We reached out to Rens and he was kind enough to grant us an interview.
>> See Also: Interview with Illogic | The Making of Garden Party
1.Could you please give us some details about what you're doing on advancing real-time graphics and on this Project!Having to learn, take risks and lean on the edge of bad performance while having to deliver in a work environment is very difficult. It is because of this, that I decided to freelance and take time to explore what is possible with real time graphics by asking, "What happens when you ignore performance, memory and basic technical limitations and purely focus on getting it right"? Would I still produce roughly the same results, or would photorealism get much closer to for instance, photorealism?
(The answer is you can get a lot closer, without really having to destroy your performance and system memory)
I wanted to avoid building just another scene for my portfolio and instead, aimed at where I wanted to be; to establish a studio focused on producing high-end games and experiences.
One of the first steps for this is to actually achieve some form of quality. For this, I tried out various workflows and reiterated for over two years. Every time I got something done, I learned what not to do. The hardest part with this is to hit "delete" and start over, but again, try getting that one step right until you go wrong somewhere else. After these two years, I hit delete! It was one of the hardest positions to be in and very demotivating to end up empty-handed again.
This time however, I have been working on and off for a year on content and I enjoy the results I am getting. There is still room for improvement, but it is a solid foundation to deliver a first experience. A Tech demo that combines scifi, AI, robotics, nature and story as a first introduction.
If all goes well, the next step is to expand. Get a small team of talented friends together to begin developing a universe and game.
The Advancing Real-time Graphics video shows high resolution captures of the volcanic island Fuerteventura. The scenes exist out of multiple 3D meshes created with photogrammetry. These were captured using an older Nikon camera and relied too much on chunks of environments. Photogrammetry results tend to get blobby when the underneath and spaces between rocks are not captured well or the mesh is too low resolution.
To avoid this, I applied another workflow when I went back several months ago. I photographed surfaces and parts of the environment like before, but also made sure to capture individual rocks. About 200 of them. With the use of physical simulations, I could let thousands of rocks fall on to a surface and generate any type of landscape variation whilst ensuring that the data found on the underneath or between spaces gets preserved. This does not only gets rid of ugly parts in the mesh but lets the light travel between the meshes, making it look much more realistic.
Exploring the limits of real time rendering
I applied the same idea to the forest I am building. No more do I rely on just a texture for the terrain or placing out a few leaves at random. By simulating hundreds of leaves and pine needles to drop to the ground, I can get the look and feel of true depth and realism.
The leaves and plants for the upcoming video are not captured using photogrammetry but are constructed using several layers of photographs. I built a light rig that allows me to capture the diffuse and subsurface information. To create the normal and height information I use Allegorithmic Substance Designer. The multi-angle node allows you to create these textures by inputting photographs with different lighting conditions.
Be out and observe. Don't just recognize what you are looking at as a rock, plant or tree, but take the time to look at the structure or how the light hits it. What shape do the branches have, can you still recognize a tree once its leaves are gone during the winter? What properties does a leaf have? What patterns of growth or appearance do you see, does something grow in lighter areas or is it along a trail, what is the ground made up of etc? Do this often enough and you build up a pretty good understanding of how to piece the world back together in a much more realistic way.
4.What are some of the important things you discovered during this Project?
You can't rush art. I have worked on plants for over two years, making thousands of captures. Every time I think I am done, I appaer to be far from finished because it is one thing to just create an environment and call it finished and another to say it's below expectations.
Which is fine, it just means you should not worry about spending more time or hitting delete and starting over. As frustrating as it is. A big part of the decision for me to have started over several times comes from the importance of a clean, strong foundation. Having a lot of average art assets would bring the whole image down and eventually only allow me to take the quality so far.
I only have positive things to say about Unreal, It is my go to engine. Not just because it's great to work with but the people behind it are approachable and very supportive. It is hard to imagine doing what I do without it. Such a powerful engine, allows for an in-depth customized workflow while maintaining an easy to use work environment. To have vast amounts of documentation and support available enables a single person to create each step of a product, which is an amazing feeling.
What comes next is hard to predict. I keep being surprised on a regular basis with what it's being applied today. All I can do is to put forth my own wish list to improve the work I do and hope to have a small impact on its future.
6.Any sneak peak about what is coming from you side? Could you please share your vision about real time graphics and the visual quality of games?
While the wide view shots of the forest and demo are still work in progress, I can show a few images displaying the quality of the content. More work and info will be available soon, including a new video. Though photogrammetric has made a huge impact on how we build content, it still requires all the skills we applied before.
Some projects will be able to separate themselves more or get closer to their vision, but there is only so much you can capture after which you still need to glue it together.
What my hopes are is to see a new focus on graphics and for 4K to truly become the standard. Looking at the Steam statistics, 4K is used by 0.45% while 1080p is used by 76% of the people playing games. Graphics cards have become stronger but still struggle at this resolution. With a new series of GPUs around the corner (fingers crossed) we might see some changes.
Another change that is allows for is for talented people and teams to separate themselves more. Not by allowing for more inefficient content to go through, but to preserve the hard work and effort that went into creating the high end content in the first place.
I am personally very excited about what we may see in 2020. Another generation will be making the difference.
Even though I can get my work to run between 30 and 60 frames per second at 4K right now, it will really take off the breaks.
More about CG Record Interview [post_ads_2]