Last week we talked about how engine change will impact our game, how it impacts development and what replication is and how it can improve our game. A lot of you expressed your love for our characters and how detailed they look, so this week we want to shed some light on how our characters are made and what makes them stand out!
Contrary to popular opinion, we are not utilizing photogrammetry in our game, that’s the Get Even and Chernobylite’s team domain. In World War 3, in the character department, we use something that’s called structured light scanners.
3D scanning is a process that analyzes an actual object to collect data about its shape and texture in order to reconstruct said object in a digital environment. There are few ways to go about that, few technologies with each having some pros and cons, but we chose structured light, as we feel it suits our needs the best.
To put it bluntly, you can point a scanner at an object and move around it (or move the object) to create its virtual equivalent. Structured light helps the process by projecting grid pattern on an object, telling the scanner how far or close it is from it. That way you can reconstruct that object, like in our example case, a military vest, and preserve its shape and all tiny imperfections and dents, as well as its colors, that make it look and feel realistic.
Some objects require more or less detail than the others to give them a feeling of realism and accuracy. For example, the British military vest is a large object, so we are using a scanner that has a bigger working area, but leaves a bit of details out of the scan. The vest doesn’t really require that much detail because it is treated more as a “background” for various attachments and items on it. While scanning boots, knee pads, knives and items that are similar in size we utilize a scanner that has much more accuracy, but has a smaller working area. This helps with bringing out all the imperfections and smaller details that would otherwise not be visible. It takes considerably more time to calculate the mesh, but still makes it worthwhile to scan the object.
That’s why we have a couple of 3D scanners, to cover both big and small subjects. Currently we have access to three Artec 3D scanners, having Leo, Spider and Eva at our disposal – though we are not using Eva lately, mainly because it was replaced in our workflow by Leo; it is the newest scanner in our arsenal. Spider is used for the smaller objects. Its working distance – how far from an object it has to be to capture the detail and shape – fits between 0.2m and 0.3m and the volume it can scan (which means how big an object can be) is up to 2,000 cm³. It works flawlessly with boots, smaller pouches, helmets and objects of similar size. Smaller working area and volume allows Spider to scan with the most accuracy, up to 0.05mm. In comparison, Leo, our newest scanner, has working distance between 0.35m to 1.2m and volume of 160,000cm³, which means that if we wanted, we could easily scan a car and still have a great scan quality. This scanner can work with a lot bigger objects, but has a point accuracy up to 0.1mm, which is still amazing, though not as detailed as Spider. It also scans much quicker than Spider and Eva, capturing frames at 44fps. The scanner it replaced, Eva, has the same 3D point accuracy as Leo, at 0.1mm, but has a working distance between 0.4m and 1m with volume capture zone of 61,000cm³. Also it only captured frames at 16fps, so we had to scan really carefully with it. Key difference between both machines is the connectivity. Spider requires constant connection to a PC/laptop for it to work, Leo is completely wireless. Which technically means we could just go out and scan whatever we wanted, but it being fragile and really useful to us, we would rather not!
All Artec scanners can also capture textures, though here at World War 3 we are doing textures by hand. It just looks better and gives 3D artists more freedom.
Using structured light helps cut down the time it would take to make the 3D model by hand and helps bringing down the production time of a High Poly version of an object from a couple of days to just a few hours, and sometimes even shorter.
After scanning and first pass of cleanup of the model, such scan goes further down the rabbit hole into the hands of a character artist.
3D artists, using specialistic 3D sculpting software create a High Poly model from received scan file. Such model is then processed and worked on by hand to eliminate any errors and mistakes that made it through the first pass of cleaning up. Its form and silhouette is fixed to match the general guides we use, so it works with different models. For example, when new boots are added to the game, artist has to make sure it fits into the pants, so there would not be as few parts sticking outside of the model as possible. Next step is to create a simplified model – the less polygons there are, the better for the overall performance of the game. Such model does not have all the details that got preserved during the scanning process, it is it’s simplified form, but using the High Poly model as a reference and a source, the artist is able to “bake” normal textures onto the Low Poly version, so it looks like all the details are still there, even though they’re not.
Then, the model is basically ready for texturing. Texturing process, which basically means painting all the colors and how an item shows up in the game, helps with defining the look of the surface of the model. It also encompasses assigning physical materials to the object. Last part of the process is to create various customization options for the model, and currently (as of 0.6 Warzone Giga-Patch Update) each item that can be changed in our game has more than 20 unique camouflages and color variants.
Just before we end the report, here’s how the character looks in game after final touches:
As usual, see you next week for our weekly update!