Most interesting and inspiring artist on the business.
Great interview! I appreciate you make this kind of material that inspire and helps people in the same path :) Thanks 80 level and Mats! Ancelmo Toledo ;)
Thanks for sharing.I found a lot of interesting information here. A really good post, very thankful and hopeful that you will write many more posts like this one. - earn to die
If you’re doing some work with Unreal Engine 4, you’ve probably seen the most recent scene “Kitchen 1930” by Michał Kubas. It’s an incredible project with very vivid, colorful details, wonderful assets and incredible lighting. We’ve contacted Michał and he was kind enough to talk about the production of this scene. Michał has a vast experience in 3d art, having worked on Dying Light, Dead Island: Riptide and Painkiller: Hell and Damnation. He has a lot to share, so we’re incredibly honored to feature some of his advice on our website.
I fell in love with computer graphics at a time, when I got my first 3D computer game „Croc: Legend of the Gobbos”. Back then, when I was around 8 years old. I was pretty sure that every frame is ‘drawn in MSPaint’ and computer was responsible only for choosing between them, depending on a pressed key. However my adventure with graphics has started much later. When I was 17 I got my first paid remote job, which was connected with mobile games. Next step in my career was working in a studio, where I spent 2.5 years.
Presently I work solely as a freelancer. My main activity is creating guns strictly for first person view, although as often I make some props or complex environment elements for games.
For me, environment should be something more than kind of decoration, collection of stuff which tells us where action is placed. Good level can easily give us story, cause feelings, emotions and carry the strong message. Game world is extremely essential, at last we’re talking about VIDEO games.
The Kitchen Environment
Impulse, which inspired me to create kitchen was a photography of an old fridge. When I decided what I want to make, I’ve started searching movie shots, pictures from catalogues, books, photographs of the reconstruction of the museum etc.
Design from the thirties contains all the elements which I appreciate and which are in harmony with medium of the video game. All the forms are thick and expressive. It can be noticed in massive hinges or in pumped details. Everything is clear and obvious at first glance. Unnecessary details, requiring close focus, have been omitted.
During planning of environment, essential is choosing elements which are characteristic. I focused mainly on a furniture (fridge, stove), popular color juxtaposition back in the days and also design of labels. Labels design from the thirties is rather unique, properly reflects visual atmosphere and décor.
Choice of objects with different size gives much more possibilities regarding to presenting and building natural arrangement than in case when we’ve got objects with similar size.
Creation of the Interior
Interior is bigger than typical kitchen of those years, considering customization to game standards (space to walk, width of door etc). Naturally it was work aimed to portfolio so I could skip gameplay requirements, however in my humble opinion if something is supposed to look like a game, I’m obligated to try to think in category production-ready.
Shape itself has changed during production. Originally it was a little bit smaller room, but with time it turned out that room has not very interesting composition and looked firmly flattened. It was necessary to add more vertical lines of wall, thanks to that exertions, space took on a different character, it was no longer that wide (spilled) and light set up in more interesting way.
I’ve imposed large rigor if we’re considering block out. In prototype phase I determined not only room layout and necessary equipment with their sizes, but also:
– size of minimal detail (screws, indentation etc.) which will be allowed to keep readability even though it will be much zoomed out.
– target number of pixels per 1 m^2
– target level of density (“roundness”) of geometry, in such way to avoid situation in which some of the objects are too much round and the others are edgy.
Struggle with Lighting
The next step was struggling with lightning, which has lasted till the end of the production.
Unfortunately in terms of operations with light I’m a inexperienced and I was forced to carry out experiments to obtained satisfying effect. Similar situation was with frames (setting of cameras), which corrections have introduced till last moment.
When I more or less knew how light will be spread I started to provide more details of room whitebox – adjustment and interesting composition small details, smaller tweaks. I wanted to obtain natural – not museum- arrangement. At this stage it’s better to imagine yourself standing there and doing something, like simulation of your real life in this room. If it would be my home, where would I keep cans? Where would I put my coffee grinder?.
Sometimes it’s better to not exhibit something at first glance, despite great effort put in model, it will create much more beneficial impression. Labels are good examples for that – in shop they are directed toward us – while we’re looking at our houses, it’s rather rare to experience such perfect setup. Unfortunately I failed with it in few places.
Of course, I think that lighting is one of the most important issue we have to deal with, but I’m still trying to figure out my own way. When I’m looking for an inspiration, usually references to light are autonomic from props references. Here I was basing on Peter Ilsted or Vermeer’s paintings. (I’m not fan of that type of paintings, but imo that scheme of light stacked up perfectly in case of my kitchen)
Light had bland blue shade to contrast it with environment which was kept in rather warm coloration.
If we’re talking about modeling I use Blender. Despite of fact, that most of assets are just simple polygonal modeled, creating them was pretty time-consuming. For sure, creating high polygonal spaghetti was challenging.
Later, after bake, all textures went to dDO. It saved much time and energy. After creating smart-materials it was over the hump. It’s obvious that I couldn’t stop at that stage. Textures from generators are usually pretty raw, they’re not unique, let’s say without emotions. Sometimes it’s enough to cover texture with small stains or other additions, everything well placed. That’s why few parts of texture had to be refined in Photoshop, repaint, mixed with photos, etc.
When I was planning environment I wanted to get full scale of materials. From unpolished to shiny, from nonmetallic to chrome etc. Such variety of surroundings gains a lot.
When I was looking for the answer on my own, I compared textures and materials with paintings, especially impressionists. Incredible impressions, rich colors and beautiful facture. And probably this is the secret, in reality nothing is uniform, ‘mathematically supreme’. Sometimes it’s necessary to take a closer look to see it. Generally I obtain reality by mixing photographs (sometimes even absurd connections – have you ever seen how it’s effective to make rust on metal from bread photographs?), noises or correction painting with brush. But it’s rather better to not overuse it, we should combine sensitively detail and rest places.
Of course, I personally prefer in my works something kind of stylized realism, where artistic values tells about real world. However I don’t think that any other artistic convention is worse.
In my opinion key to the success is throwing away all PBR values arrays. Maybe I will sound like a rebel but it limits my creative work. It’s better to consider environment/model as a set of planes, elements, accents, silhouette and contrasts rather than connecting to it realistic substances/materials. I’m trying to reduce tonal contrast in albedo, especially when light play a key role in scene. Contrast of shot comes from light and shadow, not from the black and white objects in frame – thanks to that everything is consistent.
I focused strongly on unique assets. After few tests I abandoned idea with masking elements on material. In this case, solution with dedicated textures seemed to be more reasonable.
Material itself was pretty simple. In comparison with previous generation, it’s much more easier to obtain nice, various materials. I remember creating material in UDK (for stylised basement) was much more complicated and time consuming.
Exceptions are labels on bottles and cans. Here, I used unsophisticated colour-masking system and thanks to that I was able to create infinity number of versions of one label.
I’ll be honest, floor in the kitchen defeated me so badly. I didnt prepare precised references. I’ve just made floor and thought “seems cool”. I should have added vertex blending layer with layer more destroyed and do some visible scratches in specified places, add stains combined with dust etc. After collecting feedbacks from wider public, it became obvious for me, however I didnt think about it earlier. I thought it was ‘ok’.
Working With Unreal Engine 4
I’ve been working with Unreal Engine for few years (since UE3), so I’m such familiar with it that I’m not truly able to speak objectively. However when I started to use UE4 difference was easy to notice. Unreal is a tool simple enough to use and much more intuitive, I will say it became approachable for everyone. Especially in UE/UI Epic matter they made huge progress.
Unfortunately working with engine which has static lighting can be irritating. Most frustrating was waiting for results even though changes were little. Good solution is working on much smaller resolutions of light map, when we’re interested in baking on higher production quality.
Quality Control in Environment Design
I think that most important issue is regular checking everything together as many times as it’s possible. Many computer graphic designer have such problem: they’re able to make great model and texture seperately- nice table, lamp, sofa or whatever but when they put them together it looks really bad. I’m trying to treat environment as living structure, organism in which there are many correlations between its elements. It’s similar to drawing or sculpuring figure- you’re supposed to do it as a whole object, not partially and then connect them at the end.
Worst enemies are things that we’re sure about. If we’re wondering it’s the matter of time when we’ll find solution. In the end usually I observe every shot in huge and small scale. I make a list of problems, things that must be corrected and I’m trying to remedy them till I decide that something is good enough.