to mimic posters, a shield-and-sword panel made of BSP. They were scatttered around generously, along with a couple other eagles and a panel of concrete flags (more soviet than fascist-looking). For the rain, I used one of the new actors from patch 227. As it can only create rain in a whole zone or in a cubic volume around it, I used many of them in each room to cover the many odd angles. Lighting and item placement were both made in 10 minutes, without devoting too much effort or time to the process. For sounds, I used the rainfall sound of NyLeve as a rain sound, a few wind and fountain sounds on the ground level to mimic the impact of raindrops on concrete and mud. The custom mud texture was given "flitch-flotch" footsteps sounds and its hitsound replaced with a grass footstep sound, so that players jumping into the mud from the upper levels would not make a big "blomp" as if landing on stone.
The baby was released to a candid world on the 21st of June 2011, more than two years after DmExar. The next day, version 1.1 was released with a few changes: the outdoor lighting was slightly changed to a pale orange - it makes the inside lighting, which is completely white, look much harsher by comparison. A possible collision bug was corrected on the stairs of the eastern room and the walls of the lower level were covered with a new texture. A lighting bug was fixed on the lift.
And thus DmRiot finally ended. It's not great, but it's not that bad either. There are ideas I always wanted to use there, but they are not fully developed. Many aspects of it were rushed. Yet I like it as it is. I guess the best way to see what a mess it really is it to take a look at it in UnrealEd. Go ahead.
Things were not as over as I thought however: in 2012, Smirftsch decided to include both DmRiot and DmExar in the next releases of patch 227 for Unreal, and some of his comments on both maps convinced me to tweak a few things. The emitter actors were modified by the 227 team specifically to allow the use of particles from another emitter on impact so that I could create impact effect where particles collided with geometry.